OpenAI is putting images into ChatGPT ads. On October 5, 2026, it announced a visual ad format that shows up while ChatGPT generates an image. But the targeting that would make those ads worth a bootstrapper’s money depends on a setting most advertisers never think about.

Until they reliably reach the right buyer, I’d put that budget into creating content that you own and that compounds over time.

OpenAI also expanded its measurement integrations, said it’s developing brand-suitability pilots with DoubleVerify and Integral Ad Science, and shared early advertiser results.

Most of that is about formats, measurement and brand safety. I think OpenAI should fix targeting before it starts adding new ad formats. ChatGPT can already use past chats and memory to match an ad to the right buyer, but only for users who’ve switched on ad personalization, and advertisers have no way to know how many have. Until that changes, I’d tell bootstrapped founders, and anyone advising them, to wait before putting their limited budgets into ChatGPT ads.

How ChatGPT ads use your past chats and memory

Bootstrapped founders have very limited budgets, and they can’t waste them on ads that reach a broad audience waiting for their images to be generated. Every dollar has to get to their ICP: the person who fits and is ready to buy.

With its huge context window and memory, ChatGPT knows a lot about whoever it’s chatting with. Buyers explain their situation to it at length, the way they would to a salesperson on a call. It could personalize ads better than anyone right now.

OpenAI’s help page even says personalized ads can draw on past chats and memory, through a toggle called Past chats and memory.

If it’s on, past chats and memory can contribute to ad relevance. If it’s off, ads rely more on your current chat thread, and past chats and memory won’t be used to inform the ads you see.

The ChatGPT ad personalization toggle advertisers can’t control

When you buy a ChatGPT ad, you pick the copy, the landing page, the bid and your targeting. The one thing you can’t pick is whether the person seeing it has switched on ad personalization. That’s their call, made in Settings > Ad Controls, and it decides how much ChatGPT can use to match your ad to them.

ChatGPT only uses past chats, memory and ad history to pick ads when ads personalization is enabled. With it off, ChatGPT picks ads from the current chat thread plus basic context like general location and language. That’s plain contextual matching: one conversation, one guess.

It also changes by market. OpenAI says personalized ads aren’t initially available in the European Economic Area or Switzerland at all, and Help Net Security reported in August that OpenAI asks users for a choice before personalized ads begin.

Asking people for permission rarely goes well for advertisers, either. When Apple started asking iPhone users whether apps could track them, just 4% of US users opted in in the first 2 weeks, according to Flurry data. I’d bet most ChatGPT users have never even opened Ad Controls.

Digiday reported that advertisers keep raising the same complaints about ChatGPT ads: weak measurement, too few formats and not enough inventory. It quoted Zeno Group’s Shamsul Chowdhury, who said ChatGPT’s smaller prompt volume and single ad placement create a disconnect between supply and demand. The new format and measurement tools take aim at the first 2 complaints. With inventory that tight, I’d want every impression reaching the right buyer first, and right now nobody can promise that.

Do ChatGPT ads work? Early results aren’t a pattern yet

OpenAI backed the announcement with partner-measured results for WeightWatchers, Dose and Portland Leather. For Dose, WorkMagic reported a statistically significant lift, with 67% of incremental purchases coming from net-new customers.

If it works for Dose, it doesn’t mean it’s going to be the same percentage for everyone else, and a handful of results published by OpenAI itself isn’t a pattern yet. OpenAI also says its work on incrementality is still in its early stages.

And I want to see more under the hood, especially how ChatGPT actually decides whether to serve an ad to someone at all, and how many of the people who saw those ads had personalization on.

I judge new AI ad channels by what founders I know actually spend and what comes back. A friend of mine, an award-winning interior designer for residential and commercial projects, put quite a big budget into a ChatGPT ads test, and unfortunately, absolutely nothing came back.

That surprised me even more because ChatGPT now has its own interior design option in the Images tab, called “Redesign any room.” People thinking about a renovation are already showing ChatGPT exactly what kind of room they want, which should make it easier than ever to match them with a designer. Even with that, her ads brought nothing in.

How to get recommended by ChatGPT without paying for ads

Until ChatGPT gets there, I’d put a bootstrapper’s limited time and money into organic content. Content keeps working for you, while paid ads are rented visibility that keeps you in bidding wars. ChatGPT is no exception: when more than one ad is eligible, OpenAI says it ranks them on a mix of relevance and advertiser bids.

