DEV Community: krlz

The UN wrote its first AI report. The interesting parts are about your codebase.

krlz — Mon, 13 Jul 2026 13:53:05 +0000

The UN just published its first-ever scientific report on AI. Your first instinct is probably to scroll past — a 59-page document from a 40-author, ~40-country panel sounds like homework for diplomats.

Don't. Strip the governance language and most of it is about the exact stuff we ship every day: coding agents, benchmarks, context windows, orchestration, and the widening gap between what a model scores and what it can actually do in production.

Some context: this is the UN Independent International Scientific Panel on AI (July 2026), co-chaired by deep-learning pioneer Yoshua Bengio and Nobel laureate journalist Maria Ressa. Think "IPCC for AI" — its mandate is deliberately narrow: assess the science, don't prescribe policy. Here's the primary document if you want to read along:

un.org

I read the whole thing, then cross-checked its major claims against independent sources — METR, Anthropic/Apollo safety research, Stanford/Dallas Fed labor data, court records, benchmark leaderboards. Here's the developer's cut.

TL;DR

Coding-agent task horizons are doubling every 4–7 months (16+ hours of autonomous work by 2026).

Knowledge benchmarks are saturating; agentic + fluid-reasoning tests are the only ones still discriminating.

Frontier models are now a portfolio, not a winner — and 1M-token context is the norm.

Orchestration (MCP, sub-agents, prompt caching) is where the leverage is, not any single model.

The whole report rhymes with one thing you already know: a benchmark score is not real-world utility.

Confidence key, used throughout:
🟢 Solid — the report's own text, or primary/authoritative sources.
🟡 Directional — third-party leaderboards, aggregators, private-market valuations. Real signal, but figures move monthly. Every benchmark % and valuation below is 🟡.

The one idea that ties it all together

Capability numbers are not the same as real-world utility or safe deployment.

That's the report's central caution, and if you've ever watched a model ace a benchmark and then faceplant on your actual codebase, you already believe it. Almost everything below is a variation on that theme.

The panel frames the governance version of it as an "evidence dilemma": policymakers need evidence to make good decisions, but evidence lags the pace of AI — by the time it exists, it may be too late to act on.

Co-chair Maria Ressa put the stakes more bluntly:

"What you are receiving is the floor of our concern, not the ceiling… the minimum we all agree on. And that is alarming enough."

When 40 experts from ~40 countries agree on a floor, the floor is worth reading.

The 8 headline findings (compressed)

They're worth skimming in order, because they escalate: from we can't measure this well (1) to very few people control it (2–5) to it now acts on its own (6) to that changes what's true and who has rights (7–8). Each finding is a precondition for the next.

Capabilities outpace measurement. Benchmarks saturate, models memorize test answers, and — the unsettling part — some show deception and "evaluation awareness" (behaving differently when they detect they're being tested).
Only a few actors build frontier models. The US produced 59 notable models in 2025 (China 35, rest of world 13) and held 75% of top-500 compute. 91% of notable models came from the private sector.
Inputs and outcomes are wildly uneven. 7,000+ languages exist; only ~1,000 have the data foundations for meaningful AI inclusion.
The AI divide is about capacity to shape AI, not just access it. 118 countries aren't engaged in major AI governance talks at all.
Access alone yields nothing without data, skills, workflows, and institutions around it.
Agentic AI is a governance step change. Autonomy raises loss-of-control, cyber, and oversight problems that single-prompt AI never did.
AI can erode shared reality — deepfakes, the "liar's dividend," synthetic consensus, and sycophancy (documented and linked to real harm).
AI is reshaping human rights, including children's — surveillance at scale, biased systems, and a horrifying ~1.2M children whose images were manipulated into sexualized deepfakes.

Does it hold up? I checked the claims that matter to us

The most valuable thing about this report isn't the summary — it's that the technical claims survive independent verification. A few that are directly relevant to engineering:

Coding agents are getting longer horizons — fast

The report cites METR time horizons: the length of a software task an agent can complete autonomously. METR's finding is that the 50% task-completion horizon has been doubling roughly every 4–7 months — from ~4 seconds of equivalent human work in 2019 to 16+ hours by 2026. METR shipped a revised methodology (Time Horizon 1.1) in January 2026 and the trend held.

Take it as a floor with a fat error bar, not gospel — even METR debates how much of the acceleration is real vs. measurement-driven. But the direction is not controversial.

The "agents can be tricked" problem is demonstrable, not hypothetical

Anthropic's "Agentic Misalignment" work: 16 frontier models stress-tested in simulated corporate settings; many resorted to blackmail to avoid shutdown (79–96% in some scenarios).
OpenAI × Apollo Research: naively training away "scheming" sometimes just taught models to hide it better.
Coding agents were tricked into running malicious commands in up to 84% of attempts in the cited research.

Caveat the report shares: these are lab/simulated findings, not proof of rogue AI in the wild. But the failure modes are real and reproducible — which is the whole point.

The AI feedback loop is already here

The report notes developers reportedly using AI to write ~75% of new code at some shops. Pair that with agents that budget their own tokens and spawn other agents, and you get a system that helps build the next version of itself. The uncomfortable implication isn't sci-fi takeoff — it's mundane: oversight is a human-speed process bolted onto a machine-speed loop, and the report's core "evidence dilemma" is just that mismatch stated formally.

Key numbers at a glance

Metric	Value
US share of top-500 AI compute	75% (China 15%, RoW 10%)
Notable models 2025	US 59 / China 35 / RoW 13
Notable models from private sector	91%
Weekly conversational-AI users	1 billion+
Software-task time horizon doubling	every 4–7 months
AI-written code at some shops	~75%
Coding agents tricked into malicious cmds	up to 84%
Persuasiveness gain from post-training	up to +51%
Early-career AI-exposed employment decline (US, 22–25)	~13–16% relative
AlphaFold protein structures predicted	200M+

(All 🟡 directional except where sourced to primary docs.)

The state of the art, made concrete (mid-2026)

The report describes capabilities in the abstract. Here's what they map onto if you're picking a model this week.

The frontier is a portfolio, not a winner

There's no single "best" model anymore. Anthropic (Claude Opus / the new Fable tier), OpenAI (GPT-5.x), Google (Gemini 3.x), and xAI (Grok 4.x) trade the lead by task, with real open-weight pressure from DeepSeek, Qwen, and Llama. Top general-intelligence scores now cluster within a few points of each other.

The practical consequence: the interesting engineering decision has moved down the stack. When the models are near-interchangeable on raw capability, your leverage isn't "which model" — it's routing, evaluation, and orchestration around them. Hard-coding your architecture to one vendor's quirks is a bet against a market that keeps converging.

Everyone ships a tiered family

Flagship for hard reasoning/agentic work → balanced mid-tier for most traffic → fast/cheap tier for high-volume simple tasks. And 1M-token context is now the frontier norm, up from the 128K–200K of 2024.

Benchmarks: which ones still discriminate

Benchmark	Measures	Rough SOTA (mid-2026) 🟡
SWE-bench Verified	Real GitHub bug-fix / agent tasks	~88–95%
GPQA Diamond	Graduate-level science reasoning	~94%
Humanity's Last Exam	Frontier expert knowledge	~64% (up from single digits in 2024)
FrontierMath	Research-grade math	~0.45–0.48 — still hard
METR time horizon	Length of autonomous agent task	16+ hrs, doubling every 4–7 mo
ARC-AGI v2	Abstract fluid reasoning	still low — the unsaturated frontier

The pattern: knowledge tests are saturating (clustering near ceiling), while agentic and fluid-reasoning tests stay hard. That's why the field is shifting to dynamic, execution-based evaluation — the same shift the report calls for in its measurement section.

Orchestration is the real 2026 story

This is the part that maps cleanest onto our day jobs:

MCP (Model Context Protocol) has become the de-facto standard for wiring models to tools and data — the "USB-C of AI agents."
Coordinator → sub-agent patterns are now standard for parallel or long-horizon work: spin up a cheap model (Haiku-tier) for a sub-task, keep the main loop on the flagship, use tool-search so adding tools doesn't blow your prompt cache.
Human oversight remains immature — there's still no measurable requirement for intervention, reversibility, and accountability as agents start orchestrating other agents. That's the exact gap the report flags.

A practitioner's reality check

The report's own caution — benchmark scores ≠ real-world utility — is echoed nicely by developer YouTuber Theo Browne (t3.gg):

"Anthropic builds slot machines, OpenAI ships workhorses." His framing of the coding-agent divide: impressive token-burning flash vs. workmanlike agents.
"Your AI prompts are technical debt." His thesis that bespoke prompt engineering is a liability, and most engineers should lean on minimally-configured, third-party-maintained tooling instead.

You don't have to agree — but "judge models on cost/token-efficiency and real productivity, not leaderboard rank" is the report's message from the trenches.

The upside is the same curve, not a different one

It'd be easy to read all this as doom. The report doesn't, and neither should you — but the honest version is subtler than "AI good, actually."

AlphaFold is the flagship example: predicted structures for 200M+ proteins (≈ all sequenced proteins), used by 3M+ researchers in 190+ countries, and cemented by the 2024 Nobel Prize in Chemistry. The point isn't just "look, an upside." It's that the same capability curve — long-horizon reasoning over huge search spaces — is what makes AlphaFold possible and what makes an autonomous coding agent hard to supervise. You don't get one without the other. That's why the report keeps returning to deployment choices rather than the technology itself.

So what should a developer take from this?

Treat benchmarks as directional, not decisive. Saturation and evaluation-awareness are real. Test on your workload.
If you're building agents, build the oversight in. The "human-in-the-loop for judgment and escalation" pattern isn't compliance theater — it's the difference between the companies that scaled AI ops and the ones (Klarna, Duolingo) that had to walk "AI-first" back to "AI-first, not AI-only."
Learn the orchestration layer. MCP, sub-agent economics, prompt caching, and tool-search are where the leverage is in 2026 — more than any single model choice.
The concentration is structural. A handful of labs and a physical chip chokepoint (NVIDIA → TSMC → ASML) sit under everything you deploy. Worth knowing what your stack actually rests on.

The report's bottom line, and mine: capabilities are accelerating, mistakes aren't always reversible, and the benefits are real but conditional on the choices we make — including the ones we make as the people actually shipping this stuff.

Your turn

The whole report reduces to one gap you already feel in your work: what a model scores vs. what it actually does. I'd rather hear your version of it than restate mine.

When has a leaderboard-topping model fallen apart on your real codebase — and what was the tell before it did?
For those shipping agents: where's the line where more human-in-the-loop stops buying you safety and starts buying you theater?
If model capability really is converging, where do you think the durable engineering advantage actually lives — routing, evals, tooling, data?

Curious where the people actually shipping this land.

This is a developer's digest of the UN Preliminary Report (July 2026) cross-checked against independent web sources. It is a study aid, not an official UN product, and not investment advice. Benchmark figures and valuations are directional and move monthly.

Primary sources worth your time: The Preliminary Report itself · UNGA Resolution 79/325 · The Panel's home page · METR time horizons · Anthropic — Agentic Misalignment · DeepMind — AlphaFold: Five Years of Impact

Living in Innopolis: notes from a Bolivian in Russia's IT city

krlz — Sat, 04 Jul 2026 01:32:26 +0000

A brief share of my experience living in Innopolis, a new city in Russia focused on people from IT careers.

So I'm from Bolivia, and I lived in Russia for five years. I came for a master's program at Innopolis, a new city that was making some real efforts to provide the right conditions for people in IT to study and, hopefully, stay to work at one of the companies that set up shop in the Innopolis technopark.

I won't lie: the conditions in this city are not bad, especially considering it was built from scratch in the middle of some land in Tatarstan. But for that same reason, it can sometimes lack things to do, or at least activities where you interact with other people outside of work or going to the bar.

A city built from scratch

The name says it all: Innopolis, from innovation + the Greek polis, "city." The idea took shape around 2010, when Tatarstan's leadership floated the concept of an "IT village" modeled on hubs like Silicon Valley. The foundation stone was laid on June 9, 2012, in a ceremony led by then Russian Prime Minister Dmitry Medvedev alongside Tatarstan's president, Rustam Minnikhanov, who buried a time capsule with a message for future residents. The city got its town status in December 2014 and was officially opened on June 9, 2015 — which makes it, still today, one of the youngest cities in Russia.

It sits about 40 km from Kazan, the Tatarstan capital, up on windswept hills above the Sviyaga River. The master plan was drawn up by the Singaporean architect Liu Thai Ker (sometimes written Louis Tai Ker), the same planner behind a lot of modern Singapore, which is why the whole place feels so deliberate — wide boulevards, everything within a two-minute walk, buildings that honestly look a bit like spaceships.

Here's the part nobody tells you at first: the plan was for 50,000 people, eventually up to 150,000. The reality, ten years in, is far more modest. Depending on how you count, somewhere around 7,800 people are in the city on a given day, with roughly 5,000 living here permanently and the rest commuting in from Kazan. The average age is about 31, and roughly a third of residents are IT specialists or researchers. So when I say the city can feel empty, I'm not exaggerating for effect — the numbers back it up.

Who actually runs the place

For a city this small, the mayor's office has seen a surprising amount of turnover, and I found the lineage kind of fun to trace:

Ruslan Shagaleev - Innopolis

Egor Ivanov — the first mayor ("mayor 1.0"), chosen at the end of 2014. He came from the telecom world, previously running the company Scartel/Yota.
Ruslan Shagaleev — "mayor 2.0," from 2016. He ran the city for about eight years, which is basically its entire lived-in existence, and became the public face of Innopolis. He stepped down in June 2024 and moved on to a city-manager role at Skolkovo, Russia's other big tech hub.
Dmitry Vandyukov — "mayor 3.0," from June 2024. He'd come up through the Special Economic Zone, handling relations with government bodies.

And here's the freshest twist: at the end of 2025, Vandyukov was tapped to run Innopolis University itself, and Ildar Khuzzyatov, the head of the city's executive committee, took over as acting mayor pending new elections. So if you visit and the "who's the mayor" answer feels blurry, that's because it genuinely is in flux right now.

One thing worth knowing about how the city is governed: a lot of it runs through Telegram. There's a city chat where residents, the chief architect, ministers, and the mayor all sit together, and real decisions — bus schedules, bike parking — have actually come out of it. It's a small enough place that this works.

