DEV Community: Daily Context

You’re not really that far behind.

Ryan Swift — Mon, 29 Jun 2026 16:06:48 +0000

My non-tech friends still don’t get it. Despite what you’d believe from Twitter, most people still haven’t seen the magic of AI. They don't use agents. They aren't tokenmaxxing. Most aren't really using AI at all, and if they are, it's a glorified search engine replacement.

I couldn’t be more different. I’m hopelessly addicted to AI news and model releases. I constantly message my Hermes agent. I am tokenmaxxing. And yet, somehow I am still the one feeling behind.

That gap is easy to forget when you're inside it. We're three years into the fastest tech transformation the world has ever seen, and it's still day zero for adoption. If you're at the AI Engineer World's Fair, you're among the earliest of early adopters though. By the most aggressive estimates, there are only 40 million software engineers in the world. That sounds like a lot until you realize more than a billion people are suddenly about to be able to write code. They just haven't been immersed in this for the last three years like we have.

So why do we feel so behind? If you're like me, you're overwhelmed by the sheer volume of new releases you're somehow supposed to keep up with. It's easy to get trapped in the hype cycle. What grounds me is remembering how early we actually are. If you're feeling that way this week, I hope this framing helps you too.

We’re paving the way for the wave of people coming next. Much of what we learn here won’t reach them for some time still. So beyond learning for ourselves, we should learn for others and share what we learn. If you see someone feeling lost, at AIE or out there afterward, remind them: they're much farther ahead than behind.

Two-day hackathon kicks off AI Engineer World’s Fair

Iain Thomson — Mon, 29 Jun 2026 16:06:30 +0000

While the World’s Fair officially kicks off today a bunch of keen developers were in early, taking part in a hackathon that is offering $35,000 in prizes and credits.

Teams worked to develop AI-powered apps, but with a twist. The organizers weren’t really looking for basic apps that can just perform a single function or task. Instead, the organizers explained, they were looking for code that will learn and develop with minimal user input.

For example a team calling itself SplatForge wrote an app from scratch that had a 93% success rate in identifying and manipulating objects virtually. It linked up with Google’s Gemini engine, which then began suggesting improvements to the application’s performance.

“It adds new edge cases that the AI can improve from,” said the team leader “This is something I feel is missing today.”

Team Rote claimed that they have built the fastest AI system in the world. “Today an agent computer use can operate a real browser or desktop app, but that live reasoning
expensive unit, you repeat the same workflow. Rote turns these successful runs into memory. So, once a run is executed by a computer use agent,” explained the team leader.

“Here we're using Gemini. Another agent actually reasons and records what happens, so it compiles a run into a reusable skill. It verifies it and stores it in our shared database using MongoDB, so that the next request from anyone around the world can execute the same command instantly.”

This not only increases the speed of the system, it reduces the cost, the team said. Users save on not running the same query again and again, a useful trick given the cost of tokens these days and the continuing upward trend in cost from suppliers.

There are also guardrails built into the system to stop confidential information from users who aren’t cleared. Credentials can be set for each user or rolled out across groups - both on a physical network and over the cloud.

While the rest of San Francisco was out partying for Pride, or watching soccer matches, these developers have been coding solidly since Saturday - with the traditional support from energy drinks - and judges have been keeping a close eye on proceedings.

Hackathons are famous for producing code that makes it into wider use. For example, the Facebook Like button was built in an internal hackathon, although Mark Zuckerberg hated it at first. Hopefully this weekend’s competition will produce useful results.

For AI coding, the kids are alright

Iain Thomson — Mon, 29 Jun 2026 16:06:12 +0000

The AI Engineer World's Fair has attracted a lot of adults, and they brought their kids too, for Sunday's AI Engineering Kids Day.

Scientists agree that the best time to learn new languages, be they linguistic or computer-based, is at a young age. 87 children between eight and 16 were in San Francisco to learn how to code their own material and build their own games using AI.

Cassandra Chin from the Cloud Native Computing Foundation, one of the teachers, explained that it was Minecraft that got her interested in software development. As a youngster she set up her own server and joined with others in modifying code. In the session she showed kids how to code a game and wire sensors into an Arduino Nano breadboard.

