DEV Community: Piush Gupta

AI Can't Cite a Company It Can't Identify

Piush Gupta — Tue, 01 Sep 2026 03:35:47 +0000

Getting mentioned by AI isn't an SEO problem. It's an identity problem — and most brands have never separated the two.

Most brands trying to show up in AI answers are running an SEO playbook against a problem that isn't SEO.

The reflex is understandable. For two decades, being found meant ranking: publish the page, earn the links, climb the results. So when buyers started asking ChatGPT and Perplexity instead of Google, teams reached for the same lever — more content, better keywords, tighter pages — and got very little back.

Here's why. Surveys suggest around 42% of US consumers have already used ChatGPT to research a brand or product, and B2B is moving the same direction. But an AI engine doesn't rank your pages. It synthesizes an answer and cites the sources it already trusts — and your own website usually isn't one of them. Before it can decide whether to mention you, it has to do something Google never had to do: work out who you actually are.

The first problem is resolution

That's the step almost nobody is working on. Can the machine reliably identify your company as a single, coherent entity — same name, same category, same identifiers — across the places it reads? Wikidata, LinkedIn, Crunchbase, industry directories, third-party write-ups. If those sources disagree with each other, or if your name collides with another company's, the model does the rational thing: it stays vague, or it picks the wrong company, or it leaves you out of the answer entirely. No amount of on-page optimization fixes that, because the failure happens before your pages are ever consulted.

Try it if you haven't. Ask ChatGPT, Gemini, and Perplexity to describe your company. Three different answers is the normal result. At least one is usually wrong. That isn't the model being careless — it's the model reporting, accurately, that the public record about you is inconsistent.

The second problem is truth

Once the machine can identify you, does it say the right thing? Your actual positioning, your real claims, the things you're required to phrase precisely — or a plausible-sounding guess assembled from fragments? Both problems have the same root cause. The model is improvising because nothing authoritative told it the answer.

And this is where AI discovery stops being a marketing problem and becomes a governance one. The same gap that makes an engine describe you vaguely to a buyer makes your own AI tools describe you wrongly in your own content. Ninety-one percent of marketing teams now use AI, according to Jasper's 2026 report, and an IAB study found 70% of marketers have already had an AI incident — an off-brand output, a claim they couldn't back, a hallucination that shipped. Forty percent pulled ads over it. External misrepresentation and internal drift aren't two problems. They're one problem seen from two sides: there is no machine-readable authoritative version of your brand, so everything downstream guesses.

The uncomfortable part is that the highest-leverage work here is the least glamorous. Consistent naming. Structured data. Making sure the third-party references about you agree with each other. It doesn't look like marketing and it doesn't produce a campaign, which is exactly why it stays undone while teams publish more posts into a system that can't reliably tell who published them.

You don't control the model. You never will. What you control is whether there's a clean, consistent, machine-readable version of your brand for it to find — and whether it's better than the guess it's making right now.

The AI is going to describe you either way. The only question is whether you wrote the source.

kbie is brand governance for the AI era — it turns your brand into a verified knowledge graph, so everything you and your AI tools publish stays on-brand, accurate, and safe to ship. → kbie

40% of Marketers Have Pulled Ads Because of AI. The Other 60% Just Haven't Yet.

Piush Gupta — Tue, 25 Aug 2026 03:34:28 +0000

The adoption problem is solved. The control problem doesn't even have a name yet.

Pulling an ad is an expensive, visible, slightly humiliating decision. Nobody makes it lightly. It means the money is already spent, the creative is already live, and somebody senior has concluded that the cheapest remaining option is to take the work down.

An IAB study found that 40% of marketers have had to do exactly that because of something AI produced. Four in ten teams have reached the point of pulling ads. That is not an edge case being raised by skeptics. That is a category of failure that has already happened to almost half the market.

The number underneath it is larger. The same study found 70% of marketers have already experienced an AI incident: an output that went off-brand, a claim they could not substantiate, a hallucination that made it into something published. Only 6% believe their current safeguards are enough. More than 90% say they want an independent way to check. The 60% who have not pulled an ad yet are not better protected. They are earlier in the same sequence.