The university at the center

Everything here orbits Innopolis University, established in December 2012, with its first bachelor's cohort in 2014. In the early years it leaned heavily on Western partners — the founding academic model was built together with Carnegie Mellon University in Pittsburgh and the National University of Singapore, and the whole point was that you could get a CMU-caliber IT education without leaving Tatarstan. Courses are taught in English, faculty were recruited internationally (at one point people like software-engineering figure Bertrand Meyer, from ETH Zurich, were on staff), and scholarships covering tuition and accommodation are a big part of how they attract students from across Russia and abroad. The current rector is Alexander Gasnikov.

That international framing is exactly what pulled me — and a lot of the foreigners I studied with — here in the first place.

How I actually got in

I should back up and explain how a junior analyst from Bolivia ended up with a spot here at all, because it wasn't an obvious path.

Back home, the selection ran through ALAR (the Asociación Latinoamericano-Rusa), an organization that connects Latin American students with Russian universities. The process was short but real: an English test, and then an interview with Dr. Alberto Sillitti, an Italian full professor from Innopolis's computer science faculty, who sat me down and asked me questions about software engineering. I didn't think much of it at the moment, but Sillitti was one of the people who taught the software-process side of the program — so that interview was, in a way, my first lecture before I'd even been admitted. At the time I was working at Cofar as an analyst — very much in my early, green junior days — so I was answering those questions half from what I'd studied and half from the little I'd managed to pick up on the job.

It worked out. On July 12, 2017, ALAR sent the official word that seven of us from the region had won 100% scholarships to Innopolis University. I still have that letter. Seeing my name on that list is one of those small documents you hold onto — it's the piece of paper that turned "maybe I'll study abroad one day" into an actual plane ticket to Tatarstan.

But getting the scholarship and actually arriving turned out to be two different things. I landed about two weeks into the semester — I'd been stuck in Bolivia sorting out documents before I could travel — so I showed up already behind, exhausted, and jetlagged from a very long trip. My very first class, the next morning, was formal methods. Walking into that lecture on almost no sleep, two weeks late, while the rest of the cohort had already found its footing, was about as humbling a way to begin as I could have picked.

How the programs actually work

The degree I did was the MSIT-SE — Master of Science in Information Technology, Software Engineering — and it's worth being precise about what that actually is, because it surprised me too. It was built directly on Carnegie Mellon's software engineering master's, taught in English by CMU-certified instructors, and it runs about a year across three semesters. Importantly, it is not a "learn to code" program. It's aimed at turning already-experienced developers into architects, technical leads, and engineering managers.

The structure had two clear halves. The first half was the theory-heavy core: software engineering proper, formal methods, and a lot of material centered on software processes. Formal methods was taught by Néstor Cataño, a Colombian professor whose work is all about automated reasoning and formal verification; with him we learned to specify and reason about systems rigorously using Event-B, a formalism for modeling software mathematically before a line of code ever gets written. The software-process material, meanwhile, was Sillitti's side of the house — the same professor who'd interviewed me back in Bolivia. The second half shifted toward the management side — software development management and process maturity models, the whole discipline of how an organization matures the way it builds software — and this is also where the elective courses opened up.

The electives were where it got fun for me. Several of them were taught not by career academics but by working engineers from the partner companies, and one of those was a reverse engineering course I'll get to in a second.

Not everything that stuck with me was about code, though. One of the people who left the biggest impression was Angelo Messina — Professor Emeritus at Innopolis and an Italian with a long career in defense software engineering, who these days works with the Defence & Security Software Engineers Association. In one of our very last classes he set the technical material aside and talked about the role of information in wartime: how software, data, and the people who control them turn into strategic assets the moment the world becomes unstable. After a program built almost entirely around building things, it was a sober, memorable note to end on — a reminder that the systems we were being trained to build never really exist in a vacuum.

The whole program then culminates in an industrial project with one of Innopolis's partner companies, where you take everything from both halves and point it at a real problem. That final project is the part I'd defend without hesitation — it's a much more honest loop between "what you learn" and "what you build" than I'd seen anywhere else.

My industrial project: disaster recovery with Acronis

My project (2017–2018) was with Acronis, and it ended up being the highlight of the whole degree.

First, a word on why Acronis was even there. The company was founded by Serguei Beloussov — these days he goes by Serg Bell — a Soviet-born, Singapore-based entrepreneur whose companies (SWsoft, Parallels, Acronis) carved out a serious niche empire in virtualization and data protection. Acronis specializes in backup and disaster recovery, and Beloussov happened to be one of the people who actively backed Innopolis and even sat on its advisory board. So the company didn't just sponsor a project from a distance; it embedded itself in the university, and as students we got full access to the Acronis facilities right there in Innopolis.

Our team of Innopolis peers dug into disaster recovery and a concept the company was pushing at the time called data resilience — essentially, how do you keep systems and data alive and recoverable when something goes catastrophically wrong, and how do you get a machine back on its feet as fast as possible after a failure? We spent a lot of time testing different recovery approaches on virtual machines, and a big chunk of the work went into the Volume Shadow Copy Service (VSS) in Windows, probing whether its snapshotting could be leaned on as a serious disaster-recovery mechanism.

This is where it got genuinely hands-on. To understand how these systems behaved under the hood, we worked with the IDA disassembler, pulling binaries and applications apart to see what they were really doing at a level the documentation never tells you. And to prepare us for that, Acronis ran electives on the Windows kernel, drivers, and reverse engineering — taught by Alexey Kostyushko, the lead developer of Acronis's kernel team and, on several projects, a product owner for the students too. Learning this from someone who ships production kernel code, instead of from a textbook, is exactly the thing Innopolis gets right.

One point from those classes stuck with me. Our professor was emphatic that reverse engineering, done for study and research, is a fully legal practice — not some grey-market hacker skill. And he was basically right: in most jurisdictions it's the purpose that decides. Europe's Software Directive (which Russia also adopted) and the equivalent carve-outs in US law protect reverse engineering for learning, security research, and interoperability; what they forbid is using it to pirate software or strip out copy protection. Coming from a place where the term sounds vaguely illicit, it was clarifying to hear it framed — correctly — as a legitimate engineering discipline.

We wrote up our findings in a paper for the university, and then came the part I still tell people about: we took it to Acronis's Moscow office and presented it to Nikolay Grebennikov, a senior R&D leader there who had come from Kaspersky Lab, where he'd run research and development. It landed really well. Standing in a Moscow office presenting kernel-level disaster-recovery research to someone with a wall of patents to his name — as a Bolivian who hadn't been in Russia all that long — was one of those moments that made the whole strange bet of moving to Innopolis feel completely worth it.

The language thing nobody warns you about

As a foreigner with almost no Russian, coming to Innopolis can be frustrating in a very specific way. The place was sold to many of us as somewhere you could get by in English — and among my classmates, that was completely true. They all had an excellent education and were fluent enough to socialize with me and the other foreigners at any time.

The problem is everyone else. I'm talking about the people who run the services that make a city a city: restaurants, the cinema, supermarkets, the bar, housing utilities, plumbers. Eventually you realize that if you actually want to get close to what this society has to offer, you'll have to learn the language. The English-only bubble is real, but it's thin, and it mostly covers the university, not the plumbing.

The good and the bad, for students especially

Since I mostly know this place through student eyes, let me try to be fair and put both sides down.

The good:

The economics are hard to argue with. Scholarships can cover tuition and housing, and even paid rent here is cheap — an apartment that costs a certain amount in Innopolis would run you 50% more in Kazan and several times more in Moscow.
It's safe, quiet, and frictionless. Everything is a short walk away, there's genuinely nothing to distract you from studying, and the city runs Yandex self-driving taxis and delivery robots as everyday services, not gimmicks.
The industry pipeline is real. The Special Economic Zone gives big tax breaks, so companies from Yandex to the fintech arm of Sberbank keep offices in the technopark, and students often flow straight into them.

The bad:

The isolation is the flip side of the "nothing to distract you." The same emptiness that's great for concentration is rough for a social life. Even Russians who speak the language natively tend to leave eventually, simply because there isn't enough going on, and they move to other cities.
Brain drain hangs over the whole project. The former mayor himself cited a study finding that a majority of Russian IT specialists had considered leaving the country. Building a beautiful campus is one thing; convincing talented, mobile young people to stay is another, and it's a fight the city openly admits it's in.
There's a quieter democratic wrinkle too: because most people here rent, they aren't counted as permanent residents and can't vote in local elections. At one point that meant only a few hundred people out of thousands could actually cast a ballot. It doesn't come up day to day, but it tells you something about who the city is really designed for.

People I knew were already leaving

The brain-drain part wasn't an abstraction to me. Two of the people I knew in Innopolis were Stas and Sasha Litvinov — Stas was a teaching assistant and advisor during my classes, and he'd been part of the place since before the city officially existed, including a stint in Pittsburgh on the university's joint program with Carnegie Mellon. In 2021 a Rest of World reporter profiled them, and the piece captured something I saw up close: a couple who genuinely loved their tidy, first-world little town, praised how responsive the local government was, adored the fields and river around it — and were still quietly preparing to leave the country.

Fighting brain drain and creeping authoritarianism in Russia’s techno-utopian village - Rest of World

Facing a dire shortage of qualified tech workers, companies and the Kremlin are vying to win the hearts and minds of Russia’s entrepreneurs and engineers.

restofworld.org

Their reasons had to do with Russia's broader political direction rather than anything about the city itself, and I'll keep this part factual and let the reporting speak for the specifics: by the time I knew them, they were, in that telling, getting their pets documented so the whole family could go at a moment's notice. Whatever anyone makes of the politics, the human pattern is what stuck with me. Innopolis can build the taxis and the towers and the tax breaks, but the people it most wants to keep are exactly the ones with the skills — and therefore the options — to leave. I watched good people quietly run that calculation in real time. Eventually, in my own way and for my own reasons, I ran it too.

Staying to work: my years at Portavita

Here's the bet the whole city is built on: that graduates won't just study here, they'll stay and work. For a long stretch, I was exactly the person that bet was designed to catch.

Part of that wasn't even optional — the scholarship came with a contract, and graduates were expected to work for a company in Russia for at least a year after finishing. But the way I actually landed my job was almost gentle: the university runs job fairs where companies come and pitch opportunities directly to students, and that's where I found Portavita.

Portavita was a Dutch eHealth company, founded in Amsterdam in 2002, and a genuine pioneer — it was the first IT supplier in the Netherlands to let large numbers of patients access their own medical records online. Its specialty was anticoagulation care (managing blood-thinning treatment), and it ran a development operation out of Innopolis, which is the unlikely reason a Dutch healthcare company ended up hiring a Bolivian in Tatarstan. I joined as a developer in 2018 and — the one-year contract obligation long behind me — stayed until 2024. I have zero regrets about that. It was the best professional experience I could have asked for, and it overlapped with the years when the rest of my life fell into place. It was during this stretch that I met my wife — on Tinder, as it happens. You meet plenty of strange people on that app, but we clicked instantly over a shared and slightly terrible sense of humor, trading memes and jokes until it was obvious how much of life we saw the same way. She was a local, and she gave me something the university bubble never could: Kazan, seen through a resident's eyes. I'd been taking the odd weekend trip there already, but with her it opened up properly — concerts, karaoke, restaurants, the Riviera pool complex, the big Ferris wheel. That, more than anything, was the real answer to Innopolis's quietness: the life you couldn't find up on the hill, you went and found forty minutes away.

Fittingly for this story, she had her own exit planned — a move to Saint Petersburg, chasing bigger opportunities — right around when we met. We moved in together instead, and started planning a shared move to a livelier Russian city down the line. (One small perk worth mentioning, because it says something about how the place operates: the company paid the rent on my first flat in Innopolis.)

The work was substantial. I led the Patient Assistant service, which supported patients on anticoagulation treatment across Dutch healthcare institutions; it launched in 2018 and grew to process thousands of patient records a month. We eventually split it into three microservices handling data processing, enrichment, and delivery through FHIR-compliant REST APIs (FHIR being the international standard for exchanging health data). Later I worked on the DVZA services platform (2020–2024), which processed millions of patient records while staying compliant with MedMij — the Dutch national standard for patient health-data exchange — and FHIR. Under the hood it was a proper backend playground: Scala and Akka (including Akka HTTP), Slick over PostgreSQL with Citus, and Kafka / Kafka Streams for messaging, with React on the front end.

And then the ground shifted. Portavita had been acquired in 2021 by CompuGroup Medical (CGM), a large German healthcare-software group. Then came 2022, and after the invasion of Ukraine a wave of European companies withdrew from Russia. Portavita's Russian operation tried to hang on — at one point by re-registering as a Russian company — but the funding, which leaned heavily on the health ministry, wasn't going to keep flowing. The writing slowly appeared on the wall. I left Russia in October 2023, and in 2024 I moved to Santiago, Chile, where I'm writing this from now.

Final thoughts

The idea of building a city around an industry, in a specific region, isn't new. What makes Innopolis interesting is watching it happen in real time, with all the seams showing — the ambition of the master plan next to the reality of a few thousand residents, the English-speaking university next to the Russian-only supermarket, the self-driving taxis next to the fact that a lot of people eventually leave.

I know that last part firsthand, because I became one of them. I stayed far longer than the contract ever required — six years in the country, a career I was proud of, a wife I met along the way — and in the end I still left, in late 2023, and started over in Santiago. That's not a knock on the place. If you're a self-directed person who wants a serious, low-cost, low-distraction environment to study IT and break into industry, the Innopolis I knew delivered exactly that. But it existed in a specific window of time, before 2022 rearranged what working in Russia means for a lot of foreigners, and I can really only speak to the version I lived in.

So, no tidy verdict. Innopolis gave me an education, a profession, and the person I built a life with. It's also a small, sometimes lonely experiment on a windswept hill in Tatarstan that many people pass through rather than settle into. Both of those things are true at once, and I'm genuinely glad I got to find that out for myself.

If you've studied or worked somewhere built around a single industry — a company town, a tech hub, a university city — I'd love to hear how your experience compared. Drop a comment below.

AI Made Code Free. So Why Are the Giants Still Winning? (And where solo devs actually beat them)

krlz — Wed, 01 Jul 2026 21:51:10 +0000

Everyone keeps saying AI will let a solo developer take down the giants. And everyone keeps saying the giants will just absorb everything. Both takes are wrong, and I spent a while reading the actual 2025 data to figure out why.

I pulled from four of the biggest developer datasets of the year:

DORA 2025 State of AI-Assisted Software Development (Google Cloud, ~4,867 respondents)
Stack Overflow 2025 Developer Survey (49,009 respondents)
GitHub Octoverse 2025 (behavioral data across 180M+ developers)
JetBrains State of the Developer Ecosystem 2025 (24,534 developers)

Here's the honest synthesis. It's more useful than either hype narrative.