1. Write for one buyer’s problem at a time. OpenAI says ads don’t influence ChatGPT’s answers, so getting into those answers still takes organic content.
2. Double down on what already brings customers in. That same interior designer gets most of her customers from referrals, and nothing she’s tried on Facebook, Google or ChatGPT ads comes close to that organic reach.
3. Set your own signal for when to test. Switch on personalization and memory in your own Ad Controls and watch the ads you get. When they start matching your conversations and what ChatGPT knows about you, test with a small budget.

ChatGPT already uses memory to personalize its answers, down to refining a restaurant search with your location or dietary preferences. That’s exactly what I want it doing with its ads, for every user who sees them. When it gets there, ChatGPT could be the best-targeted ad channel bootstrappers have ever had.

And the founders who spent the wait building organic content will walk in with an audience that already trusts them, so when they finally pay for ads, they’ll be amplifying something that already works.

About the Author

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post ChatGPT Ads Got Images, but Targeting Depends on a Toggle appeared first on BYVI.

Does Google pay for AI Overviews now?

Sort of, and only if it decides your page mattered enough. This month Google confirmed it’s running what it calls the “AI contribution pilot,” a Search Console feature that shows eligible publishers a running tally of earnings tied to how often their content shapes an answer inside AI Overviews, AI Mode, and Gemini.

The interface line reads: “Accumulate earnings when your content contributes significantly to AI-generated responses in participating products,” as first reported by Digiday and confirmed the same week by Search Engine Land.

Google’s own framing of the program is careful and small. A spokesperson called it “an early-stage learning pilot to test how best to reward high-quality content, on top of the traffic and tools it already provides,” according to Digiday’s reporting.

So basically it’s a test of a test. This isn’t Google’s first move here either. It teased the concept in a June blog post about paying sites that keep its AI answers “fresh and accurate,” and announced a separate “commercial partnership program” with major news publishers back in late 2025.

The Search Console rollout is the small, quiet, unglamorous version of that promise, and dozens of publishers have reportedly been invited in already, skewed toward small and midsize sites, not the large publishers Google would rather cut a real deal with.

The rollout itself tells you how much weight to put on it. Search Engine Land contributor Barry Schwartz first spotted the “AI contribution” report inside Search Console back in April, months before anyone outside Google understood there was a payment attached to it. It took until this week, and a Digiday report, for the payment mechanism to surface publicly at all.

A program this quiet, aimed at a handful of invited publishers, isn’t a policy shift you can plan a business around. It’s a pilot Google can shrink, pause, or quietly kill the moment it stops learning what it wants to learn.

If you run a bootstrapped content business and you’re wondering whether to hold out for a Google check before you change anything about your content strategy, the honest answer is: don’t.

The reasons why get clearer the closer you look at how the money actually works.

What AI Mode publisher payments actually look like from the inside

One publishing executive with early access to the program described the payout math directly to Digiday: “It’s quite black box.” Nobody outside Google can see the formula that decides which page “contributed significantly” to which answer, or how many cents that contribution is worth.

Another participant, more hopeful, said they’re “super excited to participate” because the open dialogue with Google “feels like a true partner.”

Both things can be true. A collaborative pilot and an opaque, unaccountable payout are not mutually exclusive, and for a small publisher with no seat at that table, the second part is what matters the most.

Luke Stillman, managing director at the media consultancy Madison and Wall, put the power imbalance:

“Publishers have relatively little leverage over how AI changes content discovery and distribution.”

David Buttle, founder of the publisher coalition Spur, went further, calling the whole pilot “a kind of hedge” on Google’s part, “a bit of experimentation in the event that that world comes to pass and they do actually have to pay” real money for real usage.

None of that makes the pilot worthless. It makes it a signal, not a plan. Google acknowledging it owes something for the content powering its AI answers is a real shift after two years of publishers shouting into the void about traffic collapsing under AI Overviews.

But a signal doesn’t pay your AWS bill, and independent publishers who builds a content strategy around waiting for Google to decide what your work is worth has handed away the one thing they actually control.

Why Reddit can cash this check and you can’t

Look at who actually gets paid real money for AI training content, and the pattern is obvious. Reddit signed a licensing deal with Google in February 2024 worth a reported $60 million a year, for access to what Google called “real-time content from Reddit’s vast user-authored forums.”

A separate deal with OpenAI followed months later, reportedly worth around $70 million annually. Reddit could ask that price because it owns something no other site has: millions of unscripted human conversations that AI models need and can’t fully replicate anywhere else.