So why does it keep happening to competent teams? Because almost everyone has been solving the wrong half of the problem. Enormous effort has gone into adoption: better prompts, more tools, faster generation. Jasper's 2026 report puts AI usage at 91% of marketing teams. Almost no effort has gone into control. The implicit assumption is that if the model is good enough, the output will be safe.

It will not be, because the model does not know your brand. It knows language in general. It has never read your approved-claims list, your regulatory constraints, your do-not-say words, the phrasing legal spent three weeks negotiating. It is fluent, confident, and operating blind. Ask it for a benefit statement and it will produce one. Whether you are allowed to make that claim is a question it cannot even represent.

The default safeguard is human review: catch the problems before they ship. But review is precisely what AI overwhelmed. Jasper's 2026 report found cross-functional review friction — the legal, brand and compliance back-and-forth — rose 3.4x in a single year. You have industrialized production and left inspection manual. The incidents are not a sign that the people are careless. They are a sign that the system is arithmetically guaranteed to leak.

The teams that get this under control do one thing differently. They stop treating "on-brand and compliant" as something you verify after generation, and start treating it as something the generation is grounded in. They give their tools a single, authoritative, machine-readable source of brand truth: voice, approved claims, rules, compliance constraints. The output is then shaped correctly at the moment it is made, and the risky material is caught by construction rather than by luck.

That is not a defensive move only. Research has linked brand consistency to roughly 23% higher revenue. Control and growth are the same project here; the pulled ad is just the version of the problem that generates an invoice.

A pulled ad is the visible failure. The invisible one is everything that did not get pulled: the slightly-off claim that shipped, the softened disclosure nobody flagged, the description of your product that was almost right. Those do not get filed as AI incidents. They get filed as nothing at all.

Forty percent is not a statistic about someone else's bad quarter. It is a leading indicator. The teams that move first from "we hope it's fine" to "we can prove it's fine" are the ones who will scale AI without scaling their incident count.

kbie is brand governance for the AI era — it turns your brand into a verified knowledge graph, so everything you and your AI tools publish stays on-brand, accurate, and safe to ship. → kbie

Nobody Approved It. Nobody Noticed. Your Brand Went Off-Script Anyway.

Piush Gupta — Fri, 21 Aug 2026 03:34:36 +0000

Brand drift in 2026 isn't a rogue-employee problem. It's what happens when every tool you use has to guess who you are.

Somewhere in your company, right now, there is a paragraph nobody wrote. A product description generated in a tab that was open for four minutes. A sales email that quietly softened a claim your legal team spent a quarter getting exactly right. A support reply that promised something you don't actually offer. None of it went through review, because none of it looked like it needed to. It read fine.

That is the part people get wrong about shadow AI. It doesn't announce itself. There is no breach, no incident ticket, no rogue employee to name. There is just competent, confident, slightly-wrong output, produced faster than anyone can read it.

And it is everywhere now. Ninety-one percent of marketing teams use AI in their work, according to Jasper's 2026 report. Somewhere in the last eighteen months, using AI to write stopped being a decision and became the default. What almost no one is tracking is that every one of those tools is making small, confident guesses about who your brand is, and most of them are wrong.

This is structural, not behavioural. Your brand lives in a 40-page PDF, a few Notion docs, a tone-of-voice deck someone made in 2023, and the heads of three senior people. An AI tool cannot read any of that the way a new hire would. So it improvises. It softens a claim you are legally required to phrase precisely. It invents a benefit you never promised. It flattens your voice into the same competent grey as everyone else's. Each instance is tiny. Multiplied across thousands of generated assets a quarter, it is drift: your brand slowly becoming a slightly-off version of itself, in public, at scale.

The damage shows up late, and it shows up wearing a different name. An IAB study found 70% of marketers have already had an AI incident: an off-brand output, a claim they couldn't back, a hallucination that shipped. Forty percent had to pull ads over it. Almost none of those get filed under "brand drift." They get filed under "that campaign that went sideways."

The instinct is to fix it with more review. But review is exactly what AI broke. Jasper's data shows cross-functional review friction, the legal, brand, and compliance back-and-forth, rose 3.4x in a single year. You cannot manually inspect output that is generated in seconds. The math does not close, and it never will.