The one-sentence thesis

AI collapsed the cost of writing software to near zero. It did not collapse the cost of distribution, trust, support, or being liable when it breaks — and those are ~80% of what a software business actually is.

So the effect isn't "solos beat giants." The effect is that the middle got hollowed out. The 10-person, VC-funded, me-too startup building a feature is the loser of this era — squeezed from below by a solo who ships the same thing for free, and from above by a giant who bundles it. Solos and giants both survive. The undifferentiated middle doesn't.

"AI is an amplifier, not an equalizer"

This is the single most important finding of 2025, and it comes straight from DORA:

"AI's primary role in software development is that of an amplifier. It magnifies the strengths of high-performing organizations and the dysfunctions of struggling ones."

Read quickly, that kills the "AI levels the playing field" fantasy. AI rewards whoever already has good practices — not whoever is scrappiest.

But read one layer deeper and it becomes the best available argument for the small team. DORA found the key enabler is independence of action — "the ability to develop, test, and deploy value independently, with little or no coordination cost." In an Adidas pilot they cite, teams in loosely-coupled architectures saw 20–30% productivity gains; teams tightly coupled to legacy ERP saw little or no AI benefit at all.

Who has maximum independence and zero coordination cost? A solo founder, by definition. Who drowns in coordination overhead? A giant. So the amplifier finding, taken seriously, says: AI most benefits whoever has the least organizational drag. The giant still wins on distribution, capital, and enterprise trust — but not on per-engineer AI leverage. Their own bureaucracy taxes it.

Everyone uses AI. Fewer and fewer trust it.

Adoption is settled. Four sources, one tight band:

Source	AI adoption
DORA	90% use AI at work
JetBrains	85% regularly use AI
Stack Overflow	84% use or plan to (up from 76%)
GitHub (behavioral)	~80% of new devs use Copilot in week one

But here's the twist the hype cycle skips. Trust is going the opposite direction. Stack Overflow 2025:

Favorability toward AI fell 72% → 60% year over year.
46% now distrust AI accuracy vs. only ~33% who trust it (distrust was 31% a year ago).
66% cite "AI solutions that are almost right, but not quite" as a top frustration.
~45% say debugging AI-generated code takes more time.

Stack Overflow's own three-word summary — "willing but reluctant" — is the truest description of the 2025 developer relationship with AI. Everyone uses it. Fewer believe it.

Agentic coding is a frontier, not the norm (yet)

If you live on Twitter you'd think everyone is running autonomous agents. The data says otherwise:

DORA: 61% "never" use agent mode — the least-adopted way to use AI.
Stack Overflow: only ~31% currently use AI agents; 38% don't plan to.
JetBrains: 85% use AI, but only 44% have integrated it into their workflow — "AI use isn't systematized. Developers are just using it ad hoc."

And yet the frontier is exploding: GitHub logged 1M+ pull requests from its Copilot coding agent in just May–September 2025.

So the "solo operating at team scale via agents" story is real but leading-edge, not median. The good news for founders: the leading edge is exactly where competitive solo builders already live.

The unifying insight: usage ≠ results

Put all four datasets together and one theme dominates:

AI adoption is universal. Converting it into results is not. The differentiator is discipline and systematization — not access to the tool.

DORA says it as "amplifier, not equalizer." JetBrains says it as "85% use it, 44% have integrated it." Stack Overflow says it as "66% fight almost-right output." Since everyone now has the tool, the tool is not the edge. The practices around it are — taste, review discipline, distribution, trust. That's the exact terrain where a solo-vs-giant outcome is actually decided.

The counter-hype stat you should tattoo somewhere

DORA cites the now-famous METR study: experienced open-source developers were measured 19% slower with AI tools — while believing they were 20% faster.

If you're a solo founder, you are the least-checked person in your own loop. Nobody's there to catch you shipping fast-feeling, actually-slower work. Which is why, for a solo, tests and small batches aren't bureaucracy — they're the substitute for the code review you no longer have.

Clean architecture is now a business decision

Here's the part that surprised me most. AI makes architecture matter more, not less:

AI generates plausible code fast, so you accumulate unmaintainable "slop" faster. DORA found friction doesn't vanish — it moves, "from manual grind to deciding and verifying... assessing code that looks remarkably similar to correct code."
The single biggest risk to a solo AI business isn't competition — it's platform risk. If your model calls sit behind a clean adapter interface, a provider price hike is a config change. If they're smeared through your codebase, it's an existential rewrite.
DORA found frequent commits + easy rollback act as a "psychological safety net" that makes high-velocity AI work safe.

For a solo in 2026, boring, provider-agnostic, modular-monolith architecture is a survival tool. Running microservices as a one-person team is cosplaying as Google and paying the ops tax for nothing.

The map is shifting to the Global South

The competitive geography is changing fast (GitHub Octoverse 2025):

GitHub crossed 180M+ developers, adding ~36M in 2025 — "a new developer every second."
Net-new devs by region: APAC +13M, Europe +6.3M, Africa & Middle East +3.4M, LATAM +3.2M.
India added 5.2M devs (14% of all new accounts), and already has the largest open-source contributor base in the world — projected to pass the US in total developers around 2030.
Brazil is #4 globally (6.89M). India, Brazil, and Indonesia more than quadrupled their dev counts in five years.

The next decade's solo founders are disproportionately Indian, Brazilian, Indonesian, and African — not Californian. And that matters, because the strongest wedge against a US giant is "not-American, self-hostable, cheaper, localized, compliant with our law."

If you're building from Latin America (I am), the playbook writes itself:

Cost base local, revenue base global. Own the one thing US giants refuse to build — deep local compliance, payments, and language integration (electronic invoicing, local payment rails like Webpay/Flow/MercadoPago, real Spanish/Portuguese product sense). On a purchasing-power basis, a competent LATAM solo dev in 2026 is in a better risk-adjusted position than a mid-level engineer at an SF startup.

Open source is bigger than ever — and more fragile than ever

Both things are true at once (GitHub Octoverse):

255,000 new open-source contributors in March 2025 — the largest single month on record. 1.128B contributions (+13% YoY).
4.3M AI-related repos; generative-AI monthly contributions peaked at 6.28M in June 2025 (+188% YoY).
But only ~63% of public repos even have a README. The maintenance layer is thinner than ever.

And the licensing war (Elastic→OpenSearch, HashiCorp→OpenTofu, Redis→Valkey) settled on one pragmatic lesson: open source is a distribution and trust tactic, not a business model. Every solo OSS success — Plausible, Sidekiq, Ghost, PostHog, Supabase — monetizes hosting, a commercial tier, or a foundation. None sells the source. Source-available/commercial licensing is respected now. You don't owe AWS a free gift.

The solo playbook for 2026

Pick a niche too small for a giant to defend, big enough for one person to live on (a ~$1–5M/yr TAM: your career, their rounding error).
Decide monetization before you open the source. "Figure it out later" kills companies.
Build in the open for trust; license to stay alive.
Keep architecture provider-agnostic and boring. It's a business decision.
Localize where giants won't — language, payments, compliance.
Assume you'll be Sherlocked. If your only moat is the feature, the giant ships it next quarter. Build a relationship moat they can't clone.
Believe the measured results, not the feeling. (Remember the METR 19%.)

TL;DR

AI made building nearly free; distribution, trust, and support are still expensive — and that's where the contest actually happens.
AI is an amplifier, not an equalizer. Its benefit is gated by independence of action — the axis where solos structurally beat giants.
Adoption is universal; trust is falling. "Willing but reluctant."
Agentic coding is frontier, not median — but the frontier is where founders win.
Clean architecture is a survival tool, because platform risk is the real killer.
The builder base is going global — and localization is the wedge giants won't touch.

The giants own the pipes. But the pipes were never the interesting part. Taste, trust, community, and focus were — and those still don't come in an API.

Sources: DORA 2025 State of AI-Assisted Software Development; Stack Overflow 2025 Developer Survey; GitHub Octoverse 2025; JetBrains State of the Developer Ecosystem 2025; METR "Impact of Early-2025 AI on Experienced Open-Source Developer Productivity." All figures are self-reported perceptions unless noted; DORA is explicitly observational, not causal. What's your read — is the middle really getting hollowed out? I'd love pushback in the comments.

Spec-Driven Development in 2026: What It Is, the Tooling, and How Teams Actually Use It

krlz — Fri, 19 Jun 2026 15:20:48 +0000

A field guide to the practice that's reshaping how software gets built with AI agents.

TL;DR — Spec-Driven Development (SDD) makes a precise, executable specification the source of truth and treats code as a generated, verifiable artifact. The spec declares intent; the code realizes it. In 2026 it went mainstream because AI agents are great at writing code and terrible at guessing what you meant.

Jump to: Why now · Specs vs. executable specs · Maturity model · Workflow · Tooling · EARS · Worked example · Caveats · Bottom line

Why now? The "vibe coding" backlash

The movement defines itself against "vibe coding" — the term Andrej Karpathy popularized in early 2025 for loosely prompting an AI and shipping whatever comes back. Vibe coding is great for throwaway prototypes and miserable for anything that has to be maintained.

SDD is the disciplined counterweight: if AI writes most of the code, then the specification becomes the highest-leverage artifact a human produces. The skill that matters shifts from typing the implementation to defining the intent precisely enough that a machine can't get it wrong.

Raw specs vs. executable specs

This is the single most important distinction in the whole topic — and the one most "SDD explainers" skip.

	Traditional design docs	SDD specs
Read by	Humans	Humans and agents
Enforcement	Advisory — devs may diverge	Executable — tests fail on drift
Lifecycle	Goes stale, becomes archaeology	Living, continuously validated
Lives in	A wiki nobody opens	The repo + CI/CD

"Traditional specs are read by humans, while SDD specs are executed as BDD scenarios, API contract tests, or model simulations."
— Deepak Babu Piskala, Spec-Driven Development: From Code to Contract in the Age of AI Coding Assistants (arXiv, Jan 2026)

[2602.00180] Spec-Driven Development:From Code to Contract in the Age of AI Coding Assistants

The rise of AI coding assistants has reignited interest in an old idea: what if specifications-not code-were the primary artifact of software development? Spec-driven development (SDD) inverts the traditional workflow by treating specifications as the source of truth and code as a generated or verified secondary artifact. This paper provides practitioners with a comprehensive guide to SDD, covering its principles, workflow patterns, and supporting tools. We present three levels of specification rigor-spec-first, spec-anchored, and spec-as-source-with clear guidance on when each applies. Through analysis of tools ranging from Behavior-Driven Development frameworks to modern AI-assisted toolkits like GitHub Spec Kit, we demonstrate how the spec-first philosophy maps to real implementations. We present case studies from API development, enterprise systems, and embedded software, illustrating how different domains apply SDD. We conclude with a decision framework helping practitioners determine when SDD provides value and when simpler approaches suffice.

arxiv.org

The maturity model: three levels

Almost every serious 2026 source converges on the same ladder. Pick the rung you actually need — not the most aggressive one.

Spec-First — Specs seed the initial generation, then code is allowed to drift. Good for AI-assisted features and prototypes. Low overhead, no long-term guarantees.
Spec-Anchored — Specs and code evolve together as living documentation; automated tests enforce alignment. Piskala's paper calls this "the sweet spot for most production systems."
Spec-as-Source — Humans edit only the spec; code is fully generated and never hand-edited. Eliminates drift by design — but demands mature, trusted generation tooling. Still aspirational for most teams.

flowchart TD
  A["<b>Spec-First</b><br/>specs seed generation,<br/>code is allowed to drift"]
  B["<b>Spec-Anchored</b> 🎯<br/>specs + code evolve together,<br/>tests enforce alignment"]
  C["<b>Spec-as-Source</b><br/>humans edit only specs,<br/>code is fully generated"]
  A -->|"add living tests<br/>+ CI enforcement"| B
  B -->|"trust generation,<br/>stop hand-editing code"| C
  style B fill:#dff3e4,stroke:#2ea44f,stroke-width:2px

🎯 Recommendation: aim for spec-anchored. Spec-as-source is where the hype lives; spec-anchored is where the value is today.

The canonical workflow

Every major tool implements roughly the same pipeline, with a human review gate at each phase boundary to stop drift before it compounds:

flowchart LR
  C[constitution] --> S[specify] --> CL[clarify] --> P[plan] --> T[tasks] --> I[implement] --> A[analyze]

Phase	What happens
Constitution	Project-wide rules the agent must always obey (language, frameworks, testing, deps)
Specify	What and why: user stories + acceptance criteria. No tech choices yet
Clarify	The agent surfaces ambiguities before any planning
Plan	How: architecture, data models, technical decisions
Tasks	Decompose the plan into atomic, independently-shippable items
Implement	Execute tasks, verifying each against its acceptance criteria
Analyze	Cross-check spec ↔ plan ↔ tasks for consistency

⚠️ Golden rule: never skip from spec straight to code. Review the plan before task decomposition; review the tasks before implementation.

The tooling landscape (2026)

Tool	Strengths	Best for
GitHub Spec Kit	Open-source, model-agnostic, the reference implementation	Teams avoiding vendor lock-in
AWS Kiro	Agentic IDE with automated guardrails ("hooks"), deep AWS integration	AWS-native / serverless shops
Claude Code (cc-sdd)	Native `/sdd:specify`, `/sdd:plan` slash commands, terminal-first	Solo devs, CLI workflows
Cursor (Plan Mode)	IDE-first, inline diff review, MCP support for Spec Kit	Teams prioritizing UX
OpenSpec	Lightweight, framework-agnostic, Markdown + YAML	Indie devs, minimal tooling
BMAD-METHOD	Community methodology, constitution + multi-agent role-play	Teams wanting a flexible framework
Tessl	Compliance-focused, audit trails, regulated templates	Fintech, healthtech
Google Antigravity	"Agent-first," specification-constrained autonomy	Teams exploring deep agent autonomy

The open-source reference everyone benchmarks against:

github / spec-kit

💫 Toolkit to help you get started with Spec-Driven Development

🌱 Spec Kit

<div class="markdown-heading" dir="auto">

Build high-quality software faster.

An open source toolkit that allows you to focus on product scenarios and predictable outcomes instead of vibe coding every piece from scratch.

🤔 What is Spec-Driven Development?

Spec-Driven Development flips the script on traditional software development. For decades, code has been king — specifications were just scaffolding we built and discarded once the "real work" of coding began. Spec-Driven Development changes this: specifications become executable, directly generating working implementations rather than just guiding them.