"What's important when working with kids is you inspire them to enjoy the technology," she said. "They don't have to learn all the syntax and specifics, just as long as you show them that technology is fun. Then, when they go home with their parents, they'll want to continue doing it."

Kaitlyn Hornbuckle, from Oregon State University, also taught a class on using Claude and Godot to develop 3D games. The talk covered how to build prompting skills, build GDScripts, and build a storyline that encourages other players.

"If you give them time to explore on their own, tinker around, and teach them how to be safe online, they might surprise you with what they create," she said.

The Kids Day was sponsored by graph intelligence builder Neo4j. Stephen Chin, VP of Developer Relations, explained that the company was meeting a need.

"There is a significant gap in the current school curriculum," he said. "Students who learn AI early in their career will be much more successful and have more job opportunities after graduation."

Pragmatism in an Age of Infinite Code and Unavoidable Bottlenecks

Ben Halpern — Mon, 29 Jun 2026 16:05:56 +0000

Leading into the AI Engineer event in San Francisco, I’m looking forward to having my mind blown. That being said, I’m also compelled to think deeply about how we actually get things done in today’s software development landscape.

There is competitive pressure to align on the principles of AI-driven software development. As developers and technical leaders, we are constantly trying to balance our pragmatism in the moment with our visions for the future.

Throughout the history of software development, our collective immune system against hype has been our greatest asset. The safest, most reliable strategy has almost always been to not get swept up in the current fad. You let the early adopters bleed on the bleeding edge, you wait for the dust to settle, and then you adopt the tools that actually survive contact with production. Separating hype from reality is a critical skill for developers at all levels, and erring on the side of hype rejection has usually been smart money in the long run.

In the AI revolution, separating real from hype is still valuable, but our heuristics are failing us, because the revolution is clearly here. Yet, even though there is obviously a massive amount of substance to AI-assisted development, the actual daily discourse is still hype-driven. It is an unhelpful, deafening mix of extreme utopian sentiment on one end and cynical over-dismissal on the other.

Finding the signal in that noise requires a hard look at what has actually changed, and more importantly, what hasn't.

The Fallacy of Infinite Code

AI has permanently rewritten the rules for how much software we can produce. In a purely quantitative sense, we now have the capacity to generate effectively infinite amounts of code. But if you’ve spent any time maintaining real-world systems, you know that raw code generation is rarely the true blocker to success.

Actual value delivery is governed by choke points.

These choke points are almost always driven by complex, human-centric factors: decision-makers needing consensus, cross-team collaboration, the friction of merging conflicting ideas, and the overriding need for architectural cohesion. Cohesion is that critical phase where the raw, sprawling net of potential features actually has to be distilled into a product that makes sense.
In software, more is not necessarily more. In fact, unguided “more” usually just means accelerating your technical debt. We still need to deliver precise, focused, and intentional value.

While AI is a massive help in attacking individual choke points — it can scaffold the boilerplate, write the tests, or debug the syntax — the bottleneck doesn't actually disappear. It just moves somewhere else. If we write code 10 times faster, the bottleneck shifts to code review. If we review faster using LLMs, the bottleneck shifts to product alignment and deployment infrastructure. The constraint always moves.

Any development shop that still operates in a “traditional sense” is undeniably falling behind. Writing a spec the way we traditionally have is starting to feel completely redundant—that spec should simply be described directly to your agent of choice. We know there is a need to collaborate at a higher level on the problem at hand, allowing the individual developer to execute many cycles of development without the constant need for specification realignment.

Realizing the value of infinite code in a competitive environment where the bar has been raised for everyone is a fundamentally unsolved problem.

The “Wait for the Next Model” Trap

Because this bottleneck keeps shifting, it creates a specific kind of developer paralysis. Sometimes it feels like building tooling or workflows to resolve the bottlenecks we are seeing today is inherently just a stopgap.