The real fix is upstream. The reason AI goes off-script is that it has no authoritative source for what is on-script. It is not being careless; it is guessing, because nothing told it the answer. So the move is not to check the output harder. It is to give every tool a single, machine-readable version of the truth: your voice, your approved claims, what is off-limits, what compliance requires. This is the problem kbie is built for. Get it right and "on-brand" stops being a manual gate at the end and becomes a property of the output itself.

Shadow AI is not coming. It is already in your stack, writing in your name. The only question is whether your brand is governing it, or hoping.

kbie is brand governance for the AI era - it turns your brand into a verified knowledge graph, so everything you and your AI tools publish stays on-brand, accurate, and safe to ship. https://kbie.ai

Your Next Buyer Will Meet a Paragraph an AI Wrote About You

Piush Gupta — Tue, 18 Aug 2026 03:35:28 +0000

AI has quietly become the layer buyers use to discover and vet brands. It is describing your company right now — and nobody on your team approved the copy.

Your brand's first impression is no longer something you designed. Increasingly it is a paragraph, generated on demand, delivered to a buyer you will never meet, assembled from whatever an AI could piece together about you. You didn't write it. You didn't approve it. You will probably never read it.

Try it before you read on. Open ChatGPT, Gemini, and Perplexity, and ask each one to describe your company. You'll likely get three different answers. At least one will be vague. At least one will be wrong. And if your name resembles another company's, one of them may be describing someone else entirely.

That used to be a party trick. It is now a distribution problem, because the search box is no longer the front door. Surveys suggest around 42% of US consumers have used ChatGPT to research a brand or product, and the behaviour is climbing fast in B2B — where the buyer doing quiet research months before they ever fill in a form is the norm, not the exception.

Citation is not ranking

Here is the part most teams miss: getting cited by an AI is not the same as ranking on Google. Google ranks your pages, so the work is making your page the best answer. An AI engine doesn't rank — it synthesizes an answer and cites the sources it already trusts, and your own website usually isn't one of them. What it leans on instead is what it can resolve: a consistent identity across the places it reads.

So the work splits into two problems most brands have never separated.

The first is resolution — can the machine reliably identify you at all? Same name, same category, same identifiers across Wikidata, LinkedIn, Crunchbase, industry directories, all agreeing with each other. This is unglamorous entity hygiene, and it is simultaneously the most skipped and the highest-leverage step in the whole game. If the model can't confidently tell who you are, it stays vague, picks the wrong company, or simply doesn't mention you.

The second is truth — when it does describe you, does it say the right thing? Your actual positioning, your actual claims, the things you are required to phrase precisely, rather than a plausible-sounding guess. This is the same failure mode teams are already living with internally. An IAB study found 70% of marketers have had an AI incident — an off-brand output, a claim they couldn't back, a hallucination that shipped. Forty percent had to pull ads over it, and only 6% believe their current safeguards are enough. The mechanism outside your walls is identical to the one inside them: a fluent model, filling a gap you left.

The same root cause

Both problems have the same root cause. The model is improvising because nothing authoritative ever told it the answer. And the instinct — check the output, correct the record afterwards — doesn't survive contact with the volume. Jasper's 2026 report found cross-functional review friction rose 3.4x in a single year, and that is for content you actually control. You cannot review an answer generated privately, on demand, for a buyer you can't see.

You don't control the model. You do control whether there is a clean, consistent, machine-readable version of your brand for it to find. Brands that build that get described accurately and cited by name. Brands that don't get summarized by an algorithm's best guess — to buyers who never knew a better answer existed.

The AI is going to describe you either way. The only real choice is whether you wrote the source it is reading from.

kbie is brand governance for the AI era — it turns your brand into a verified knowledge graph, so everything you and your AI tools publish stays on-brand, accurate, and safe to ship. → https://kbie.ai

You Can Make a Hundred Ads Before Lunch. Legal Still Needs Four Days.

Piush Gupta — Fri, 14 Aug 2026 03:36:30 +0000

The constraint moved, and most teams are still optimizing the part that's no longer the problem.

For twenty years the scarce resource in marketing was production. Making the thing — the copy, the creative, the page — was slow and expensive, so that's where the tools, the headcount, and the budget went. AI quietly ended that era. Production is now effectively free and effectively instant. Ninety-one percent of teams use AI to make things, per Jasper's 2026 report.