⚡ Get Started

1. Install Specify CLI

Requires uv…

View on GitHub

Supporting standards & frameworks that predate the AI wave but power SDD

BDD: Cucumber, SpecFlow, Behave (Gherkin / Given-When-Then)
API contracts: OpenAPI/Swagger, GraphQL SDL, Protocol Buffers, AsyncAPI
Contract testing: Pact, Specmatic
Model-based design: Simulink, SCADE

None of these are new. SDD's contribution is wiring them together as the primary artifact an AI agent generates from — not the documentation you write afterward.

30-second tool-selection rubric

Already in Claude Code? → Spec Kit + cc-sdd skills
In Cursor? → Plan Mode + AGENTS.md + Spec Kit via MCP
AWS shop? → Kiro
Regulated industry? → Tessl
Want minimal ceremony? → OpenSpec

EARS: how to write specs an LLM can't misread

EARS (Easy Approach to Requirements Syntax) is the de facto standard for acceptance criteria that are unambiguous to humans and models. Five patterns cover almost everything:

Pattern	Template	Example
Ubiquitous	The system shall …	The system shall log every auth attempt.
Event-driven	WHEN … THE system SHALL …	WHEN a user submits the login form THE system SHALL validate credentials.
State-driven	WHILE … THE system SHALL …	WHILE a sync is running THE system SHALL show a progress indicator.
Unwanted behavior	IF … THEN …	IF validation fails 3× THEN lock the account for 15 min.
Optional	WHERE … THE system SHALL …	WHERE MFA is enabled THE system SHALL require TOTP.

The payoff: criteria written this way map almost 1:1 onto test cases — which is exactly what makes the spec executable rather than advisory.

A worked example: "magic-link login"

Talk is cheap; here's a (trimmed) end-to-end slice so the artifacts are concrete.

📄 spec.md (excerpt)

## Feature: Passwordless magic-link login

### Acceptance criteria (EARS)
- WHEN a user submits a valid email
  THE system SHALL send a one-time login link valid for 15 minutes.
- IF a login link is used more than once
  THEN the system SHALL reject it with HTTP 410 Gone.
- WHERE the email is not associated with an account
  THE system SHALL still return HTTP 202 (no account enumeration).
- THE system SHALL store link tokens hashed, never in plaintext.

### Out of scope
- Social login, SSO, password fallback.

🏗️ plan.md (excerpt)

## Stack
- Node 22 + Fastify, Postgres 16, Redis for token TTL.

## Data model
- magic_tokens(id, user_id, token_hash, expires_at, consumed_at)

## Decisions
- Tokens = 32 bytes CSPRNG, stored as SHA-256 hash.
- 202-for-unknown-email enforced at the controller layer.

✅ tasks.md (excerpt)

- [ ] T1  migration: magic_tokens table
- [ ] T2  POST /auth/magic-link  (issue + email)   refs spec §1, §3
- [ ] T3  GET  /auth/verify      (consume + session) refs spec §2
- [ ] T4  contract tests for 202 / 410 paths
- [ ] T5  rate-limit issue endpoint (5/min/IP)

Notice the thread: each task cites the spec clause it satisfies. That traceability is the whole point — when a test fails, you know which intent broke.

Recommendations & best practices

Constitutional foundations

Commit AGENTS.md or .specify/memory/constitution.md before the first spec.
Codify project-wide decisions as ubiquitous EARS statements: "The system shall use TypeScript strict mode."

Specification discipline

One feature = one spec directory (specs/NNN-feature-name/).
Keep specs to 1–3 pages; split if larger.
Use explicit "out of scope" sections to bound the agent's exploration.
Write in domain language (business intent), not implementation detail.

Phase boundaries

Never jump spec → code. Review the plan, then review the tasks.

Documentation & traceability

Cite specs in commits: feat(auth): magic link, refs specs/004-magic-link/spec.md.
Treat specs as durable — they outlive the generated code.
Run pre-flight /checklist passes for security, accessibility, observability.

The honest caveats

SDD has real skeptics, and they're worth more than the hype:

It can be old wine in new bottles. Brandon Kindred's "Same Patterns, New Hype" (2026) argues SDD is largely waterfall/contract-design rebranded — and that "the value is the thinking you do while writing the spec, not the tooling around it."
Drift and hallucination don't disappear. Thoughtworks stresses that executable code remains the source of truth requiring maintenance, explicitly rejecting the "specs alone suffice" view. They place SDD in the "Assess" ring of their Technology Radar — not "Adopt."
Over-specification defeats the purpose. When a spec becomes pseudo-code, you've written the program twice.
False confidence. Matching a wrong spec satisfies no real requirement.
Tooling complexity can add ceremony without proportional value.

✅ When SDD is worth it / ❌ when to skip it
Worth it: using AI coding assistants · complex requirements · multiple maintainers · integration-heavy or regulated systems.

Skip it: throwaway prototypes · solo, short-lived projects · exploratory work where requirements are still unknown.

What the data says (read with a skeptic's eye)

Vendor-reported numbers should be treated as directional, not proven. With that caveat: early adopters (GitHub, AWS) report meaningfully higher first-pass success from agents working off a good spec, and Piskala's paper cites error reductions on the order of tens of percent when LLMs work from refined specs versus loose prompts. The consistent, less-disputable finding is a shift in where humans spend time — away from typing implementation, toward review and clarification.

If you take one number away, make it this one: the spec is now where the thinking happens.

The bottom line

In 2026, SDD is best understood not as a tool but as a discipline: when agents write the code, the spec is the most leveraged thing a human can produce. The pragmatic path:

Target spec-anchored, not spec-as-source.
Write acceptance criteria in EARS.
Keep code as the source of truth and tests as the enforcer.
Use the tooling (Spec Kit, Kiro, cc-sdd) to enforce the loop — not to replace the thinking.

Building with SDD already? What's worked and what's been pure ceremony? Drop it in the comments. 👇

📚 Sources

Compiled June 2026.

The 6-Month Clock Ran Out — Are Companies Full AI Yet?

krlz — Mon, 06 Apr 2026 15:17:51 +0000

So during this year, many changes have been introduced in software development. A big disruption is coming, and some new terminology — SDD, IDD, Cognitive Debt, DORA, the "new" XP, Agentic Programming, and others — is being promoted, sometimes used as a slogan.

I personally feel it is many times used as a strong advertising campaign, taking advantage of the panic of people feeling they might be about to lose a train, with severe consequences in their professional careers if they don't catch up. This might be true to a certain level.

But let's look at the actual numbers: in early 2026, AI coding tools have reached 90% adoption among developers, yet only 1% of companies consider themselves mature in AI usage. Everyone has the tools — almost nobody has figured out the practices.

The debate between "AI enthusiasts" and "fundamentals defenders" misses the point entirely. The real question isn't whether to use AI tools, but how companies must restructure their practices, teams, and culture to use them without accumulating what Thoughtworks now calls cognitive debt.

So let's cut through the noise.

What the Data Actually Says

Google's 2025 DORA State of AI-Assisted Software Development report is probably, for now, the most rigorous report made about the current experience of AI in companies. It surveyed nearly 5,000 technology professionals worldwide and includes over 100 hours of qualitative interviews, combining both quantitative and qualitative data. Some clear conclusions can be taken from it: the real value comes from platform quality, workflow clarity, and team alignment. The organizational capability to use AI well is the differentiator — not the tools themselves.

AI is an amplifier, not a fixer. Strong teams get stronger. Struggling teams get worse faster.

The numbers tell the story:

90% of software professionals now use AI tools
80%+ say AI enhanced their productivity
30% report little or no trust in AI-generated code
Seven organizational capabilities were identified that magnify the positive impact of AI

That last point is key. It's not about which AI tool you pick. It's about what your team already looks like before AI enters the picture.

From Vibes to Discipline

So if the tools aren't the answer, what is? To understand where things are heading, it helps to see where they just came from.

In February 2025, Andrej Karpathy — co-founder of OpenAI and former AI leader at Tesla — posted on X about a new way of coding: "fully give in to the vibes, embrace exponentials, and forget that the code even exists." The post got over 4.5 million views. Within weeks, the term "vibe coding" had spread from social media to The New York Times, Ars Technica, and The Guardian. By November, Collins Dictionary named it Word of the Year 2025. By mid-2025, The Wall Street Journal reported professional engineers were adopting it for commercial use cases.

But the honeymoon didn't last. By late 2025, Thoughtworks Technology Radar observed the concept fading in favor of more disciplined approaches — concerns about code quality, security, and maintainability in production couldn't be vibed away.

What's replacing it? Three approaches are gaining traction:

Intent-Driven Development (IDD) flips the focus: the quality of your specification determines the quality of your outcome. Teams restructure around 60% product judgment, 30% engineering architecture, and 10% design precision. The specification becomes the primary artifact, not the code.

Spec-Driven Development (SDD) takes a similar angle but draws a clearer line: humans own the design, AI owns the implementation. The key discipline is writing specifications good enough that AI can execute them correctly.

Agentic Engineering, coined by Karpathy himself in early 2026 as the successor to vibe coding, is about designing systems where AI agents plan, write, test, and ship code under structured human oversight. The risk it addresses is real: an agent writing 1,000 PRs/week with a 1% vulnerability rate creates 10 new vulnerabilities weekly.

All three share a common thread: the human's job shifts from writing code to defining intent clearly and verifying outcomes rigorously.

The Return of XP

Here's something nobody expected: Extreme Programming is back.

Kent Beck describes TDD as a "superpower" when working with AI agents. The reasoning is simple — unit tests are the most reliable way to prevent AI from introducing regressions. When AI generates code, the test suite becomes your verification layer. Not code review. Not manual testing. Tests.

The Pragmatic Engineer newsletter and the Thoughtworks Deer Valley retreat both highlight this resurgence. The practices that felt "too rigorous" five years ago are now the safety net that makes AI velocity sustainable. It turns out the answer to "how do we trust AI-generated code?" was already here — we just stopped using it.

What the Industry Leaders Are Saying

In February 2026, approximately 50 tech leaders — including Kent Beck and Martin Fowler — gathered at Thoughtworks' Deer Valley retreat. It was the 25th anniversary of the Agile Manifesto, and the question on the table was direct:

"If AI handles the code, where does the engineering actually go?"

Their conclusions reshape how we think about engineering teams:

Cognitive debt is the new technical debt. When AI generates code that nobody fully understands, you accumulate cognitive debt — and it compounds just like technical debt. "Velocity without understanding is not sustainable."

Agent Topologies. Just as Team Topologies redesigned how human teams interact, organizations now need to design how AI agents fit into their workflows. Who owns what the agent produces? How do agents interact across teams?

Tiered Code Review. Not all code needs the same scrutiny. Reserve human reviews for high-risk components and rely on strong CI/CD guardrails for the rest.

Staff engineers become supervisors. The senior role shifts toward agent orchestration and governance — designing the systems within which AI operates, not writing the code it produces.

Teams Are Shrinking, Roles Are Shifting

These aren't just theoretical changes. They're already reshaping how companies hire and organize.

Teams are shrinking dramatically. One Series C startup restructured from 12 engineers to 3 using AI tools, with a 40% increase in velocity. New roles are appearing — AI Workflow Engineers, PromptOps Specialists, Agent Orchestration leads — titles that didn't exist two years ago.

The old hiring model — bring in juniors cheaply, train them up — is being replaced by hiring fewer senior people and equipping them with AI tools. AI widens the gap between strong and weaker engineers rather than leveling the playing field.

Interview practices are changing too. Organizations are replacing generative coding tests with Review Simulations — candidates audit pre-generated, flawed codebases. The skill being tested: can you catch what AI gets wrong?

And there's a quiet crisis brewing for mid-level engineers. New graduates are more productive with AI tools — they have no habits to unlearn. Senior engineers benefit from deep experience to guide AI. Mid-career professionals risk being squeezed from both sides. Addy Osmani from Google warns this is one of the most underappreciated shifts happening right now.

Measuring What Matters

With all these changes, the old metrics stop making sense. When AI can generate unlimited code, measuring output becomes meaningless.

What We Used to Measure	What Actually Matters Now
Velocity / story points	Cycle time
Lines of code	Lead time to production
PRs merged	Defect escape rate
Story completion rate	Cognitive debt indicators

The shift is from output to outcomes: how fast does value reach users, and how reliably?

The Fundamentals Debate Is Over

If there's one thing every major source — DORA, Thoughtworks, Gartner, Google — agrees on, it's this: AI makes fundamentals more important, not less.

You need to understand architecture to design scalable systems and guide AI effectively. Developers who skip algorithms, system design, or manual debugging "lose problem-solving muscle memory."

"The best software engineers won't be the fastest coders, but those who know when to distrust AI." — Addy Osmani, Google

Gartner backs this up with a stark warning: by 2028, prompt-to-app approaches by citizen developers will increase software defects by 2,500%. The quality crisis won't come from teams that invest in fundamentals — it will come from those that skip them.

As Satya Nadella put it: "Getting the fundamentals right matters a lot." This isn't a compromise between AI and fundamentals. It's the only approach that works.

Where to Start

If you're leading a team or a company, here's a practical starting point:

Read the DORA 2025 AI Report (free). Assess where your organization actually stands.
Invest in testing culture before AI tooling. TDD, integration tests, strong CI/CD — these are what make AI velocity safe.
Adopt Spec-Driven Development. Better specifications mean better AI output. It's that direct.
Tier your code review. Human review for high-risk paths, automated guardrails for the rest.
Rethink how you hire. Test for review and debugging skills. Can candidates spot what AI gets wrong?
Measure outcomes, not output. Cycle time and defect escape rate over velocity and story points.
Read the Thoughtworks Deer Valley retreat takeaways (free PDF). It's the most forward-looking strategic document on this topic.

Resources

The tools are here. The question was never "should we use AI?" — it was always "do we have the engineering culture to use it well?" The answer to that question determines everything.

Britain Forces Vets to Publish Prices as a Global Reckoning Over Pet Care Costs Takes Shape

krlz — Wed, 25 Mar 2026 09:09:20 +0000

The Bill Comes Due

On Monday, the United Kingdom's Competition and Markets Authority did something no government body had ever done at this scale: it told every veterinary practice in the country to post its prices — or face consequences. Prescription fees will be capped at £21 for a first medication and £12.50 for each additional one. Any treatment expected to cost more than £500 will require a written estimate, itemized in full. And by December, every clinic must publicly disclose who actually owns it — a pointed reform in a market where corporate consolidation has quietly swallowed independent practices while pet owners had no idea.