The nagging thought is always there: Why spend cycles optimizing this workflow or building around this friction? We could instead just wait for the models to get smarter. Give it six months, and the agents will just talk to each other and sort this out natively.
This is the trap of anticipating the S-curve. A year or two ago, this logic actually held up. Some of yesterday’s bottlenecks genuinely weren’t worth solving because the next model intelligence leap made those local optimizations totally irrelevant. Building highly specific, complex wrappers around early LLMs was often a waste of time once the next foundational model dropped.

But I think we are at a point where that logic is failing us because we’ve had enough time to collectively learn how we work with our core AI productivity tools. Things are settling down — for the moment. The bottlenecks we face today in software development and value delivery are inherently complicated on a human level, and they extend beyond the scope of near-term AI.

Nothing short of Artificial Superintelligence (ASI) is going to overcome the natural, messy bottlenecks of real-world impact. Until an AI can sit in a room, navigate the political dynamics of a stakeholder meeting, understand the company’s runway, and empathize with the end-user’s actual day-to-day frustrations, these choke points remain tethered to human reality.

We Don’t Need Autopilot, We Need Ergonomics

Yes, we’ve automated some things away. Most things are still semi-autonomous at best, and if those things exist in a black box that surprises us too often and doesn’t hand things off in a usable way to move them along productively, we’re not actually beating our bottlenecks.

Because of this reality, I don’t believe we are at the point where we resolve bottlenecks by taking our hands off the wheel. The prevailing sci-fi vision — that we just let the AI have full, unsupervised control of our computers as a magic bullet for productivity — misses the point of how good software gets built.
Instead, the frontier of real productivity is about intuitive, human-in-the-loop workflows. It’s about ergonomics.

We need environments and command centers where we can ergonomically understand the entire system at a glance. The goal isn’t to completely remove the human from the loop; it’s to make the loop so seamless that the human can operate at a fundamentally higher level of abstraction. We need tooling that helps reduce the cognitive burden of orchestrating developer agents, while still leaving us firmly in the driver’s seat. We need to be able to effortlessly send signals out, course-correct the AI, and manually clear the downstream blockers that the system can’t contextualize.

When a developer can easily see what an agent is attempting, guide it with a single keystroke, and merge that work cohesively into a larger system architecture — that’s when the real value unlocks.

On Skating Where the Puck is Going

We don’t gain much waiting for massive, ground-up system design rethinking to save us from our current bottlenecks. The solution is resolving the friction where we see it, today, with the tools we have right now.

Yes, the models will get better. Yes, the agents will get smarter, and our methods for getting things done will evolve from our successes and failures. But we have reached a spot in the current S-curve with enough maturity to make solving today’s friction highly valuable.

Skate to where the puck is going, not where it has been (to quote Wayne Gretzky). It’s true in AI as well. We can absolutely skate to where the puck is going, but we don’t need to pretend the puck is in a whole different arena. Pragmatism in AI tooling today means accepting the incredible leverage we have right now, building the ergonomic, human-in-the-loop systems to actually harness it, and getting back to delivering precise, cohesive value.

Today we are radically capable of solving yesterday’s problems with tremendous efficiency. However, the tug and pull between day-to-day iterative progress and step-change innovation is still generally a problem we have to manage ourselves. The most pragmatic thing you can do right now is stop waiting for the perfect autonomous system to clear your blockers and start orchestrating the tools you have. Build out your own command centers today — interfaces and workflows that give you high-level observability over your agents and complex workstreams. Invest in the tooling that keeps you effectively in the loop, rather than trying to engineer yourself out of it. The developers who win this cycle won’t be the ones with the smartest unsupervised agents; they will be the ones who build the most ergonomic systems to manage them.

Evolving Our Outcomes

It’s reasonable to build a human-in-the-loop productivity flow with the next phase in mind where the idea of “in the loop” is expected to keep changing. It’s reasonable to build for the future in this sense, but it is not practical to push off today’s problems in favor of waiting around to solve for a hypothetical future.

Build for your people, your team, and your productive customers, not around and in spite of them. They’re more capable than ever. Recognizing this is how we push our outcomes forward.