When a constraint disappears, it doesn't vanish — it moves. And it moved to the one step nobody industrialized: approval. The brand check, the legal review, the compliance sign-off. Jasper's data shows cross-functional review friction rose 3.4x in a single year. That's the sound of a bottleneck relocating. You can generate a hundred variations before lunch and then wait four days for someone to confirm they're allowed to exist.

Most teams are responding by optimizing the wrong end. They buy another generation tool, write better prompts, add more output — pouring water into a funnel that's already overflowing at the bottom. The pile of unapproved content grows. The review queue grows. The cycle time barely improves, because the limiting step was never production.

The reason approval is slow is worth being precise about. Review is slow because it's where brand knowledge finally gets applied — and that knowledge lives in people's heads and static documents, not in the system that made the content. So a human has to manually re-apply it, asset by asset, after the fact. It's slow because it's a manual re-injection of context the generator never had.

Which points at the fix. You don't speed up approval by hiring more reviewers; you speed it up by moving the brand knowledge upstream, into the generation step, so most of what review used to catch never gets created. Encode the voice, the approved claims, the rules, the compliance constraints into a single machine-readable source every tool draws from. Now the output arrives mostly-right, and review shrinks from "inspect everything" to "handle the genuine edge cases." That's the difference between governance as a brake and governance as a rail — same control, a fraction of the drag.

The teams that win the next phase of AI won't be the ones generating the most. Everyone can do that now. They'll be the ones who fixed the bottleneck everyone else is still ignoring.

kbie is brand governance for the AI era — it turns your brand into a verified knowledge graph, so everything you and your AI tools publish stays on-brand, accurate, and safe to ship. → https://kbie.ai

Only 6% of Marketers Think Their AI Safeguards Are Enough. The Other 94% Are Right.

Piush Gupta — Tue, 11 Aug 2026 03:36:19 +0000

The AI brand incident isn't a future risk. It's already the majority experience — and more review won't fix it.

There's a number in an IAB study that should have stopped the industry cold, and mostly didn't: when marketers were asked whether their current safeguards for AI-generated content were sufficient, only 6% said yes.

Six percent. In a profession that is otherwise very good at telling itself things are under control.

The rest of that study explains why the confidence is so low. An IAB study found that 70% of marketers have already experienced an AI incident — an output that went off-brand, a claim they couldn't substantiate, a hallucination that made it into something published. Forty percent had to pull ads as a result. And more than 90% said they want an independent way to check AI output before it ships.

Read those together and the picture is unambiguous. This isn't a fringe risk raised by skeptics on the sidelines. It's the majority experience of teams already shipping with AI, and they know their defenses don't hold. They're describing the present, not worrying about the future.

So why does it keep happening?

Because almost every team is solving the wrong half of the problem. Enormous effort has gone into adoption — better prompts, more tools, faster generation. In Jasper's 2026 report, 91% of teams now use AI in their work. Almost none of that effort has gone into control.

The implicit assumption is that if the model is good enough, the output will be safe. It won't, because the model doesn't know your brand. It knows language in general. It has never read your approved-claims list, your regulatory constraints, your do-not-say words. It is fluent, confident, and operating blind.

Review is the thing AI broke

The default safeguard is human review — catch the problems before they ship. But review is precisely what AI overwhelms. Jasper's 2026 report found cross-functional review friction, the legal and brand and compliance back-and-forth, rose 3.4x in a single year. You have industrialized production and left inspection manual.

The incidents aren't a sign that the people are careless. They're a sign the system is mathematically guaranteed to leak.

That is what the 6% is actually measuring. Not pessimism — arithmetic. Those marketers can see that the only thing standing between generated content and the public is a human being reading it, and they can see how much content there now is.

The fix is upstream, not downstream

The teams that get this under control do one thing differently. They stop treating "on-brand and compliant" as something you verify after generation, and start treating it as something the generation is grounded in. They give their tools a single, authoritative, machine-readable source of brand truth — voice, approved claims, rules, compliance constraints — so output is shaped correctly at the moment it is made, and the risky material is caught by construction rather than by luck.

Notice what that does to the review queue. It doesn't remove the humans; it changes what reaches them. Instead of inspecting everything on the chance that something is wrong, they handle the genuine edge cases. Governance stops being a brake and becomes a rail.