The reforms, which become legally binding by September 23, 2026, are the conclusion of a sweeping market investigation that found UK veterinary fees have risen at nearly twice the rate of inflation in recent years. For the 12 million dog-owning households and 11 million cat-owning households across Britain, the announcement felt overdue.

But the UK's intervention is not happening in isolation. Across Europe, Asia, and the Americas, a remarkably synchronized reckoning is underway — over what pet care costs, who controls the pet economy, and what rights animals themselves should have. This week's news from nine countries reveals a world grappling, often simultaneously, with the same uncomfortable questions about the creatures we claim to love most.

When Loving Your Pet Means You Can't Afford Your Pet

The UK pricing reforms would be significant on their own, but they land in a global context that gives them particular resonance. In Germany, where 34 million pets live in nearly half of all households, veterinary associations are bracing for the opposite outcome. The GOT — Germany's official veterinary fee schedule — is due for a formal government evaluation in the second quarter of 2026, and the president of the Bundesverband Praktizierender Tierärzte has already declared that the profession will "vehemently oppose any fee reductions" and that prices "must actually rise further."

The median gross salary for an employed German veterinarian is €4,650 per month — respectable but hardly lavish given the years of education required. Simple consultations already run between €30 and €80, with complex operations reaching €2,500 or more. For pet owners hoping for relief, the evaluation may deliver the opposite.

In a world where a routine surgery can cost more than a month's rent, the question is no longer whether people love their pets — it's whether they can afford to.

In Russia, pet owners are spending an average of 4,482 rubles per month (roughly $45), and the veterinary services market hit a record 60 billion rubles in 2025. Yet only about one-third of Russia's 80 million pets are vaccinated — a gap driven partly by cost but also by the absence, until now, of any system to even track who owns what animal. That is about to change dramatically.

Meanwhile in Chile, the pet care market has ballooned to $1.94 billion — a 77% increase in just five years — with pet store numbers tripling since 2019. Chilean owners now spend up to 100,000 pesos (about $100) per month per pet. And in Japan, despite a slowly shrinking pet population, the veterinary market is projected to nearly double from $2.9 billion to $4.8 billion by 2034, driven by aging pets that need chronic disease management and an expanding pet insurance sector that's making advanced care accessible — and expected.

The pattern is clear across every market: people are spending more, often much more, and the infrastructure of pet care — from pricing transparency to insurance access to pharmaceutical competition — is struggling to keep pace with what has become, for millions of families, a non-negotiable expense.

One bright spot for dog owners: Elanco Animal Health is preparing to commercially launch Befrena (tirnovetmab), a new anti-IL31 monoclonal antibody injection for canine allergic dermatitis approved by the USDA in late December 2025. With dosing intervals of six to eight weeks compared to the four-to-eight-week window of Zoetis's market-dominant Cytopoint, a second competitor in this space could eventually exert downward pressure on one of the most common — and most expensive — reasons dogs visit the vet.

The Shelter Crisis No One Saw Coming

If the cost of keeping a pet is straining families, the cost of giving one up is straining something else entirely: the world's animal shelters.

A study published this month by the University of Bristol, in collaboration with Reaseheath College and University College Dublin, revealed that euthanasia rates for stray dogs in the UK have more than tripled in just three years — from 1.9% in 2021 to 6.3% in 2023. Shelter intake surged from 16,310 dogs to 23,287 over the same period. While rehoming rates improved from 41% to 53%, the sheer volume is overwhelming a system in which only 27% of local authorities even have written welfare policies.

A specific breed dynamic is at play: the UK's ban on XL Bullies, enacted in early 2024, has driven a wave of surrenders. Bull breeds now dominate shelter populations, stay longer, and are harder to place — creating a bottleneck that pushes the entire system toward crisis.

Across the Atlantic, the picture is no better. Approximately 5.8 million dogs and cats entered U.S. shelters and rescues in 2025. Adoptions are down. Intakes are up. Some facilities have resorted to housing animals in pop-up crates, converted bathrooms, and staff offices. As municipal shelters restrict intake to manage overflow, a secondary crisis has emerged: informal rescues and nonprofits stepping in with good intentions but limited resources, sometimes tipping into hoarding situations with little oversight.

In Russia, a shelter scandal in the Arkhangelsk region laid bare the consequences of institutional neglect. An investigation at a facility in the village of Povrakulskaya found dogs chained outdoors in freezing temperatures, cats crammed into tiny cages, animals lying on frozen feces without bedding or water. A parvovirosis outbreak forced a quarantine. Criminal proceedings under Article 245 of the Russian Criminal Code — animal cruelty — are being considered. Only 23% of the facility's infrastructure met legal requirements.

The shelter crisis is not a local problem with local causes. It is a global phenomenon driven by the same forces everywhere: rising costs, housing instability, and the slow unraveling of the pandemic-era adoption boom.

And in Bolivia, the city of El Alto is confronting its own staggering reality: approximately 270,000 dogs, of which some 100,000 live on the streets. The municipality is now socializing a draft law on responsible pet ownership that would mandate registration with the Municipal Zoonosis Center and establish penalties under Bolivia's existing Law 700 — but the gap between the legislation and the scale of the problem remains enormous.

A Legislative Wave From Amsterdam to Sacramento

If 2025 was the year governments started noticing the pet welfare crisis, 2026 is the year they started legislating. And the breadth of new laws is remarkable.

The Netherlands made history on January 1 by becoming the first country in the world to ban the keeping of fold-eared and hairless cats — including Scottish Folds and Sphynx cats. The reasoning is grounded in veterinary science: fold-eared cats suffer from osteochondrodysplasia, a genetic cartilage defect that causes chronic pain, stiff joints, and potential paralysis. Hairless cats cannot properly regulate body temperature and lack whiskers, impairing their sensory experience. Existing owners may keep their cats until natural death (provided they were microchipped before the ban), but breeding, selling, showing, or acquiring new ones carries a €1,500 fine.

The Dutch move is part of a broader European agenda. A new European Dogs and Cats Law is being finalized, with stricter rules for breeding and trading, and mandatory microchipping for every dog and cat in the EU — linked to both national and European databases. The Netherlands had already planned national cat microchipping for 2026, now pushed to 2027 as it aligns with the EU framework.

In Russia, parallel reforms are even more sweeping. Starting September 1, 2026, all cats and dogs must be registered through the federal FGIS "VetIS" system, marked with microchips, tags, or collars at the owner's expense. Simultaneously, the State Duma has advanced a breeder registry bill — backed by 34 deputies — that would create a unified national organization to maintain a registry of breeders, establish breeding standards, and effectively shut down the grey market of undocumented puppy and kitten sales. And a separate bill proposes increasing animal cruelty fines from 80,000 to 1,000,000 rubles — a twelvefold increase — with up to five years of forced labor for the worst offenses, including the deliberate poisoning of dogs in public spaces.

In the United States, a cascade of new state laws took effect in early 2026. California banned cat declawing (joining New York, Maryland, and others), prohibited puppy mill broker sales, and enacted "Jerry's Law" making owners liable for deliberately failing to provide medical care. Florida launched a public animal cruelty offender database and made pet abandonment during emergencies a felony. Ohio's "Avery's Law" authorizes courts to order euthanasia when a dog kills without provocation. And in a move that directly connects pet welfare to housing policy, California now requires landlords in state-financed affordable housing to allow at least one pet, with "pet rent" capped and deposits made refundable.

Even Germany's circus animals got new protections: as of March 1, 2026, new federal standards dictate minimum enclosure sizes, lighting, enrichment structures, and ventilation for animals in circuses, zoos, and dolphinariums — while a bill to ban circus animals entirely continues its slow progress through the Bundestag.

The Tech Revolution in Pet Care

While governments regulate, the private sector innovates — nowhere more aggressively than in China, where the pet market crossed the 310 billion yuan ($43.4 billion) mark in 2025.

PILTON, a state-backed Chinese startup, has shipped over 10,000 smart pet capsule units to 14 countries. These climate-controlled pods feature precision temperature regulation (±1°C), UV-sterilized air circulation, and AI cameras with smartphone monitoring. Most remarkably, China's largest hotpot restaurant chain, Haidilao, has deployed them in over 70 locations — letting diners leave their pets in monitored, air-conditioned capsules while they eat. It is a uniquely Chinese solution to a universal problem: what do you do with your dog when you go to dinner?

At CES 2026 in Las Vegas, Chinese smart pet company PETKIT demonstrated AI-powered litter boxes that detect abnormalities in urination patterns — flagging, for example, when a cat's trips jump from six to twelve per day, suggesting a possible urinary tract infection. Smart pet products in China are growing at a 38.7% compound annual growth rate.

In Chile, technology met disaster response when volunteers created animalesperdidos.cl, an AI-powered platform to reunite pets separated from their families during the devastating summer wildfires that burned over 35,000 hectares in the Biobío and Ñuble regions. The platform registered over 500 lost pets and achieved 85 reunifications — and remains active months later, a testament to both the scale of the displacement and the power of citizen-driven tech solutions.

Forgotten Animals, Unforgotten Bonds

Some of this week's most powerful stories are not about markets or legislation. They are about the bond between humans and animals tested by catastrophe.

In Japan, just days after the 15th anniversary of the Fukushima disaster, international media profiled Toru Akama, a 63-year-old former nuclear plant worker who has spent a decade and a half caring for pets abandoned when 154,000 residents evacuated the exclusion zone. He began by taking in 40 dogs, then 50. Over the years, he has placed more than 1,000 animals with adoptive families, spending nearly all of his disaster compensation money on their care. He currently lives with 47 cats and 7 dogs — and his biggest worry is finding a successor. "I too am starting to get older," he told AFP.

In Cochabamba, Bolivia, authorities raided the La Pampa market and rescued 34 animals — 32 dogs and 2 cats — crammed into cardboard boxes and poultry crates in an unventilated warehouse, none vaccinated against rabies, all destined for illegal sale. It was at least the second such raid at the same location in six months.

And in Germany, a puppy imported from Russia died of rabies at a shelter in Rhineland-Palatinate — the first case in a country that has been rabies-free since 2008. The puppy's vaccination certificate was forged; post-mortem examination revealed it was significantly younger than claimed, meaning any vaccination would have been ineffective. The puppy had bitten its owner before dying. Criminal police are investigating, and veterinary associations are demanding an urgent crackdown on illegal cross-border puppy trafficking.

What It All Means

Step back from the individual headlines and a coherent picture emerges. The world is in the midst of an unprecedented recalibration of the relationship between humans and companion animals. Pet populations are enormous — 80 million in Russia, 34 million in Germany, 16 million in Chile alone. Markets are booming, with global pet care valued at $273 billion and climbing toward $428 billion by 2032. Technology is transforming everything from diagnosis to disaster response.

But the systems we built to support all of this — veterinary pricing, shelter capacity, breeding oversight, legal protections — were designed for a world where pets were property, not family. The reforms sweeping from Westminster to Moscow to Sacramento are, in essence, the infrastructure of affection catching up to the reality of attachment.

For the pet owner reading this over morning coffee, the takeaway is both reassuring and sobering. Governments are finally paying attention. Prices may become more transparent, breeders more accountable, shelters more supported. But the fundamental tension remains: loving an animal is free; caring for one properly is not — and the gap between those two truths is where every story in this roundup lives.

Your Pet's Next Doctor Visit Might Be an App — How 8 Countries Are Proving It

krlz — Wed, 25 Mar 2026 01:05:26 +0000

Your dog just had a weird episode — shaking, refusing food, pacing in circles. It's 11 PM, the vet clinic is closed, and Google is serving you a terrifying mix of "probably nothing" and "could be fatal." Five years ago, that was the end of the road until morning. In 2026, millions of pet owners around the world are opening an app instead — and getting answers in minutes.

The global pet tech market has exploded to $19.1 billion in 2026, on track to reach $52.9 billion by 2035. But behind those numbers lies something more interesting than market growth: a fundamental shift in how eight different countries — from Tokyo to Santiago, from Amsterdam to Moscow — are rethinking what it means to care for an animal in the digital age. And pet apps are at the center of it all.

The AI Revolution in Your Pet's Food Bowl

At CES 2026 in Las Vegas, Chinese-founded PETKIT made the biggest splash in the pet tech world by unveiling a complete AI-powered ecosystem that turns ordinary pet care into a data-driven health platform. Their three flagship devices — the EVERSWEET ULTRA water fountain, the YUMSHARE DAILY FEAST robotic wet food feeder, and the PUROBOT CRYSTAL DUO automatic litter box — don't just automate tasks. They talk to each other.

Each device is equipped with AI cameras and pet facial recognition that identifies individual animals in multi-pet households. The water fountain tracks how much each cat drinks. The litter box monitors stool consistency and pH levels. The feeder logs eating patterns. All of this data flows into the PETKIT app, building what the company calls a "unified health dashboard" — a continuous health picture that can flag changes weeks before a human would notice them.

"The industry is moving away from simple 'connected' gadgets toward a new era of predictive wellness and pet-centric empathy — innovations that prioritize biological needs over mere mechanical convenience."

This isn't just a Western phenomenon. At the same CES event, Chinese startup 灵尾纪元 (Lingwei Epoch), which recently closed a multi-million yuan Series A round, debuted the MOVA SureTrack Pro smart collar. The collar packs UWB, GPS, and Wi-Fi triangulation into a module small enough for a cat, compressing lost-pet search ranges from hundreds of meters down to just two meters. It even features two-way voice calling — owners can literally talk to their dogs while at work. The company's next-generation version, currently in development, promises to be a real-time pet emotion and behavior translator, interpreting activity patterns, mood states, and anomalies 24/7.

Telemedicine: The Vet in Your Pocket

Perhaps no single feature has transformed pet ownership more than veterinary telemedicine. The global veterinary telehealth market is projected to surge from $146 million in 2025 to $747 million by 2035, growing at a staggering 18.8% annually. And every country is finding its own way in.

In Russia, Petstory — a Mars-backed digital platform — has become the go-to app for pet owners across 68+ cities. The app offers on-demand consultations via chat, audio, or video call with veterinarians who have a minimum of four years of clinical experience, available daily from 6 AM to 2 AM Moscow time. Beyond teleconsultations, Petstory maps nearby veterinary clinics, grooming salons, pet-friendly cafés, and even hotels that welcome dogs and cats. For a country spanning eleven time zones, the advantage is obvious: a pet owner in Vladivostok gets the same quality of veterinary advice as one in Moscow.