Coding Agents Play Favorites With Your Dependencies

Adam DuVander — Mon, 29 Jun 2026 16:05:43 +0000

Deep into a coding session, you realize you want beta testers to try some new functionality first. You ask your agent to add feature flagging to your app. It offers you LaunchDarkly’s experimentation solution and a reasonable-looking plan to implement it. With a skim of a Markdown document, you accept its recommendation, and it begins writing code.

This is a realistic scenario, because Claude, ChatGPT, and Gemini all recommend LaunchDarkly. But when you ask these questions of your agent, the response comes from a single model that was asked just once. It’s subject to the same training bias and nondeterminism as any prompt. In my research, the tool recommendations can vary considerably.

How Dependencies Are Chosen

Regardless of whether it’s the agreed-upon leader, the model’s favorite arrives with the same confidence, you give the plan the same light read, and you probably react with the same “looks reasonable.”

That’s a dependency decision. And it was mostly abdicated to your agent.

On one hand, this makes sense. You trust that agent to plan and write the code to build your app, so it’s reasonable to trust its other decisions too. You also can review the code it writes and suggest alternative approaches. More and more, engineers give the code a cursory scan similar to the implementation doc.

Indeed, there are multiple sessions at the AI Engineer World’s Fair that either pronounce code review dead or declare an intention to kill it. With human review as a bottleneck, engineers work toward automated review solutions we can trust. Most organizations will require a robust approach, perhaps at the level of a mathematical proof, as Erik Meijer may suggest in his keynote.

Many dependencies also have a life outside of your codebase, beyond the full visibility of any review process. Before this modern era, two-ish short years ago, dependencies weren’t adopted with academic rigor. But there were usually multiple sets of eyes, and a decision took significantly longer.

Pre-2024, looking for a new tool started with a web search or a chat message to collaborators. You would research alternatives, investigate maintenance issues, and scrutinize open source licenses. There might be an RFC, or at least a teammate’s gut check. There were conversations and data gathering that went into choosing a new tool for your project.

Even with AI agents writing most of the code, some engineers or teams may still use a less automated, well-researched decision-making process. But the modern tools don’t encourage it. The de facto approach is the feature flag story: Ask the agent, get the recommendation, and start the build.

How Top Models Rank Dev Tools

With more engineers turning to their agents for research, I’ve been tracking what gets recommended across common categories such as application databases, managed hosting, and, yes, experimentation platforms like LaunchDarkly. I run the same set of prompts multiple times on the top models and publish the results publicly at llmrank.fyi every month.

Each category ranking is an average of the results from the latest Claude, ChatGPT, and Gemini models. Though LaunchDarkly has remained atop the experimentation leaderboard for three months, No. 2 and No. 3 have been distinct each time. There’s even more change when you look at the differences between models. For example, the latest data saw both Gemini and Claude rank Split.io second, behind LaunchDarkly. ChatGPT did not list the product at all. If you’d asked your agent for multiple feature flag options, you’d get different results based on which model you’re using.

There are many of these disagreements across models. In one fun example, I asked for dev-friendly AWS competitors. ChatGPT returned Azure as one of its responses 100% of the time. Gemini did not include Azure in any of its answers. Conspiracy theories abound.

These disagreements between models are meaningful because it’s not just typical AI noise. The methodology I’ve used represents distributions of recommendations rather than one-offs that an engineer would get from their agent. Based on roughly 50,000 pairwise run comparisons, all three models shared a top-three grouping (regardless of order) in about 58% of cases. In other words, they agree only slightly more than at least one disagrees.

What the Disagreements Actually Mean

Engineers are used to nondeterministic outputs. Two code reviews might provide slightly different responses. That’s expected variation, even within the same model. Dependency recommendations are a little sneakier. They arrive with an authoritative plan and are remarkably consistent within a model.

Gemini provides the same top recommendation 97% of the time. ChatGPT and Claude are within a percentage point. Each agrees with itself, though it frequently disagrees with the others. At minimum, it’s worth a second model’s opinion before you commit.

Claude commands roughly 54% of the enterprise AI market share, according to a Menlo Ventures report from Q4 2025. It’s also the most frequent outlier. Claude is most likely to give you a different top three than the other two. The tool your agent just recommended with confidence may be one that a second opinion would contradict.