Between August 2024 and June 2025, Reddit became the single most-cited domain across both Google AI Overviews and Perplexity.

That’s the model. Reddit didn’t get paid because it waited patiently for a fair pilot program. It got paid because it holds a monopoly on a specific kind of data, and it negotiated from that position of strength, before Google or OpenAI ever offered a Search Console widget to anyone.

Your blog or publication does not have that leverage, and neither does mine.

The moat is the citation

Founders keep reporting the same strange pattern this year: rankings holding steady while traffic quietly drains away. The data backs them up. Ahrefs compared Search Console data across 300,000 keywords and found that AI Overviews cut click-through rates for top-ranking pages by 58%, up from 34.5% in their earlier 2025 study.

As they put it: “For every 100 clicks you could historically earn for a top-ranking page, Google now keeps 58.” That’s the mechanism behind every founder post wondering why months of stable positions produced falling visits.

Stable rankings, falling traffic.

That gap is the whole AI-search story in one sentence, and it’s happening to founders who’ve never heard of the AI contribution pilot and never will.

Research from Ahrefs, analyzing nearly 17 million citations across ChatGPT, Perplexity, Gemini, and AI Overviews, found that AI assistants cite meaningfully fresher content than traditional Google search results do, roughly 25.7% fresher on average.

ChatGPT showed the strongest bias toward recency of any platform tested. That’s a concrete, actionable lever available to any founder today, no pilot invitation required.

How to get cited by AI without waiting on Google

What actually moves the needle, while Google figures out its own math, is more boring than a licensing deal and far more useful to a founder with no leverage to negotiate with in the first place.

Where this leaves founders

Google’s AI contribution pilot is worth watching, and it’s worth opting into if you’re invited, since free money is still money. But build your business as if that check never clears. The founders who’ll matter most in this next stretch of AI search are the ones who made themselves the source worth citing long before any platform decided to pay for the privilege, and who never stopped building the parts of their business no algorithm gets to price.

About the Author

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post Google’s AI Contribution Pilot: Founders, Don’t Count on It and Do This Instead appeared first on BYVI.

A signed licensing deal didn’t save Reddit’s visibility when OpenAI quietly changed how ChatGPT searches. If a company that just joined the S&P 500 can vanish from the answer overnight, your startup can too. The only durable defense is owning what the algorithm can’t revoke.

The short version

• Reddit’s share of ChatGPT citations fell from 3.8% to 0.5% in about a week, an 86% drop, after an unannounced August 8 change to how ChatGPT searches.
• A licensing deal doesn’t protect visibility. Reddit has paid contracts with OpenAI and Google and still lost most of its presence in the answer.
• This is platform risk, and it has happened before. Reddit’s ChatGPT citations collapsed once already in September 2025.
• The founder move is owned media plus diversification. Depth on domains you control is the only visibility an algorithm can’t take back on a Tuesday.

Reddit spent the past year as one of the sites ChatGPT cited more than almost anything else. Then, in a single week this month, it nearly stopped showing up at all.

Between July 18 and August 7, Reddit accounted for 3.8% of every source ChatGPT cited. By the week of August 14 that number was 0.5%, an 86% drop in about a week, according to data from the analytics firm PromptWatch. No announcement. No warning. Founders who had built a slice of their AI-search visibility on Reddit threads woke up to a channel that had quietly closed.

Every founder I talk to wants their product showing up in ChatGPT’s answers. Almost none of them have asked what happens when the answer changes its mind. This is what that looks like.

What actually happened to Reddit’s ChatGPT citations

On August 8, ChatGPT changed how it searches. Instead of running broad web queries, it started aiming far more of them at specific domains using the site: operator, the same trick you’d use to search inside one website on Google. Domain-scoped queries jumped from 0.37% to 16.8% of its background searches in a single day, a 46-fold increase. When the model stops casting a wide net and starts asking named sites directly, whatever isn’t on its shortlist falls out of the answer. Reddit fell out.

The timing isn’t clean. The search change landed August 8, the sharpest Reddit drop hit August 14, and PromptWatch itself calls the number provisional and says it can’t rule out a problem on its own end. OpenAI hasn’t explained any of it.

Why this is platform risk, and why it’s bigger than Reddit

Reddit itself will be fine. It signed a reported $70-million-a-year deal with OpenAI and a $60-million one with Google, it spent the better part of a year as the second most-cited domain in ChatGPT, and it joined the S&P 500 on August 18, the same week its ChatGPT visibility cratered.