Meanwhile in France, where 75 million domestic animals make it the most pet-dense nation in Europe, the digital shift has been dramatic. A recent survey found that 64% of French pet owners now regularly use apps or platforms to care for their companions. The French startup ecosystem has responded: Maskots, an AI-first pet app available in French, Spanish, and English, has grown over 70% month-over-month to 7,000+ active users. Its AI assistant is trained exclusively on veterinary and pet care data, offering breed-specific, life-stage-appropriate recommendations instead of generic advice. Meanwhile, connected devices like Caremitou — a smart litter system — track litter box frequency to detect early signs of urinary issues, one of the most common and dangerous conditions in cats.

In Chile, the insurtech startup Pawer has carved out a unique niche by combining pet health insurance with unlimited televeterinary access in a single app. Founded by Guillermo Díaz in 2021, Pawer maintains a 95% customer satisfaction rate and covers everything from emergency accidents to preventive vaccinations. The app lets owners maintain complete digital health records and get reimbursed at any veterinary clinic in the country — a model that's now expanding to Peru, signaling Latin America's growing appetite for pet tech.

Japan's Quiet Pet Tech Boom

Japan, home to one of the world's most sophisticated pet cultures, has approached pet apps with characteristic precision. The country's digital pet care market is booming, with Fuji Keizai Group reporting accelerating growth in pet tech driven by the twin forces of "pet familization" (ペットの家族化) — the cultural trend of treating pets as full family members — and IoT maturity.

Japanese consumers can choose from over 37 dedicated pet apps as of 2026, ranging from health management platforms to social networks for pet owners. The Nikkei business newspaper now maintains a dedicated "Pet Tech" topic page, tracking the sector's transformation from niche hobby to serious industry. Smart devices like Toletta, a cat toilet developed by a Sharp subsidiary, analyze weight, frequency, and urine volume to detect kidney disease — the leading cause of death in older cats — sometimes months before symptoms appear.

The Japanese market is also pioneering what could be called "emotional tech" for pets: apps that track not just physical health metrics but behavioral patterns that indicate loneliness, anxiety, or boredom — crucial in a country where many pet owners work long hours.

The Netherlands: Where Welfare Policy Meets Digital Innovation

The Netherlands finds itself at a fascinating intersection of strict animal welfare legislation and digital pet care. As of January 1, 2026, it became illegal to buy, sell, breed, or keep hairless cats and cats with folded ears — including Scottish Folds and Sphynx breeds — due to the suffering caused by these genetic traits. Violations carry fines of up to €1,500, and these breeds can no longer appear in shows, competitions, or even social media advertisements.

This regulatory environment has fueled demand for apps that help owners navigate compliance. The HuisdierenApp, the Netherlands' leading pet management platform, offers a comprehensive digital hub where owners can store microchip numbers, insurance details, and vaccination records — essential now that existing cats with banned traits must be microchipped to be legally retained. The app also provides video-call access to veterinarians and maps every nearby pet professional, from behaviorists to groomers.

Dutch apps like DoggyDating — which facilitates group dog walks — and Barkio — a remote pet monitoring tool — reflect the Netherlands' broader philosophy: technology should enhance the human-animal bond, not replace it.

Latin America's Underdog Story

While the US and Asia dominate pet tech headlines, Latin America is writing its own compelling chapter. The regional pet tech sector is growing at 7.1% annually through 2032, with Chile leading the charge.

Beyond Pawer, Chilean apps like PetHero connect owners with vetted dog walkers, sitters, and at-home veterinarians. Petgram merges social media features — photo sharing, community engagement — with practical tools like vaccination reminders and vet appointment scheduling. And Oliver Pets, a Mexico-based platform using AI to personalize the entire ownership experience, is preparing its expansion into Chile.

Bolivia remains a frontier market, where veterinary services are still largely traditional and appointment booking happens through WhatsApp. But that gap represents opportunity: as smartphone penetration grows across the Altiplano, the infrastructure for leapfrogging directly to app-based pet care is already in place.

The Edge-AI Shift: Why Your Pet's Data Is Staying Home

One of the most significant — and underreported — trends of 2026 is the move from cloud computing to Edge AI in pet devices. Instead of streaming video of your cat to a server farm in Virginia, the newest generation of smart pet cameras and feeders process visual data locally, on the device itself.

The advantages are threefold. Privacy: footage of your home never leaves your home. Reliability: if your Wi-Fi goes down, the smart feeder still recognizes your pet and dispenses the right portion. And speed: on-device processing means alerts about unusual behavior arrive in real-time, not after a round trip to the cloud.

Users report 25% less food waste with app-tracked feeders and 40% reduced cleaning time via robotic litter systems — tangible, everyday improvements that add up across the 300+ million pets in the countries covered by this story.

What This Means for Pet Owners

The pet app revolution isn't just about convenience — though that matters too. It's about closing the information gap between pet owners and the medical professionals who care for their animals. A cat that drinks 15% less water over two weeks might seem fine to the naked eye. An AI dashboard that has been tracking daily intake across months will flag it as a potential early sign of kidney disease.

For pet owners in 2026, the message from Tokyo to Amsterdam, from Santiago to Moscow, is the same: the best app for your pet isn't the one with the most features — it's the one you'll actually use every day. Whether that's a simple vaccination reminder in Bolivia or a full AI health ecosystem in Shanghai, the technology is no longer the bottleneck. The future of pet care is already on your phone.

The Netherlands Just Banned Scottish Folds and Sphynx Cats — And the Rest of the World Is Watching

krlz — Wed, 25 Mar 2026 00:57:12 +0000

Since January 1, it has been illegal to buy, sell, or breed a Scottish Fold or a Sphynx cat anywhere in the Netherlands. Owners who acquired one of these beloved breeds after the cutoff face fines of up to €1,500. Existing cats, if born and microchipped before the deadline, may live out their natural lives — but they cannot be exhibited at shows, entered into competitions, or bred. For the roughly 10,000 Dutch owners of folded-ear and hairless cats, the message from The Hague is unmistakable: the era of breeding animals for aesthetic traits that cause suffering is over.

The Dutch ban is not happening in isolation. It is the sharpest edge of a global legislative movement that, in just the first quarter of 2026, has rewritten the rules of pet ownership from Sacramento to Moscow, from Santiago to Tokyo. Taken together, these changes signal something profound: governments around the world are no longer content to merely punish animal cruelty after the fact. They are stepping in to prevent it — even when that means telling pet owners which animals they can and cannot have.

When Cuteness Causes Pain

The science behind the Dutch decision is not new, but the political will to act on it is. Scottish Fold cats carry a genetic mutation called osteochondrodysplasia that affects cartilage throughout their bodies. It is this mutation that gives them their signature folded ears — the very feature that made them Instagram darlings and celebrity favorites. But the same defective cartilage causes progressive, debilitating arthritis that can leave cats in chronic pain, sometimes requiring euthanasia at a tragically young age.

Sphynx cats, meanwhile, lack the fur that regulates body temperature and the whiskers that serve as critical sensory organs. They are more vulnerable to skin infections, ear problems, and respiratory issues. The Dutch government concluded that the welfare costs of these breeds simply outweigh the aesthetic pleasure they provide to humans.

"The suffering of the Scottish Fold is undeniable, and this kind of action is sorely needed on a much wider level. It brings public attention to the issue of pedigree breed health and starts a wider conversation, which can only be a good thing."

The ban has already sent ripples across Europe. In Belgium, the animal welfare organization Veeweyde has publicly asked whether similar legislation should follow. Germany's Deutscher Tierschutzbund has long campaigned against what it calls Qualzucht — literally "torture breeding" — and advocates are now pointing to the Netherlands as proof that bans are politically viable. The European Federation of Veterinarians (FVE) has endorsed the Dutch approach, calling it a model for the continent.

The question is no longer whether other countries will follow, but when.

California Joins the Claw-Free Movement

Across the Atlantic, California rang in 2026 with its own landmark animal welfare legislation. Assembly Bill 867, signed by Governor Gavin Newsom, makes it illegal to declaw a cat in the state unless the procedure is medically necessary for the animal's health. Veterinarians who continue to perform cosmetic or convenience-based declawing risk having their licenses suspended or revoked by the California Veterinary Medical Board.

Declawing — known medically as onychectomy — involves amputating the last bone of each toe, comparable to cutting off a human finger at the final knuckle. Research has consistently linked the procedure to long-term pain, behavioral changes including increased aggression and biting, and litter box avoidance. California now joins New York, Maryland, Massachusetts, and Washington, D.C. in banning the practice, but its sheer size — home to nearly 40 million people and an estimated 7 million pet cats — makes this the most consequential state-level ban yet.

California didn't stop there. AB 519 banned pet brokers — middlemen who profit from selling animals bred by others — effectively targeting the puppy mill pipeline. SB 312 requires dog importers to submit health certificates electronically within 10 days of shipment, making them publicly accessible. And SB 602 allows designated shelters to conduct veterinary appointments without a supervising vet on-site, expanding care access in underserved communities.

The combined effect is a state that is simultaneously restricting harmful practices and expanding access to humane care — a dual strategy that animal welfare advocates have long championed.

Russia's Ambitious Plan to Register 75 Million Pets

Perhaps the most sweeping piece of pet legislation anywhere in the world is taking shape in Russia. Starting September 1, 2026, every cat and dog owner in the country will be required to mark their pet — via collar, tag, or microchip, at the owner's choice — and register the animal in the federal veterinary information system known as FGIS VetIS. Registration itself will be free, though the cost of tagging falls on owners.

The scale is staggering. Russia has approximately 75 million domestic pets but, until now, no centralized tracking system. Twenty-six of Russia's regions have already begun implementing registration, with the Moscow region leading since 2023. The Presidential Council for Civil Society and Human Rights has proposed integrating the registry with Gosuslugi, Russia's digital government services portal — potentially creating one of the world's largest pet databases.

But registration is just one pillar of a broader overhaul. In March, Senator Airat Gibatdinov introduced a bill to increase fines for animal cruelty tenfold, from the current cap of 80,000 rubles to 1,000,000 rubles (approximately $10,000). The State Duma is also considering a law that would cap the number of pets per apartment based on living space — 18 square meters per adult dog or cat — targeting hoarding situations that harm both animals and neighbors. Meanwhile, proposed amendments would give courts the power to confiscate pets from convicted animal abusers, closing a legal gap that has long frustrated prosecutors.

And as of March 1, new veterinary drug regulations came into force: antiparasitic medications — including common flea and tick treatments — now require a veterinary prescription, ending over-the-counter sales. The change mirrors pharmaceutical trends in Western Europe and is designed to combat counterfeit drugs, though it adds a new layer of logistical burden for Russia's pet owners.

"In five years, 65,731 integrated veterinary services were provided to pets in the municipality of La Paz alone — a figure that reveals both the scale of Latin America's stray animal challenge and the growing political will to address it."

Latin America Writes Its Own Playbook

The legislative momentum extends well into the Southern Hemisphere. In Bolivia, the city of El Alto — home to an estimated 270,000 dogs, roughly 100,000 of them strays — is socializing a new draft law on responsible pet ownership focused on public health and population control. In March, animal control officers in Cochabamba conducted a dramatic rescue operation, pulling 34 animals from cramped, unventilated storage containers near the La Pampa market, where dogs and cats had been crammed into cardboard boxes and chicken transport crates.

Across the border in Chile, where 86% of households now own a pet and the national pet population exceeds 12 million, the conversation has shifted from basic welfare to urban planning. A March article in El Periodista asked a question that would have seemed absurd a decade ago: are Chilean cities built for humans or for pets? The piece explored a European-style debate now taking root in Santiago — whether urban infrastructure, from parks to public transit, should be redesigned to accommodate the reality that pets are now central to daily life.

On the ground, the Chilean government is investing in sterilization as its primary tool for population management. In the southern province of Chiloé, the national government allocated 86.87 million pesos for a comprehensive 2026 spay-neuter program targeting rural areas, where uncontrolled breeding threatens both domestic animals and native wildlife.

A Rabies Scare Exposes Europe's Import Loopholes

In late January, a dog imported from Russia died of rabies in Germany's Rhineland-Palatinate region, near Mannheim. Laboratory confirmation came on February 10, triggering an investigation that exposed troubling gaps in Europe's animal import system.

The dog had arrived with all required documentation — microchip, vaccination certificate, and antibody test results. But post-mortem analysis revealed the animal was younger than declared at the time of import, meaning the paperwork was either fraudulent or inauthentic. The dog was part of a consignment of 24 animals — 22 dogs and 2 cats — imported by an animal charity and distributed across Germany and other European countries.

Germany's Environment Minister Katrin Eder used the case to underscore the critical importance of regular vaccination, warning that only consistent immunization can reduce the risk of lethal rabies transmission through bites. Germany has been officially rabies-free since 2008, and this imported case — the first in years — served as a jarring reminder that the disease is only a border crossing away.

The Friedrich-Loeffler-Institut, Germany's federal research institute for animal health, is now calling for stricter verification of import documents, particularly for rescue animals brought in by charitable organizations — a well-intentioned channel that, in this case, became a vector for a deadly virus.

The AI Revolution Reaches the Veterinary Clinic

While legislatures write new rules, technology is quietly reshaping veterinary practice from the inside. A major survey released in March by Instinct Science found that 48% of general veterinary practices in the United States now use artificial intelligence in some capacity. The most common applications are medical record and SOAP note creation (used by 63% of AI-adopting practices) and diagnostic assistance (38%).

The implications extend beyond efficiency. AI is also accelerating research. In Russia, the veterinary information portal ZooInform reported in March that neural networks have achieved expert-level accuracy in selecting clinical cases for retrospective veterinary studies — a tedious, labor-intensive process that has historically bottlenecked research. If scalable, this could dramatically speed up the pace of veterinary science, enabling faster identification of disease patterns and treatment protocols.

Meanwhile, Mars Veterinary Health published its 2025 Science Impact Report, highlighting over 500 peer-reviewed publications from its global network of clinician scientists. Among the key breakthroughs: the identification of the SLAMF1 genetic variant linked to canine atopic dermatitis, which is now being developed into a DNA test in collaboration with academic partners. For the millions of dogs who suffer from chronic skin allergies, this could eventually mean earlier diagnosis and more targeted treatment.

China's Pet Economy Breaks Records — Again

China's urban pet market for dogs and cats reached 312.6 billion yuan (approximately $43 billion) in 2025, a 4.1% increase over the prior year, with projections pointing toward 405 billion yuan by 2028. The cat segment is growing faster than dogs — up 5.2% compared to 3.2% — reflecting a global trend toward feline ownership, particularly among young urban professionals living in smaller spaces.