The cost of adopting the dependency your model recommends is often just a click of a button. Undoing a dependency that you later regret is significantly more work — even if you make the model apologize and help fix its mistake. There may already be other systems that now require what the tool delivers. You’ll have to update them as well. That’s 42% of cases where a second opinion would have pointed you somewhere else.

Many of the AI Engineer World’s Fair sessions are focused beyond the model. Harness engineering, context optimization, and software factories are about improving workflows and generating more predictable outcomes. Dependencies are one place where model disagreement shows up clearly, but probably not the only one. Find these gaps and fill them with these new engineering disciplines.

Not every engineering team will implement the multi-model approach. The default path is where most dependency decisions will continue to be made. And it's where developer tool companies either show up or don’t. If you work with a product or marketing team, this is the data gap they probably don’t know exists yet.

That feature flag decision, and thousands more like it, gets made whether you think about it or not. The 42% disagreement number will change as models converge or diverge and the market shifts with them. What stays the same is that the decision lands somewhere: Is it with your agent, a single model, a sub-agent, a model routing algorithm, or some human in the loop? All of these are being advocated at the AI Engineer World’s Fair. The best sign is that we’re talking about it.

What to Expect at the AI Engineer World’s Fair 2026

Rachael Berkey — Mon, 29 Jun 2026 16:05:26 +0000

The AI Engineer World’s Fair returns to Moscone West in San Francisco from June 28 through July 2, 2026. It is the largest technical AI conference in the world, with 29 tracks, more than 400 sessions, 100-plus expo partners, and thousands of AI engineers, founders, and VPs of AI in one building. The week opens with New Engineer Orientation on Sunday evening, a full day of hands-on workshops on Monday, then three days of keynotes and up to 12 parallel tracks.

Nobody sees all of it. The attendees who get the most out of the week are the ones who plan before they walk in. Here’s how to do that.

The Basics: When, Where, and How Long

The Fair runs four days at Moscone West, a convention venue in downtown San Francisco. Everything lives under one roof across three floors connected by a single lobby: keynotes, breakout tracks, the expo, the labs, networking, and coffee. An attendee is never more than an escalator ride from the next session. You can find all the necessary details on both the AIE World Fair’s site or the conference app.

The real start of the conference is the evening of Sunday, June 28, when badge pickup and New Engineer Orientation open. Main programming hours run roughly 8 a.m. into the early evening on conference days, with the program grid spanning into the night for receptions and side events. Expo access covers three and a half days, from June 29 through July 2.

Callout! San Francisco is hosting World Cup matches the same week, including a quarterfinal on July 1. Hotel rooms near Moscone are scarcer and pricier than a normal conference week, so lodging and transit deserve more lead time than usual.

Day Zero and Day One: Orientation, Then Building

Sunday, June 28 is New Engineer Orientation (NEO). Early registration and badge pickup run from 5 to 9 p.m., with the orientation itself from 7 to 9 p.m.. NEO exists for first-timers: Pick up a badge, meet a small peer group, ask the questions that feel obvious, and map out the week before the crowds arrive. Attendees who have never done a multitrack conference of this size tend to find it worth the early evening. Sign-ups go through the NEO registration page.

Monday, June 29 is Workshop Day. More than 45 hands-on workshops run across 10 rooms, plus lunch-and-learn sessions, capped by an expo welcome reception in the evening. This is the day to build rather than watch. Workshops fill up, and a beginner who commits to one or two hands-on sessions here will start the keynote days with momentum.

The Main Event: Keynotes and Up to 12 Parallel Tracks

The core of the Fair runs Tuesday through Thursday. Each morning opens with keynotes, then the program splits into 10 engineering tracks plus two leadership tracks running at the same time.

Across the week that adds up to 29 distinct tracks and over 400 sessions. The daily keynote sets the theme: a Coding Agents keynote on Tuesday, an Autoresearch keynote on Wednesday, and a Harness Engineering keynote on Thursday.