Seventy percent isn't a statistic about somebody else's bad quarter. It's the baseline. And 6% is the honest internal assessment of what is currently protecting you. The teams that move first from "we hope it's fine" to "we can prove it's fine" are the ones who will scale AI without scaling their incident count.

Shadow AI Doesn't Look Like a Breach. It Looks Like Slightly-Wrong Copy.

Piush Gupta — Tue, 28 Jul 2026 03:35:50 +0000

The riskiest thing in your stack isn't a rogue employee. It's a thousand small, confident guesses about who your brand is.

When people hear "shadow AI," they picture a security incident: someone pasting confidential data into a chatbot, a policy violation, a breach with a date and an owner. That version is easy to imagine and easy to write a policy about. The real one is duller and far more expensive. It looks like copy that is ninety percent right.

Somewhere in the last eighteen months, using AI to write stopped being a decision. It became the default. A product description here, a sales email there, a deck, a support reply, a landing page. Ninety-one percent of marketing teams now use AI in their work, according to Jasper's 2026 report. The output is faster than anything that came before it.

Here's what almost no one is tracking: every one of those tools is making small, confident guesses about who your brand is, and most of them are wrong.

This isn't about rogue employees. It's structural. Your brand lives in a 40-page PDF, a few Notion docs, a tone-of-voice deck someone made in 2023, and the heads of three senior people. An AI tool can't read any of that the way a new hire would. So it improvises. It softens a claim you're legally required to phrase precisely. It invents a benefit you never promised. It flattens your voice into the same competent grey as everyone else's. Each instance is tiny. Multiplied across thousands of generated assets a quarter, it's drift: your brand slowly becoming a slightly-off version of itself, in public, at scale.

The damage shows up late and looks like something else. An IAB study found that 70% of marketers have already had an AI incident: an off-brand output, a claim they couldn't back, a hallucination that shipped. Forty percent had to pull ads over it, and only 6% believe their current safeguards are enough. These don't get filed under "brand drift." They get filed under "that campaign that went sideways."

The instinct is to fix it with more review. But review is exactly what AI broke. Jasper's data shows cross-functional review friction, the legal and brand and compliance back-and-forth, rose 3.4x in a single year. You cannot manually inspect output that's generated in seconds. The math doesn't close.

The real fix is upstream. The reason AI goes off-script is that it has no authoritative source for what's on-script. It's guessing because nothing told it the answer. So the move isn't to check the output harder. It's to give every tool a single, machine-readable version of the truth: your voice, your approved claims, what's off-limits, what compliance requires. Get that right and "on-brand" stops being a manual gate and becomes a property of the output itself.

Research has linked brand consistency to roughly 23% higher revenue. That number used to be an argument for discipline. It is now an argument for infrastructure, because consistency at machine speed is not something a style guide can enforce.

Shadow AI isn't coming. It's already in your stack. The only question is whether your brand is governing it, or hoping.

Marketing Doesn't Have a Production Problem Anymore. It Has an Approval Problem.

Piush Gupta — Fri, 24 Jul 2026 03:36:56 +0000

AI made content free. It just moved the bottleneck to the one step nobody automated.

Count how many pieces of content your team touched today — drafts, variations, first passes, whatever an AI tool produced before a human opened the doc. Now count how many of those are actually live, approved, published. The gap between those two numbers is where your team's time is going, and it has nothing to do with writing.

For most of marketing's history, the expensive part was making the thing. Copywriters, designers, production cycles — building content was slow, so companies built org charts around producing less of it, more carefully. AI broke that constraint completely. Ninety-one percent of teams now use AI to create marketing content, per Jasper's 2026 report, and a single person can generate more drafts before their first coffee than a team used to ship in a week.

You'd expect that to make marketing faster. Instead, a lot of teams report feeling more backed up than ever. Here's why: a constraint doesn't disappear when you remove it, it just relocates to whatever's next in line. And the next step in line — brand check, legal sign-off, compliance review — was never built to absorb this kind of volume. Jasper's data shows cross-functional review friction is up 3.4x in a year. That's not a rounding error. That's the whole system's bottleneck changing addresses.

Most teams respond to this by doing more of what got them here: another generation tool, a better prompt library, higher output. It doesn't help, because the queue was never short on drafts. It's short on people who can say "yes, this is safe to publish" fast enough to keep up.