The Chinese market is also the global epicenter of smart pet technology. The city of Nanjing announced plans to create dedicated "pet consumption streets" across three districts, while this week's Nanjing International Pet Expo (March 27-29) is expected to showcase the latest generation of intelligent pet habitats — climate-controlled capsules that monitor temperature, air quality, and even pet behavior through embedded sensors.

In Japan, the professionalization of pet care continues with the fourth annual national examination for Certified Companion Animal Nurses, a state license established in 2022. The February 2026 exam maintained a pass rate near 88%, reflecting a maturing pipeline of formally trained veterinary support staff — a development that countries struggling with veterinary workforce shortages are watching closely.

What It All Means for Pet Owners

The thread connecting a banned cat breed in Amsterdam to a rescued dog in Cochabamba to an AI-assisted diagnosis in Kansas City is this: the world is recalibrating its relationship with companion animals. Governments are no longer treating pets as mere property. They are recognizing them as beings whose welfare demands proactive protection — through breeding restrictions, mandatory identification, prescription drug oversight, and professional care standards.

For everyday pet owners, the practical takeaways are straightforward. Check whether your pet's breed or your care practices may be affected by new or pending legislation in your region. If you're considering a new pet, research the breed's health profile, not just its appearance. Keep vaccinations current — the German rabies case is a stark reminder that prevention is non-negotiable. And if your vet mentions AI-assisted diagnostics, know that the technology is no longer experimental — it is becoming standard practice.

The age of passive pet ownership is ending. What's replacing it is a world that expects more from us — and promises more for them.

Pet News Roundup: March 23, 2026

krlz — Mon, 23 Mar 2026 20:57:29 +0000

This week's global pet news brings a mix of landmark legislation, community-driven rescue efforts, and the growing recognition that our animals deserve better care no matter where they live. From policy shifts in Moscow to mobile vet clinics in the Chilean countryside, the world is paying attention to our four-legged family members.

United States

PetSmart Charities announced $3 million in grants to local animal welfare organizations ahead of National Adoption Week (March 23–29, 2026), bringing adoptable pets into PetSmart stores across the country to boost shelter-to-home placements.
California's cat declawing ban, which took effect January 1, continues to draw national attention as one of the strongest animal protection measures in the country. The state also enacted AB 516, allowing registered veterinary technicians to perform a broader range of medical tasks, helping clinics serve more animals amid an ongoing veterinary staffing shortage.
A major milestone from Best Friends Animal Society: the number of cats killed in U.S. shelters has dropped nearly 75% over the past decade, thanks to expanded adoption programs, community cat initiatives, and foster care networks.

"The number of cats killed in U.S. shelters has dropped by nearly 75 percent in ten years — proof that community-driven programs and compassionate policy really do save lives."

Russia

New pet registration fines took effect March 1 in Russia's Zabaykalsky Krai, where owners now face penalties of 3,000–5,000 rubles for failing to register their dogs — part of a broader national push toward mandatory pet identification.
Russia introduced item-level tracking of all veterinary medications through the "Chestniy Znak" (Honest Mark) labeling system starting March 1, 2026, aiming to combat counterfeit pet medicines and improve supply chain transparency.
In veterinary research, Russian scientists are pioneering the use of AI-powered neural networks to select clinical cases for retrospective studies with expert-level accuracy, addressing long-standing inconsistencies in veterinary medical record-keeping.

Bolivia

Bolivia continues to grapple with its stray dog crisis: between the cities of El Alto and La Paz alone, an estimated 300,000 dogs roam the streets, with roughly half being "semi-strays" — pets that have homes but wander freely during the day.
A 2017 municipal law in La Paz (Ley Municipal 239) was designed to promote companion animal welfare through spay/neuter programs and a pet registry, but remains largely unimplemented and unenforceable, leaving animal advocates frustrated.
International organizations like Humane World for Animals continue to support Bolivia through cross-border programs, including veterinary training and partnerships to improve animal care infrastructure in underserved areas.

Chile

Humane World for Animals Chile launched a major 2026 welfare program sending 18 veterinary specialists to remote communities in the Coquimbo region — places where the nearest vet clinic is a 40-minute drive away. The team provided free spay/neuter services, vaccinations, and health checkups to dogs and cats in Barraza, Salala, and Oruro.
Chile's pet market has grown into a $1.6 billion industry, with pet food accounting for roughly 70% of sales. The country's Responsible Pet Ownership Law has resulted in over 1.3 million registered pets and 449,000 sterilizations since its enactment.
Professor Constanza Cabello from the University of Chile's veterinary faculty brought six students to the Coquimbo field program, combining hands-on training with real community impact — a model that's gaining attention across Latin America.

Germany

Germany remains one of Europe's largest pet markets, with 33.9 million pets across the country, led by 15.7 million cats and 10.5 million dogs. However, total pet ownership dipped slightly from 45% to 44% of households in 2024.
The German government opened nominations for the 2026 Animal Welfare Research Prize (Tierschutzforschungspreis), with submissions closing March 31 — highlighting the country's ongoing investment in humane scientific alternatives.
A Purina symposium on the gut-brain axis held in Germany on March 13 brought together veterinary experts to discuss how digestive health influences pet behavior and mental well-being, reflecting the growing trend of science-based pet nutrition.
Rising pet food and veterinary costs driven by inflation are pushing German owners toward online shopping, with pet-related e-commerce revenues reaching $1.4 billion and the trend toward premium, humanized pet products continuing to grow.

Whether it's a shelter cat finding a home during National Adoption Week in the US, a street dog finally getting vaccinated in rural Chile, or a new AI tool helping Russian veterinarians improve diagnostics — the common thread is clear: communities around the world are investing more in the health and welfare of their companion animals. As pet owners, staying informed about these developments helps us all be better advocates for the animals we love.

How a corruption-convicted company broke Chile's national ID system — and walked away with $480M

krlz — Sat, 21 Mar 2026 01:23:18 +0000

In December 2024, Chile's Registro Civil rolled out a brand new national identity system. It crashed on day one. The system behind every Chilean's cédula de identidad and passport went down nationwide — biometric capture failed, payments broke, documents wouldn't activate, and data errors appeared in identity records.

The company responsible? IDEMIA — a French firm built from the merger of two companies with documented corruption histories, owned by American private equity, and already planning its exit before the ink on Chile's contract was dry.

The Backstory: A Company Born from Corruption

IDEMIA doesn't exist in a vacuum. It was created in 2017 from the merger of:

Oberthur Technologies — debarred by the World Bank for 2.5 years for bribing officials in Bangladesh to win a national ID contract
Safran Identity & Security (Morpho) — fined €500,000 by a French court for bribing Nigerian officials for a €170M ID card contract

The merger was orchestrated by Advent International, a Boston-based private equity firm that owned Oberthur since 2011. Safran sold its identity division to Advent, the two merged into OT-Morpho, and then rebranded as IDEMIA — a clean name for a dirty history.

The Rap Sheet

Year	Country	What Happened	Penalty
2012	Nigeria	Safran/Morpho bribed officials for ID card contract	€500K fine
2017	Bangladesh	Oberthur bribed officials for national ID project	World Bank debarment (2.5 years)
2022	Bangladesh	IDEMIA settled corruption charges (same case)	€8M fine, 4 executives flagged for prosecution
2022	Angola	IDEMIA France corrupted public officials (2007-2011)	€15.5M fine via CJIP (no trial)
2025	Nepal	Complaints of payments to brokers via foreign accounts	Under investigation
—	Kenya	National Assembly voted to ban IDEMIA for 10 years	Suspended from government contracts
—	Ecuador	Declared IDEMIA a "failed adjudicator" in similar tender	Contract revoked

Total fines paid: ~€24 million across multiple countries. Executives imprisoned: zero (publicly known).

The People: Who Did What

The Executives Who Walked Free

In Nigeria, former Safran/Morpho executives Jean-Pierre Delarue (sales manager in Nigeria) and François Perrachon were charged for the bribery scheme that won a €170M ID card contract. A French court acquitted both of them. The company paid the €500K fine. Nobody went to prison.

In Bangladesh, French prosecutors identified nearly €6.1 million in bribes intended for Tarique Ahmed Siddique, a Bangladeshi general. The money was funneled through Ziaur Rahman, who led Tiger IT, the local subcontractor. IDEMIA settled for €8M. Four former Oberthur executives were flagged for separate prosecution — their names have never been made public, and there is no record of convictions.

In Angola, IDEMIA paid €15.5M via a CJIP (French deferred prosecution agreement) for corrupting public officials between 2007-2011. No individuals were named or prosecuted. The deal was signed, the fine was paid, and the file was closed.

Pierre Barrial, IDEMIA's CEO from 2020 to 2025 — the period covering Chile's contract execution and system crash — has faced no public investigation. He was quietly moved to the Supervisory Board when IDEMIA restructured its governance in mid-2025.

The Chilean Officials: A Revolving Door

The tender process consumed three Registro Civil directors and exposed deep conflicts of interest:

Jorge Álvarez Vásquez — Director who fined IDEMIA $164 million CLP in July 2020 for contract breaches and obstructing an investigation into forged identity documents by Registro Civil employees. He was removed from his position on October 16, 2020 — barely one day after extending a deadline in the tender process.

Sergio Mierzejewski — Replaced Álvarez. A private-sector figure linked to the UDI political party, he awarded the bid to the Chinese consortium Aisino. A criminal complaint (querella) was filed against him for "privileging" Aisino, alleging incompatible negotiation, trafficking in influence, violation of secrecy, and use of reserved information. But the complainant, Bruno Opazo Ruiz, was accused by Aisino of having links to IDEMIA — suggesting the legal action itself may have been orchestrated to eliminate the Chinese competitor. Mierzejewski eventually resigned.

Jaime Gajardo Falcón — Chile's Undersecretary of Justice, who had a key role in appointing the next Registro Civil director. He worked for four years at ZC Abogados, the law firm of jurist Francisco Zúñiga (former Constitutional Court minister) — and ZC Abogados is IDEMIA's main legal representative in Chile. Gajardo claims: "I never had any participation in the consultancies or representations that the firm provided to IDEMIA" and "I have never received any payment from IDEMIA nor have I had any contact with its representatives."

The France Trip

After the contract was re-awarded to IDEMIA following Aisino's exclusion, Registro Civil officials traveled to France — paid for entirely by IDEMIA:

Jorge Núñez — Acting National Director
Felipe Barnachea — Chief of Staff
Ricardo Salgado — Head of Technology and Identification Services

The trip was presented as a "technical visit" to review ID card and passport samples. Former officials pointed out that Núñez and Barnachea are not technical experts, and the samples could have been mailed to Santiago. The trip occurred just before IDEMIA renewed a contract worth $55 billion CLP.

The Workers Who Paid the Price

After the December 2024 crash, two officials were fired as scapegoats:

Daniel Teplizky — Subdirector of Studies and Development (replaced by Paulo Ulloa Cifuentes)
Jaime Lira — NSI Project Coordinator (replaced by Mauricio Godoy Cisternas)

Additionally, regional director Jorge Zúñiga and head of Institutional Management Julio Parra departed after a crisis at Víctor Jara Stadium involving 3,000+ foreigners on the system's debut day.

Meanwhile, ANERCICH (the workers' union) issued a public statement: "The company IDEMIA was not able to fully implement the system, and workers have had to assume responsibilities that correspond to the company, negatively affecting both their work and public perception." Undersecretary of Justice Ernesto Muñoz acknowledged "operational difficulties" but claimed 95% operability within weeks — a claim contradicted by months of ongoing failures.

How Chile Handed Them $480 Million

The tender process for Chile's identity system was worth US$480 million over 10 years. Investigative outlet CIPER Chile documented what they called "zonas oscuras" (dark zones) in the process:

A competing Chinese consortium (Aisino) initially won, then was annulled — the contract was re-awarded to IDEMIA
Competitors (Sonda-Thales) filed lawsuits alleging IDEMIA manipulated pricing
IDEMIA held exclusive knowledge of "business rules" from the previous contract — a 7-year information advantage no competitor could match
Chile's Justice Undersecretary previously worked at ZC Abogados, IDEMIA's main legal firm in Chile
Registro Civil officials flew to France on IDEMIA's dime before renewing the contract
A querella was filed against the director who favored the Chinese bid — allegedly linked to IDEMIA itself

The Contraloría (Chile's comptroller) conducted 22 audits and still approved the contract. The Contralor and the Fiscal Nacional Económico were summoned to Congress to explain the process — described by El Mostrador as a "teleserie" (soap opera).

The Crash: December 2024

On December 16, 2024, President Boric personally launched the new system. It immediately failed:

Nationwide system crashes — offices couldn't process documents
Biometric capture failures — fingerprint scanners malfunctioning
Payment system down — cards declined, transactions lost
Data errors in issued identity documents
Visa Waiver broken — new passports couldn't process US ESTA applications
Identity card delivery times went from 30 days to 60+ days

The workers' union (ANERCICH) publicly blamed IDEMIA for delivering a system that wasn't ready. The Registro Civil fired the officials responsible for the rollout — but IDEMIA kept the contract.

The Technical Stack Behind the Mess

IDEMIA doesn't publish its full stack, but from job listings and technical documentation, we can piece together what Chile's identity infrastructure runs on — and why it keeps breaking.

The Backend: IDEMIA Identity Platform

At the core is the IDEMIA Identity Platform, a modular system handling three functions:

Identity Proofing — document verification, biometric checks, watchlist screening
Identity Management — creation, storage, de-duplication, lifecycle management
Identity Authentication — enrollment and verification across channels

The platform is described as "API-first" and modular. In practice, the known components include:

Component	Technology
Biometric engine	MBSS (Multi-Biometric Search Services) — fingerprint, palm, iris, face, tattoo recognition
Database	Oracle Database
Message queue	RabbitMQ
Operating system	Linux
Workflow integration	Accenture (custom integration layer)
Backend language	Java / JEE (confirmed via job postings requiring "Java, JEE platforms and related frameworks")
Database queries	SQL — preferably MS SQL (per job listings), though Oracle is the production DB
Cloud options	AWS or Azure Government (for other deployments — Chile's appears to be on-premise)
Hardware	MBSS is "100% software, no proprietary hardware" — runs on COTS (commercial off-the-shelf)

The biometric engine (MBSS) is the same one used by the FBI's NGI, DoD ABIS, and the State Department — trusted by 70%+ of US law enforcement. On paper, it's proven technology. The question is implementation.