If a company that size, with signed contracts, can lose most of its presence in the answer overnight, your startup can lose all of it, and nobody will ever notify you.

Reddit is protected in a way you never will be. It gets paid whether or not it shows up in ChatGPT, because it licensed its data separately from how it gets cited.

You get neither. Your visibility in an AI answer is a favor the model is currently doing you, and favors get withdrawn. When one platform decides how buyers find their answers, one back-end tweak can erase a channel you spent months building.

It already happened once, in September 2025

If this feels familiar, it should. Reddit’s ChatGPT citations collapsed once before. Semrush studied 230,000 prompts and more than 100 million AI citations across a 13-week stretch in 2025 and watched Reddit’s ChatGPT citation frequency fall from roughly 60% in early August to around 10% by mid-September, a slide that lined up with Google quietly removing a search parameter called num=100.

Two lessons from that one.

First, the pattern repeats: a back-end change nobody flagged, a visibility cliff nobody chose.

Second, and this is the part worth holding onto, Reddit recovered enough to stay one of ChatGPT’s two most-cited domains through the end of 2025. So re-rankings are volatile in both directions. And that is exactly why you can’t run a business on one.

What survives an AI re-ranking: the founder playbook

You can’t control how ChatGPT decides to search. You can control whether your visibility depends on it. Here is exactly what I tell startup founders to build, in order.

The bottom line

AI search is turning into a single point of failure, and this week made it visible. Build for the answer, absolutely, but never bet the company on one engine’s mood. Reddit just took an 86% hit with a $130-million safety net under it and a spot in the S&P 500. You don’t have either. Own your depth, spread your bets, and watch your ChatGPT referral traffic in Google Analytics.

Questions founders are asking me

About the author

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post Reddit’s ChatGPT Citations Collapsed: 86% Gone in a Week appeared first on BYVI.

You raised money, and the first thing the money wants to buy is content that makes you sound exactly like the companies you raised it to beat.

There’s a pitch sitting in my inbox right now. 40 blog posts a month, flat fee, “AI-assisted, human-edited.” Third one this month. They’re all addressed to founders who just announced a round, because of course they are. That’s who has budget, deadlines, and a board asking about the content engine.

I want to talk you out of signing it.

The swap-the-logo test for commodity content

Take any post on your blog. Swap your logo for your closest competitor’s. If nobody would notice, not your customers, not your own team, you have published commodity content.

“Top 10 Tips for SaaS Growth.” “The Ultimate Guide to Whatever.” The ultimate guide is never written by the person who knows the most. It’s written by the person who was assigned the keyword.

This is not new, by the way. Agencies produced this for years before ChatGPT, mostly through no fault of their own. A freelancer who has never sat in one of your sales calls physically cannot write anything except the average of the internet. And AI didn’t change the quality of that content, it removed every limit on how much of it you can produce.

What’s new is that you can buy it at volume now. 40 posts a month. 50. The number in the pitch changes, the promise doesn’t. And funded founders are the customers for this, almost always, because a bootstrapper looks at the invoice and feels it.

How AI search treats commodity content

Your buyers changed how they search and I don’t think most founders have absorbed it yet. They don’t type keywords. They open ChatGPT and describe their situation (the stage, the issue, the thing that broke) the same way they’d describe it to a salesperson on a call. Long, messy, but very specific.

And the model synthesizes one answer rather than ranking a page of links. It has read everything: your blog, your competitors’ blogs, your G2 reviews, the Reddit thread where someone complained about your onboarding (though as of this week, ChatGPT cites Reddit far less).

Then AI forms an opinion, and it hands that opinion to your buyer with total confidence. There’s no page 2. You’re either in the answer or you don’t exist.

Generic content is how you don’t exist. There is nothing in a swap-the-logo content for a pattern-matching machine to hold onto. No pattern that belongs to you.

Why depth gets cited and commodity content evaporates

Best proof I have is a story I’ve told so many times my clients can recite it, and I’m telling it again, sorry.

A recruiting agency came to us. And instead of “we place top talent” content, we wrote one genuinely in-depth article: a full walkthrough of how to hire a founding engineer, written for a founder doing it for the first time.

First page of Google within a couple of days.

Then ChatGPT started recommending the agency by name to founders asking hiring questions. And it was all done with just one article. I checked with the client twice because I didn’t believe it either.

Would I promise you that result? No. The channel moves monthly and anyone promising you specific AI-visibility numbers is selling something. The direction I’ll promise: depth gets cited and fluff doesn’t get you anywhere.