It's worth asking why approval is slow in the first place, because the answer points straight at the fix. Review takes time because it's the moment brand knowledge finally gets applied to the content — and that knowledge doesn't live in the tool that generated the draft. It lives in a person's head, or a style guide nobody opens, or a compliance doc from two reorgs ago. So a human has to manually reconstruct all of that context, asset by asset, after the content already exists. That's not a people problem. That's a sequencing problem.

The fix isn't hiring more reviewers to keep pace with AI-generated volume — that's a losing race by design. It's moving the brand knowledge earlier in the pipeline, into the generation step itself, so the draft that comes out the other end is already mostly right. Encode the voice, the approved claims, the regulatory constraints, the things that are always off-limits, into one machine-readable source every tool pulls from. Review stops being a full inspection of everything and becomes what it should've been all along — a check on genuine edge cases.

The teams pulling ahead right now aren't the ones generating the most content. Every competitor can do that. They're the ones who noticed the bottleneck moved and actually went and fixed where it moved to.

kbie is brand governance for the AI era — it turns your brand into a verified knowledge graph, so everything you and your AI tools publish stays on-brand, accurate, and safe to ship. → kbie

Your Homepage Isn't the Front Door Anymore

Piush Gupta — Tue, 14 Jul 2026 03:35:32 +0000

Buyers now meet your brand through a paragraph an AI wrote about you — and it's writing that paragraph today, with or without your input.

Somewhere in your pipeline right now is a buyer who will never visit your website. They asked ChatGPT — or Gemini, or Perplexity — what your company does, whether it's credible, and how it compares to the alternatives. What they got back was a confident, synthesized paragraph. That paragraph is your first impression now. Surveys suggest around 42% of US consumers have used ChatGPT to research a brand or product, and the same behavior is climbing fast in B2B.

Run the test yourself. Ask all three engines to describe your company. You'll likely get three different answers. At least one will be vague. At least one will be wrong. And if your name resembles another company's, one of them may be describing someone else entirely. Two years ago that was a party trick. Today it's a pipeline problem.

Here's the part most teams miss: getting cited by an AI is not the same as ranking on Google. Google ranks your pages. An AI engine synthesizes an answer and cites the sources it already trusts — and your own website usually isn't one of them. It leans on what it can resolve: a consistent identity across the places it reads. If the model can't confidently tell who you are — same name, same category, same identifiers across Wikidata, LinkedIn, Crunchbase, industry sites — it stays vague, or it picks the wrong company, or it doesn't mention you at all.

So the work splits into two problems most brands have never separated. The first is resolution: can the machine reliably identify you at all? This is unglamorous entity hygiene — consistent naming, structured data, third-party references that agree with each other. It's the most skipped and highest-leverage step in the whole game. The second is truth: when the machine does describe you, does it say the right thing — your real positioning, your actual claims — rather than a guess?

Both trace back to the same root cause. The model is improvising because nothing authoritative told it the answer. You don't control the model. You do control whether there's a clean, consistent, machine-readable version of your brand for it to find. Brands that build that get described accurately and cited by name. Brands that don't get summarized by an algorithm's best guess — to buyers who never knew there was a better answer.

The AI is going to describe you either way. The only choice is whether you wrote the source it's reading from.

Everyone Solved AI Adoption. Almost No One Has Solved AI Control.

Piush Gupta — Fri, 10 Jul 2026 03:40:39 +0000

In 2024 the question was whether to use AI. In 2025 it was which tools. In 2026 the only question that matters is quieter: can you trust what it produces? Nearly every team has answered the first two — 91% of marketing teams now use AI in their work, per Jasper's 2026 report. Almost nobody has answered the third.

An IAB study put a number on the gap: 70% of marketers have already experienced an AI incident — an output that went off-brand, a claim they couldn't substantiate, a hallucination that made it into something published. Forty percent had to pull ads as a result. And when asked whether their current safeguards were enough, only 6% said yes. More than ninety percent said they want an independent way to check.

Sit with that for a second. This isn't a fringe risk raised by skeptics. It's the majority experience of teams already shipping with AI. The incident already happened. They're describing the present, not a worry about the future.