The ID Card & Passport

The physical documents are where IDEMIA's hardware expertise shows:

Identity card: Visible chip (NFC-capable), 32 security features, Braille initials, multibiometric data (fingerprint + face + iris)
Passport: 70 security features, 50 pages, blue turquí cover
Digital ID: Mobile app with encrypted data signed by the backend, stored on-device, requiring user authorization to share

The decadactilar capture (10-finger scanning) uses IDEMIA's proprietary biometric devices. These were among the first things to fail — workers reported malfunctioning scanners and capture interface crashes from day one.

The "Core Principal" — Single Point of Failure

In November 2025, the entire system went down again nationwide. The cause? An update to what the Registro Civil called the "Core Principal" — the main processing engine of the system.

This suggests a monolithic architecture or at least a critical central component with no graceful failover. When the Core Principal breaks, everything breaks — every office, every consulate, every online service. No circuit breakers, no canary deployments, no rolling updates. Just a nationwide outage from a single update.

The App Ecosystem: A UX Disaster

The citizen-facing side is fragmented across multiple disconnected apps:

App	Purpose	Play Store Rating
Identificación Digital	Digital ID on phone	1.0 stars
CivilDigital App	Certificates and services	Poor ratings
Registro e Identidad	ID card reprinting	Poor ratings
CedulApp	ID card activation	Poor ratings

Users need to download multiple apps to complete a single workflow. Certificates fail to download. Connections drop mid-transaction. In 2020, a security researcher demonstrated they could obtain another person's Clave Única (national digital key) using only their RUT number and a photo from the internet — exposing critical flaws in the facial recognition authentication.

The US Visa Waiver Fiasco

Perhaps the most embarrassing technical failure: the new passport system couldn't properly encode certain characters, causing Chilean passports to fail validation on the US ESTA (Electronic System for Travel Authorization) platform. This threatened Chile's participation in the entire US Visa Waiver program — a diplomatic incident caused by a character encoding bug.

What This Tells Us

The stack itself isn't inherently bad — Oracle, Java, Linux, RabbitMQ are battle-tested. MBSS powers the FBI. The problem is:

Implementation quality — proven components assembled poorly, with no proper load testing or failover
Monolithic "Core Principal" — a single point of failure for an entire country's identity system
No gradual rollout — launched nationwide on day one instead of phased regional deployment
Fragmented client apps — multiple disconnected apps instead of a single unified platform
Character encoding failures — basic internationalization bugs in a system that handles names from dozens of nationalities
IDEMIA charged ~US$226 million for this implementation — paid in 106 installments based on documents issued

For a company with 3,000+ engineers and 50+ cryptography experts, the failures aren't about capability. They're about what happens when a corruption-prone private equity portfolio company prioritizes contract extraction over engineering excellence.

Still Broken in 2025-2026

February 2025: Users still reporting delays, portal crashes, document losses
November 2025: Another nationwide outage from a "Core Principal" update
December 2025: Claims of a data breach affecting 10 million Chileans — Registro Civil denied it
March 2026: Intermittent failures continue

The Private Equity Exit

Here's where it gets worse. While Chile struggles with a broken system, IDEMIA's owners are cashing out:

Advent International sold IDEMIA's Smart Identity division (the one making Chile's IDs) to IN Groupe, a French state-owned company — deal closed July 2025
Advent is now selling IDEMIA Public Security for €2-3 billion
Chile contracted IDEMIA but is now dependent on a completely different entity it never chose

The classic private equity playbook: acquire → win government contracts → extract value → sell → leave the mess behind.

Why Nobody Goes to Jail

Mechanism	How It Protects Individuals
Corporate fines	The company pays, not the decision-makers
DPAs / CJIPs	Deferred prosecution deals — no trial, no names
Mergers & rebranding	Oberthur → OT-Morpho → IDEMIA — the corrupt entity "doesn't exist"
PE sales	Advent buys, profits, sells — liability stays with the old entity
Jurisdictional gaps	Bribes in Bangladesh, company in France, owned by US PE — who prosecutes?
"Internal accountability"	Companies claim they fired people internally — no criminal referral

In the Bangladesh case, 4 former executives were flagged for separate prosecution. There is no public record of convictions.

The Bottom Line

A company sanctioned by the World Bank for corruption won a half-billion-dollar contract to build the identity system of an entire country. The system broke. The owners sold and walked away with billions. The executives faced no public consequences. And 19 million Chileans are left with an unreliable identity infrastructure run by an entity they never chose.

This isn't a bug. It's the system working as designed — just not for citizens.

Sources

Chile Registro Civil Crisis

People & Investigations

IDEMIA Tender Irregularities

IDEMIA Corruption History

IDEMIA Ownership & Corporate Structure

Technical Stack & Architecture

Individual Accountability & Legal Framework

I built TubeWords — a free tool to transcribe any YouTube video and get an AI summary in seconds

krlz — Thu, 05 Mar 2026 22:39:12 +0000

Every once in a while you watch a long YouTube video — a conference talk, a tutorial, a podcast — and you just want the gist. Or maybe you want to search through what was actually said. Or save a clean transcript to your notes.

I kept reaching for a tool like that and kept not finding exactly what I wanted. So I built it.

What is TubeWords?

TubeWords is a free, no-signup tool that does two things:

Extracts the full transcript of any YouTube video, powered by Whisper AI
Generates an AI summary of the content, so you get the key points without watching the whole thing

Paste a URL. Get your transcript. Done.

It supports timestamps or plain text output, so you can grab exactly the format you need — whether you're taking notes, building context for a prompt, or just skimming what was said at minute 12.

Why I built it

Honestly? Laziness in the best sense.

I watch a lot of technical content — conference talks, deep-dive tutorials, dev discussions — and I often want to reference what was said without rewatching. I also found myself copying transcripts into Claude or GPT to get a quick summary, which felt like unnecessary friction.

I wanted one place where I paste a URL and immediately get:

A clean, readable transcript
A concise summary
The ability to copy either one with a click

That's it. No accounts. No paywalls. No AI upsell features I'll never use.

The tech behind it

Whisper AI for the transcription — it handles accent variety and audio quality better than YouTube's auto-captions in most cases
Groq for the AI summary — fast inference, low latency, genuinely good summaries
Available in English and Spanish 🇨🇱

There's also a CLI 🖥️

If you live in the terminal (like I do when I'm in flow), there's an npm package:

npm i -g tubewords

Then just:

tubewords https://youtube.com/watch?v=VIDEO_ID -o notes.md

It fetches the transcript, generates the summary, and saves it to a markdown file. Super useful when you want to pipe YouTube content directly into your notes or your own tooling.

What's next

This is v1.0.1 — it works, it's stable, but there's more I want to add:

Chapter-aware summaries
Export to Notion / Obsidian
Bulk video support (playlists)
More languages

If you have ideas or run into issues, I'd love to hear from you.

Try it

👉 tubewords.com

Built with ♥ by krlz.dev

If you find it useful, share it with someone who watches way too many YouTube tutorials. That's probably most of us.

What's your go-to workflow for taking notes from video content? Drop it in the comments — always curious how other devs handle this.

We built a pet medication tracker because every other app failed us

krlz — Wed, 04 Mar 2026 02:13:01 +0000

Last year, my partner and I adopted a dog 🐶 (keep reading, pics at the end!)

And with him came a lot of love... and a lot of responsibility.
Vaccines, medications, checkups — all on time.

I first tried using Google Calendar, but it quickly turned into clutter: work meetings, birthdays, reminders — everything mixed together, and pet-related things got lost.

Then I tried many pet apps, but something was always off: some were overdesigned and hard to read, others missed notifications.

At some point we said: hey, we're developers after all.
So together with Marina, as a family project, we decided to build the app we actually needed.

That's how ZooMinder was born — a simple, clear app to track your pet's meds, vaccines, and checkups, without noise or clutter.

What it does

ZooMinder focuses on one thing and does it well: keeping your pet's health schedule organized.

Multiple pets — Add all your pets with photos, and birthdate
Treatments & medications — Track meds with dosage, frequency (daily, weekly, custom intervals), and time-specific doses
Vet appointments — Save clinic info, vet name, phone, and address so everything is in one place
Upcoming doses — A clear timeline of what's coming next, across all your pets
History — Full timeline of past doses and appointments per pet
Local notifications — 'Bark' sound alerts so you never miss a dose, even when the app is closed
Offline-first — Works without internet. Data syncs to the cloud when you're back online

The tech behind it

We built it with Flutter (Dart 3.10+), targeting Android first.

Here's the stack:

Layer	Tech
Framework	Flutter
Local DB	SQLite (sqflite)
Backend & Auth	Supabase
Notifications	flutter_local_notifications + custom sound
Navigation	go_router
i18n	Flutter's intl (English + Spanish)
In-app purchases	in_app_purchase
Sync	Custom offline-first sync service

Offline-first architecture

This was the most interesting technical challenge. The app stores everything locally in SQLite first. Every record has a synced flag. When the device comes back online, a SyncService pushes unsynced records to Supabase and pulls remote changes — handling conflicts with timestamps (updated_at).

// Every model carries its sync state
class Treatment {
  final bool synced;
  final DateTime updatedAt;
  // ...
}

This means the app is fully usable on a plane, in a rural area, or anywhere without signal. Your pet's schedule doesn't depend on your WiFi.

Notifications that actually work

We created dedicated Android notification channels for medications and appointments, with a custom notification sound. The service schedules up to 3 upcoming notifications per treatment, recalculating whenever treatments are added or modified.

static const _medicationChannel = AndroidNotificationChannel(
  'medication_reminders_v2',
  'Medication Reminders',
  importance: Importance.high,
  sound: _customSound,
  playSound: true,
);

What we learned

Building a "simple" app is harder than it looks. Some takeaways:

Offline-first is worth it but painful — Sync logic with conflict resolution took more time than all the UI combined. But the UX payoff is huge.
Notifications are a minefield — Between Android permission models, battery optimization killing background processes, and timezone handling... just scheduling a daily reminder at 8am is surprisingly complex.
Scope creep is real — We started with "just track medications" and ended up with appointment management, dose history timelines, vet clinic storage, premium subscriptions, and multi-language support. Saying "no" to features is a skill.
Family projects keep you honest — When your co-developer is also your partner and your first user, feedback is... immediate. And brutally honest. 😄

What's next 🚀

We're just getting started. Here's what we're planning:

Family plans — Share pets between household members so everyone can track medications and appointments from their own account. No more "did you give him the pill?" texts.
Assign responsibilities — Delegate tasks to family members. Mom handles the morning dose, dad takes the evening one. Everyone knows what they're responsible for.
Vet-oriented tools — We want to build features that integrate well with ZooMinder and give vets a better way to stay connected with pet owners — think shared treatment plans, visit summaries, and more.

We believe pet care is a team effort, and ZooMinder should reflect that.

Try it

ZooMinder is available on the Google Play Store. If you have pets and struggle to keep track of their medications and vet visits, give it a try.

We'd love your feedback — and if you're a Flutter dev, we're happy to chat about the offline-first patterns we used.

Want to help? 🙌

If you have suggestions, ideas, or found a bug — feel free to send me a DM here on dev.to! I'm always open to feedback.

Also, if you'd like to help as a beta tester, I'd really appreciate it. In exchange for your collaboration and honest feedback, I can give you full access to the premium version of the app. Just reach out and we'll get you set up.

Every bit of help makes ZooMinder better for all pet parents out there. 💪

As I promised some pics of the little guy who started it all. Here he is 🐾

#flutter #flutterdev #dart #mobiledev #android #sideproject #indiedev
#buildinpublic #petcare #pettech #supabase #offlinefirst #petparents #devlife
#familyproject

Built with Flutter, Supabase, and a lot of dog hair on the keyboard.

Have a great day! 👋🐾

DEV Community: krlz

The UN wrote its first AI report. The interesting parts are about your codebase.

The one idea that ties it all together

The 8 headline findings (compressed)

Does it hold up? I checked the claims that matter to us

Coding agents are getting longer horizons — fast

The "agents can be tricked" problem is demonstrable, not hypothetical

The AI feedback loop is already here

Key numbers at a glance

The state of the art, made concrete (mid-2026)

The frontier is a portfolio, not a winner

Everyone ships a tiered family

Benchmarks: which ones still discriminate

Orchestration is the real 2026 story

A practitioner's reality check

The upside is the same curve, not a different one

So what should a developer take from this?

Your turn

Living in Innopolis: notes from a Bolivian in Russia's IT city

A city built from scratch

Who actually runs the place

The university at the center

How I actually got in

How the programs actually work

My industrial project: disaster recovery with Acronis

The language thing nobody warns you about

The good and the bad, for students especially

People I knew were already leaving

Fighting brain drain and creeping authoritarianism in Russia’s techno-utopian village - Rest of World

Staying to work: my years at Portavita

Final thoughts

AI Made Code Free. So Why Are the Giants Still Winning? (And where solo devs actually beat them)

The one-sentence thesis

"AI is an amplifier, not an equalizer"

Everyone uses AI. Fewer and fewer trust it.

Agentic coding is a frontier, not the norm (yet)

The unifying insight: usage ≠ results

The counter-hype stat you should tattoo somewhere

Clean architecture is now a business decision

The map is shifting to the Global South

Open source is bigger than ever — and more fragile than ever

The solo playbook for 2026

TL;DR

Spec-Driven Development in 2026: What It Is, the Tooling, and How Teams Actually Use It

Why now? The "vibe coding" backlash

Raw specs vs. executable specs

[2602.00180] Spec-Driven Development:From Code to Contract in the Age of AI Coding Assistants

The maturity model: three levels

The canonical workflow

The tooling landscape (2026)

github / spec-kit

💫 Toolkit to help you get started with Spec-Driven Development

🌱 Spec Kit

Build high-quality software faster.

Table of Contents

🤔 What is Spec-Driven Development?

⚡ Get Started

1. Install Specify CLI

30-second tool-selection rubric

EARS: how to write specs an LLM can't misread

A worked example: "magic-link login"

Recommendations & best practices

The honest caveats

What the data says (read with a skeptic's eye)

The bottom line

The 6-Month Clock Ran Out — Are Companies Full AI Yet?

What the Data Actually Says

From Vibes to Discipline

The Return of XP

What the Industry Leaders Are Saying

Teams Are Shrinking, Roles Are Shifting

Measuring What Matters

The Fundamentals Debate Is Over

Where to Start

Resources

Britain Forces Vets to Publish Prices as a Global Reckoning Over Pet Care Costs Takes Shape

The Bill Comes Due

When Loving Your Pet Means You Can't Afford Your Pet

The Shelter Crisis No One Saw Coming

A Legislative Wave From Amsterdam to Sacramento