The track list is broad enough that almost any specialty has a home. A representative slice includes Security, Evals, Voice and Realtime AI, Computer Use, Context Engineering, Agentic Engineering, Robotics and World Models, Generative Media, Vision and OCR, Local AI, Design Engineering, plus leadership programming for AI-native enterprises and AI architects. New tracks this year point at where the field is heading, with dedicated programming for healthcare, finance, research, and data quality.

The planned speaker lineup includes Mike Krieger (Anthropic), Ryan Dahl (Deno), Addy Osmani, Yohei Nakajima (Untapped Capital), and Romain Huet (OpenAI), among roughly 300 speakers total. The full interactive schedule supports search, filters, and favorites, which is the single most useful prep tool for the week.

With as many as 12 sessions happening at once, the math is simple. No one attends everything live. The Fair is built to be divided and conquered, which is why so many companies send teams that split up and regroup to compare notes. Talks are typically recorded and posted afterward, so missing a session in person is not the same as losing it.

The Expo and the Hallway Track

The expo runs three and a half days and is three times larger than last year’s program, with more than 100 partners, four stages for live demos and talks, and the major AI labs on the floor, including OpenAI, Google DeepMind, Amazon AGI Labs, Minimax, and zAI. A Startup Battlefield runs on July 2, and the expo stays open through the final day, so there is no rush to see all of it on Tuesday.

Don’t miss the hallway track! The conversations that happen between sessions, at the booths, and over coffee are sometimes the most engaging and informative. Experienced attendees consistently rank these as some of the most valuable hours of the week. The Fair leans into this on purpose. There are no official afterparties on July 1 or 2, and side events and community meetups are actively encouraged instead, partly so attendees can catch the World Cup quarterfinal together.

How to Actually Have a Good Week

A few habits separate a productive week from an exhausting blur:
Pick a spine, not a buffet. Anchor each day to one or two tracks that match your current goals, then leave room to wander. Chasing every interesting session across three floors is how people end up seeing nothing well.
Favorite sessions in the app in advance. Use the interactive schedule before arriving. Decisions made in the hallway at 9 a.m. can be worse than ones made the night before.
Do NEO if it is your first time. Two hours on Sunday saves a lot of disorientation on Tuesday.
Talk to strangers. The hallway track only works for people who use it. Booths, lunch tables, and the lines for popular talks are all openings.
Pace yourself. Comfortable shoes, water, and an honest assessment of energy levels matter across a four-day, morning-to-evening event. Burning out on day one is a common and avoidable mistake.
Keep a running notes list. Capture one takeaway per session while it is fresh. By Thursday, the talks blur together, and a short list of what actually landed is worth more than a folder of slides.

The Fair rewards preparation. Pick the tracks, favorite the sessions, book lodging early around the World Cup crowds, and leave room for hallway conversations. Do that, and a week that could feel overwhelming becomes the most useful four days on the calendar.

The AI Engineer World’s Fair Spreads Globally

Iain Thomson — Mon, 29 Jun 2026 16:05:03 +0000

It has been less than three years since Shawn "swyx" Wang coined the term AI engineer in an open letter to the community, but this week thousands of engineers are converging on San Francisco for the now-annual AI Engineer World’s Fair.

As he put it, AI engineers don’t need doctorates and the ability to create LLMs. Rather, they do the important work of integrating machine learning capabilities into current software effectively. “When it comes to shipping AI products, you want engineers, not researchers,” he suggested.

While the latest AI models from the likes of Anthropic, OpenAI, and Google tend to grab the consumer press headlines, the task of integrating these new capabilities into workable, sellable software is arguably more important. And adding APIs without breaking the code is a profession that appeals to many software engineers.

The concept spawned a San Francisco summit, and in 2023 there were 500 software engineers discussing the ins and outs of using this relatively new technology, with many more attendee applications than there were places.

The demand was so great, in 2024 it was renamed the AI Engineer World’s Fair, and all the big software companies were there to show off their latest advances and advise on how to use them in applications. Not to be left out, Nvidia’s CEO Jensen Huang made a couple of surprise appearances to point out his hardware was ready for this.