One technical note because people push back here: yes, officially ChatGPT’s search runs on Bing. When I test side by side, the citations still track whatever ranks on Google’s first page. Make of that what you want. My conclusion is simple and not as exciting: classical SEO still matters, don’t skip it.

What Google does to commodity content now

Not going to dramatize this part. It doesn’t need it.

In March 2024, Google shipped an update targeting mass-produced content and reported afterward that searchers would see 45% less low-quality, unoriginal content in results.

Next to it sits a written policy, scaled content abuse, which says the value problem applies “no matter how it’s created,” whether the pages came from AI, humans, or both.

So the worst case of the 40-posts-a-month plan is not “it underperforms.” The worst case is your domain (your history, your backlinks, the thing every future campaign sits on) becomes worthless, and you paid for that outcome with investor money.

The real cost of commodity content is trust

Edelman and LinkedIn run a thought-leadership study every year. Two figures from it live in my head rent-free: 73% of decision-makers say a company’s thought leadership is a more trustworthy basis for judging it than its marketing materials, and only 15% rate the quality of what they actually read as very good or excellent.

Both true at once. That gap is the entire opportunity and almost nobody takes it.

Gartner has the other one: buyers spend about 17% of the buying journey meeting with potential suppliers at all, and if they’re comparing three of you, your slice is maybe 5%.

Underneath the numbers, though, is something simpler. Commodity content costs you the same thing the “always crushing it” LinkedIn persona costs you, and it comes from the same instinct: hide the pivot, hide the quarter you lost, cut anything that sounds human, and publish the polished version instead. But the polished version is the one nobody trusts. Buyers, candidates, the AI models, all of them have worked out by now that when something reads this smoothly, there’s usually nothing real behind it.

How to create non-commodity content instead

Four ways to turn a commodity blog into something no competitor can copy, even with a bigger budget.

1. Pull your sales call transcripts

The lost deals especially. The exact words buyers use for their pain, the objection your champion couldn’t answer internally: that’s a year of content sitting in a tool you already pay for. And it’s already in the language buyers will type into ChatGPT.

2. Describe the client, not just the win

Industry, size, constraints, the mess they were in. “ROI up 30%” proves nothing. A detailed client is a pattern the model can match the next similar buyer against.

3. Get your actual experts talking

Founder, CTO, the support person who hears everything. Nobody wants to write, fine: 5 minutes a day of them answering a few sharp questions out loud is enough raw material. True authority can’t be outsourced. But you can outsource the typing part.

4. Create content together with your prospects and partners

A prospect ghosted me once after a hard pitch. Instead of follow-up email number 6, I proposed we write something together. It restarted the whole conversation, and eventually became business. We did a version of this with Advite too: we published a practical guide on comment marketing. It ranked, and it brought them customers.

How to measure whether AI is recommending you

When a board asks how you’d ever measure any of this, the honest answer is that it’s easier than they think, and almost nobody bothers. Start with what’s already sitting in your analytics: referral traffic from ChatGPT shows up as its own source in GA4, and it’s been there for a while. Almost nobody looks. The same goes for perplexity.ai and Google’s AI surfaces. That’s your floor: people who read an AI answer, clicked through, and landed on you.

Then do the part that actually tells you something. Open ChatGPT logged out, so your own history doesn’t skew the results, and type the specific questions your buyers ask.

Note whether you get named or cited, and which of your pages it pulls from. Run the same prompts again every few weeks, and watch whether your company shows up in more of the answers than it did last time.

What I won’t hand you is a precise AI-visibility score. Anyone selling you one is selling something, the channel moves monthly. But “are we in the answer, yes or no” is a question you can check yourself this afternoon, for free.

My AI confession

I use AI for content too. Grammar, typos, polish: it’s good at that, and it touched this article for exactly that and nothing else. What I won’t do is let it think. Every time I’ve broken that rule as an experiment I get back the same recognizable smoothness I’ve spent this whole article complaining about.

Am I certain about all of this? Not all of it. The AI-search stuff shifts monthly and some of what I wrote here will age badly. I just don’t know which parts. Maybe volume comes back. I doubt it. When I picture the internet in 2 years I don’t see a shortage of content. I see a shortage of anyone worth listening to.

Go read your last 3 posts. If your competitor could have shipped them with the logo swapped, you already know what I’m going to say.

FAQs

About the author

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post The True Cost of Commodity Content for a Tech Startup appeared first on BYVI.