So why does it keep happening? Because almost every team is solving the wrong half of the problem. Enormous effort has gone into adoption — better prompts, more tools, faster generation. Almost none has gone into control. The implicit assumption is that if the model is good enough, the output will be safe. It won't be, because the model doesn't know your brand. It knows language in general. It has never read your approved-claims list, your regulatory constraints, your do-not-say words. It's fluent and confident and operating blind.

The default safeguard is human review — catch the problems before they ship. But review is precisely what AI overwhelms. Jasper's 2026 report found cross-functional review friction rose 3.4x in a year. You've industrialized production and left inspection manual. The incidents aren't a sign the people are careless; they're a sign the system is mathematically guaranteed to leak.

The teams that get this under control do one thing differently: they stop treating "on-brand and compliant" as something you verify after generation, and start treating it as something the generation is grounded in. They give their tools a single, authoritative, machine-readable source of brand truth — voice, claims, rules, compliance — so the output is shaped correctly at the moment it's made, and the risky stuff is caught by construction rather than by luck.

Seventy percent isn't a statistic about other people's bad day. It's the baseline. The teams that move first from "we hope it's fine" to "we can prove it's fine" are the ones who'll scale AI without scaling their incident count.

kbie is brand governance for the AI era — it turns your brand into a verified knowledge graph, so everything you and your AI tools publish stays on-brand, accurate, and safe to ship. → kbie

Your Brand Has a Thousand Ghostwriters Now — and None of Them Read the Guidelines

Piush Gupta — Tue, 07 Jul 2026 03:38:07 +0000

Somewhere in your company, right now, an AI tool is writing as your brand. A product description, a support macro, a sales follow-up, a landing-page variant. Nobody approved the exact words. Nobody will review most of them. According to Jasper's 2026 report, 91% of marketing teams now use AI in their work — which means the average brand has quietly acquired a thousand ghostwriters, and not one of them has read the brand guidelines.

That is the real story behind shadow AI. It is not rogue employees sneaking tools past IT. It is structural: your brand lives in a 40-page PDF, a few Notion docs, a tone-of-voice deck someone made in 2023, and the heads of three senior people. An AI tool cannot read any of that the way a new hire would. So it improvises. It softens a claim you are legally required to phrase precisely. It invents a benefit you never promised. It flattens your voice into the same competent grey as everyone else's. Each instance is tiny. Multiplied across thousands of generated assets a quarter, it is drift — your brand slowly becoming a slightly-off version of itself, in public, at scale.

The damage shows up late and looks like something else. An IAB study found 70% of marketers have already had an AI incident — an off-brand output, a claim they couldn't back, a hallucination that shipped. Forty percent had to pull ads over it. These don't get filed under "brand drift." They get filed under "that campaign that went sideways."

The instinct is to fix it with more review. But review is exactly what AI broke. Jasper's data shows cross-functional review friction — the legal, brand, and compliance back-and-forth — rose 3.4x in a year. You cannot manually inspect output that is generated in seconds. The math doesn't close.

The real fix is upstream. The reason AI goes off-script is that it has no authoritative source for what is on-script. It is guessing because nothing told it the answer. So the move is not to check the output harder — it is to give every tool a single, machine-readable version of the truth: your voice, your approved claims, what is off-limits, what compliance requires. Get that right and "on-brand" stops being a manual gate and becomes a property of the output itself.

Your ghostwriters are not going away. The only question is whether you hand them the script, or keep letting them improvise.

kbie is brand governance for the AI era — it turns your brand into a verified knowledge graph, so everything you and your AI tools publish stays on-brand, accurate, and safe to ship. → kbie

AI Made Content Free. Approval Became the New Bottleneck.

Piush Gupta — Fri, 03 Jul 2026 03:39:45 +0000

The constraint moved, and most teams are still optimizing the part that's no longer the problem.

When a constraint disappears, it doesn't vanish — it moves. And it moved to the one step nobody industrialized: approval. The brand check, the legal review, the compliance sign-off. Jasper's data shows cross-functional review friction rose 3.4× in a single year. That's the sound of a bottleneck relocating. You can generate a hundred variations before lunch and then wait four days for someone to confirm they're allowed to exist.

The teams that win the next phase of AI won't be the ones generating the most. Everyone can do that now. They'll be the ones who fixed the bottleneck everyone else is still ignoring.