In 2026 the conference has now spread over four continents, with events in London, San Francisco, and New York, as well as partner conferences in Miami, Singapore, Melbourne, Paris, and Sydney. It looks set to grow further, given the interest in the topic, and software engineers who once saw their jobs as threatened by AI are now embracing it for the roles it will provide in the future.

Welcome to AI Engineer World’s Fair 2026

swyx — Mon, 29 Jun 2026 16:04:30 +0000

Welcome to the first edition of the AI Engineer World’s Fair newspaper!

On behalf of the organizing team, I’m delighted to welcome you to AI Engineer World’s Fair 2026, and to thank MLH and DEV for helping make this newspaper possible.

There are multiple personal full-circle moments here for me.

Ten years ago, at a conference very much like this one, I began my developer writing career on DEV because of a writing challenge they inspired. My first blog post was about a conference talk: what I heard, what I learned, and what I wanted to remember. That habit of writing in public has shaped my career, my learning, my friendships, my opportunities, and, in a very real way, my life. So my first encouragement to you is simple: Write things down. Share what you learned, restate things in your own words, construct new insights by contrast and comparison. Not only because it helps your public profile, though it does, but because reflection is good for the soul (and changing the trajectory of your career).

Three years ago today, I wrote down “The Rise of the AI Engineer,” because of a very simple observation that engineers would both be enabled by AI and would be uniquely able to explore the capability overhang far more effectively than “prompt engineers” on one hand, or, more controversially, “ML engineers” on the other. Since then, the thesis bore out: Everyone from top YC startups to the Metas, OpenAIs, and Anthropics of the world are building out multibillion-dollar AI Engineering/Forward Deployed Engineering orgs and saying “the model alone is no longer the product,” agentic models and the neoclouds/“AI Clouds” have drastically shifted compute workloads from training to inference and sandboxing, and prompt engineering gave way to rigorous evals, RL environments for post-training, and context/harness engineering. While there’s a new shiny thing every month, the fundamentals of both great product/software taste and ML intuition are more needed than ever, so we set out to build a home that is both timely and timeless.

The World’s Fair is now our largest flagship event of the AI engineering community. It has grown across tracks, across topics, and across the world, roughly doubling every year since its inception. Online, AI Engineer now reaches more than 1.5 million unique developers each month through our recorded talks. But those recordings are not the real reason you are here.

You are here because the most important parts of this gathering cannot be captured by an algorithm. You are here to meet people in person, to make friends, to find collaborators, and to discover ideas you would never have clicked on at home. Even as we discuss the future of AI and robotics, this event remains intensely human.

We have worked hard to curate some of the best speakers, sponsors, researchers, builders, and practitioners in the world. Please take the time to explore what they have brought here. Just as importantly, learn from one another. The attendees of this conference are among the best AI engineers in the world, representing finance, healthcare, legal, media, telecom, enterprise, startups, consumer products, and everything in between.

As the curators of AIE, we have tried to focus the program on the most urgent and useful questions of the moment: data quality, memory, continual learning, tokenmaxxing, vertical AI in healthcare, finance, and GTM, as well as enduring foundations like evals, inference, RAG, and security.

For the first time, we are also launching poster sessions, creating a dedicated space for papers, research discussion, and academic exchange. One of the World’s Fair’s most important roles is to bring research closer to industry, because research is where we see the next horizon forming.

We will close the show with our first Startup Battlefield, supported by the HyperAgent team, featuring judges including Garry Tan of Y Combinator. We are seeing a new generation of AI founders emerge from every background, and we want this stage to remind you that you can just build things.

AI Engineer is attended by thousands of people from countries all over the world. Our deepest hope is that you stay connected long after the event ends. We are increasingly bringing AIE to more places through partner events from Miami to Paris, Singapore, Melbourne, Shanghai, and next year, Tokyo.

If I could offer one piece of advice for getting the most out of the World’s Fair, it would be this: Spend 80% of your time looking for ideas you can use at work and 20% of your time exploring things you would never normally seek out.

Increase temperature, do parallel rollouts, continually learn memories… This is what the World’s Fair is for.

Welcome.