Search your own product’s name plus the word “breach” right now. If your stomach dropped a little, hold onto that feeling, because it’s the entire lesson in this piece.

What actually happened in the OnlyFans leak

In May 2026, a threat actor using the handle Euphoric_Reply_5727 listed 340 million OnlyFans user records on a cybercrime forum for 0.313 bitcoin (about $24,000). The listing promised usernames, emails, phone numbers, follower counts, content stats, linked social profiles, and a field labeled “card” for the last 4 digits of payment cards. The headlines wrote themselves. OnlyFans, a platform sitting on some of the most sensitive user data on the internet, hacked wide open.

Except it wasn’t. The seller first claimed inside access, then admitted in messages they “didn’t breach or hack OnlyFans.” They took old leaks from X/Twitter, Instagram, and Spotify, matched those records against publicly visible OnlyFans profiles, and stapled the two together into something that looks like a breach.

OnlyFans dismissed the reports as false, and on the narrow question of “did our servers get popped,” they’re right.

Founders keep getting this part wrong. We treat “we weren’t breached” as the finish line. For the person whose data is now in that file, whether it came from a hack or a clever reassembly job makes no difference. Their information is still for sale with your logo on the folder.

Why a leak that isn’t a breach should scare you more

The OnlyFans file is a symptom of a much bigger shift, and the numbers behind it are the reason I’d rework my security roadmap this quarter.

Attackers mostly don’t break down the door anymore. They log in.

IBM’s 2025 Cost of a Data Breach report summed it up: “many attackers are logging in rather than hacking in.”

Verizon’s 2026 Data Breach Investigations Report found the human element in 62% of breaches, and stolen or abused credentials still turned up in 39% of them. The front door is your users, and they hand over the keys constantly.

The front door is your users, and they hand over the keys constantly. That’s why “human risk management” has grown into its own category: platforms like Right-Hand run continuous, AI-generated phishing and deepfake simulations against your own team, then nudge people in the moment they slip, instead of the once-a-year training everyone clicks through and forgets.

If most breaches run straight through people, the people are worth defending like infrastructure.

Those keys get stolen at industrial scale, mostly by infostealer malware. Recorded Future indexed roughly 2.95 billion compromised credentials in 2025, with the average infected device giving up 87 stored logins across every account that person had ever touched.

Worse for anyone relying on 2-factor as a safety blanket: 276 million of those credentials came bundled with active session cookies, which let an attacker walk straight past both the password and the MFA prompt.

So the raw material for a “breach” that never touches your servers is already sitting in criminal hands, cheap and searchable. Someone just has to match it to your user list.

That’s the OnlyFans playbook, and it works against any company with public profiles and reused passwords, which is to say, most of us.

What a breach really costs a startup

The cost when it lands on you is real. The global average breach fell to $4.44 million in 2025, the first decline in 5 years, but the U.S. average hit a record $10.22 million.

For a seed-stage company, that average almost doesn’t apply. The real danger is a smaller loss: the kind that drains the runway you fought for and the trust you spent years building. The good news: it’s one of the few risks here you can actually prevent.

But there’s a newer problem that hits startups specifically, because we’re the ones shipping AI features fastest. IBM found that 1 in 5 organizations reported a breach traced to shadow AI, the tools employees adopt without approval, and those breaches ran about $670,000 higher on average.

Among organizations that had an AI-related security incident, 97% lacked proper AI access controls. If your team is pasting user data into whatever model got hot this month, you’re building tomorrow’s aggregation feedstock in real time, and you won’t see it in your own logs either.

The founder playbook I’d run this quarter

Here are the 6 moves that actually change your exposure, each one drawn from a company that learned it the expensive way.

Assume every password is already leaked

23andMe didn’t get its servers breached either. In 2023, attackers ran credential stuffing, logging in with email and password combos leaked from other sites. They cracked only about 14,000 accounts directly, but a relatives-sharing feature let that spill into data on roughly 6.9 million people. The company filed for Chapter 11 bankruptcy in March 2025.

Protect the token, not just the password

MFA is necessary and it is also not a force field. When Okta’s support system was breached in October 2023, attackers used stolen session tokens lifted from support files to ride into customer environments like 1Password and Cloudflare without ever entering a password.

Turn on MFA for the boring backend accounts, where old credentials go to work

The 2024 Snowflake customer breaches hit around 165 organizations, including Ticketmaster and AT&T, and none of it required a Snowflake vulnerability. Mandiant found that 79.7% of the abused accounts had prior credential exposure, many from infostealer infections dating back to 2020, and the impacted accounts simply weren’t configured with MFA.

Collect less and expose less, so there’s nothing to reassemble

The OnlyFans file only works because so much profile data is public and so much personal data is already floating around to match it against. Every field you make publicly visible is a field an aggregator can cross-reference. And every optional data point you collect “just in case” is a row in the next database someone puts up for sale.

Watch for your users’ data in the wild, not just your logs

The OnlyFans seller had that data assembled for weeks before it hit the headlines. If your only tripwire is your own server logs, an aggregation attack is invisible to you until a journalist emails for comment.

That’s the trap: your systems look clean because they are clean, while your users’ exposure grows somewhere you’re not looking.

Recorded Future found that more than half of stolen credentials surface within a week of a device being infected, which means detection speed is the whole game.

Draft your breach statement before you need it

OnlyFans answered a global story with one word, “false,” and technically that was defensible. It also did nothing to reassure the millions of people who saw their platform’s name in a ransom listing.

What the OnlyFans breach teaches founders about protecting user data

The strange lesson of the OnlyFans story is that it separates your security effort from your fate. You can do everything right on your own infrastructure and still wake up to your brand attached to a 340 million record file built entirely from other people’s mistakes. That feels unfair because it is.

But it also points at where the leverage actually sits. You can’t stop old breaches from existing, and you can’t force your users to use a password manager. You can make reused passwords useless with MFA and breached-password screening, you can shrink the surface an aggregator has to work with, and you can have your response ready so a bad news day doesn’t become a bad news quarter.

The founders who treat “we weren’t breached” as the start of the conversation instead of the end of it are the ones who’ll keep their users’ trust when the next reassembled file drops.

Every move here is already within your reach, and the trust you protect now becomes the one thing no old leak can ever take from you.

FAQs

About the author

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post OnlyFans Leak Explained: The Breach That Never Touched OnlyFans appeared first on BYVI.

On July 30 LinkedIn started letting anyone flag your post as “AI slop” with two taps, then feed that flag straight into a model that decides who sees you next. For founders who treat LinkedIn as their main distribution channel, that little button quietly rewrote the rules of what earns reach.

Here’s what actually shipped, why it lands on us harder than on anyone else, and how to post so the button never gets pointed at you.

What LinkedIn actually shipped

Open LinkedIn, scroll for 30 seconds, and you’ll spot them by their tics. “This isn’t just a product update. It’s a masterclass in resilience.” “What a 4-hour flight delay taught me about scaling a Series A.” The “Let that sink in.” And the “Agree?” at the end of each post. And of course one of my favorites: the “Unpopular opinion:” that is, in fact, the most popular opinion on the platform.

That specific flavor of nothing now has a button attached to it. Tap the three-dots menu on any post or comment and you’ll find a “Seems like AI slop” option, as Chief Product Officer Hari Srinivasan showed in his own post announcing the change.

Hari Srinivasan, Chief Product Officer, LinkedIn Ecosystem

The button isn’t just a mute switch. Srinivasan said the flags feed LinkedIn’s classifiers directly: “We are ramping the ability for members to tell us if they believe a post or comment seems like AI slop. Slop is hard to define and the definition changes; this lets us tune our models and make better feeds.”

In plainer terms, every report is a training signal, and posts the crowd tags as slop get less algorithmic reach, especially in the suggested content LinkedIn pushes to people outside your network.

Three other moves came in the same announcement, and each one matters for how you show up.

Why the sudden urgency? Because the numbers are genuinely embarrassing.

AI detection firm Pangram scanned more than 1 million social posts and found LinkedIn was the most AI-saturated platform of all: over 40% of its long-form posts flagged as fully AI-generated, and while LinkedIn made up only a third of everything Pangram scanned, it accounted for 62% of all the AI content the firm flagged.

Why this lands on founders harder than anyone

LinkedIn is where a lot of us go to raise, hire, and sell before we have the logos to do it the easy way. For an early-stage founder, our feed is the pitch deck that runs 24/7.

Let’s start with the obvious risk. If you’ve been scaling your “thought leadership” by feeding bullet points into a model and posting whatever came out, you were already living in the 40%.

Now the readers you’re trying to win, the ones who were going to remember your name at fundraise time, can label that output slop in two taps, and the algorithm will believe them.

The channel you leaned on to look credible can now make you look generic. And unlike a paid campaign you can pause, a reputation for slop compounds quietly: the classifier remembers, and your reach decreases before you notice the flag in your dashboard.

The deeper shift is what LinkedIn is protecting. Srinivasan said the goal is to let “members get feedback from real humans on what sounds authentic.” That’s a trust economy, and trust is the one asset a pre-revenue founder actually has to trade.

And no, this isn’t a LinkedIn-only tantrum. Substack just shipped a tool with Pangram that tells readers when a newsletter was written by AI.

Pinterest added toggles to cut AI content from feeds, and Snapchat said it will stop recommending AI slop outright. Pangram itself raised $9 million to chase this problem across the web.

The LinkedIn posts I’d publish now

The slop button punishes exactly the content that was never going to convert anyone anyway. So your authentic, specific, founder-voiced posts just caught a tailwind.

Write the post only you could write.

The fastest tell of slop is that literally anyone could have posted it. Your defense is proprietary: a real churn number, a deal that face-planted and why, a metric nobody else can touch.

Look at how Tom Hunt (I’m a huge fan) writes about his company Fame: they ship around 1,000 client assets a week, the reviews that used to need people now run on AI, and a job that once meant a high-paying hire is “now a $7 API call.”

You can’t prompt that post into existence, because the numbers only live inside his business. A model can fake your tone all day. But it can’t come up with specific figures you pulled from your own Stripe account.

Kill the AI polish pass.

LinkedIn just deleted its own “enhance your post” button because the polish was the problem, so take the hint. Draft in your own words, then use AI the way LinkedIn’s replacement tool now does: to proofread, not to rewrite.

The second a model buffs your sentences into that smooth, generic filler, you’ve traded your voice (the thing that survives the new ranking) for the thing it now punishes.

If you don’t want to use LinkedIn’s proofreader, try this prompt below with any AI model. It edits like a line editor, not a ghostwriter, protects your numbers, and strips every tell up top without flattening your voice.

ROLE. You are my line editor, not my ghostwriter. Make my draft cleaner and clearer while keeping it unmistakably mine. The test: if someone who knows my writing couldn't tell I wrote it, you failed. When in doubt, do less.

VOICE PRIMING (optional, recommended). Before editing, study 2 to 3 of my past posts (pasted below) for my sentence length, rhythm, the words I reach for and the ones I never use. Match that fingerprint. Do not average me toward "good LinkedIn writing."

YOU MAY: fix grammar, spelling, punctuation, and typos; tighten only genuinely redundant or confusing sentences (cut filler words, never ideas); point out (don't fix) any claim that reads as vague or unsupported.

YOU MAY NOT: change my word choices, rhythm, or structure to sound "more professional"; add adjectives, transitions, hype, hooks, or summary lines I didn't write; smooth out my fragments, asides, or short punchy sentences; touch any number, name, quote, or date (treat them as load-bearing); use em dashes or en dashes (use commas, colons, periods, or parentheses).

BANNED PATTERNS (remove if present, never add): "This isn't just X, it's Y" contrast framing; throat-clearers like "Here's the thing" or "In today's fast-paced world"; engagement bait like "Let that sink in," "Agree?," "Unpopular opinion:"; single sentences stacked one per line for drama; corporate gloss like "leverage," "unlock," "game-changer," "masterclass in," "humbled to announce"; emoji used as punctuation.

PROCESS: read the whole draft first to learn my voice; make the lightest edit that does the job; if a change might alter my meaning or voice, leave it and flag it.

OUTPUT: (1) the edited draft; (2) "Changes," each fix in one line with why; (3) "Flags," anything vague, unsupported, or that sounds like me on autopilot rather than at my sharpest.

MY DRAFT: [paste it here]

Put your face on it and show up with proof.

With 100 million verified profiles, LinkedIn is all-in on identity. Post from your verified personal profile and bolt on proof a machine can’t hallucinate.

• The actual customer message
• The real graph
• The awkward group shot from the offsite
• The team photo where nobody’s looking at the camera
• Imperfect behind the scene moment.
  

Trade cadence for evidence.

The old growth hack was volume, and volume is precisely what the automation crackdown is built to shut down. LinkedIn is already cracking down on engagement pods and automated comments, blocking hundreds of thousands of comment attempts a day. So post 3 times a week with proof instead of daily with filler. One post carrying a number you earned for sure beats 5 that sound like everyone else on LinkedIn.

Treat comments like the product.

LinkedIn is spending real engineering effort removing bot comments because fake engagement damages trust faster than fake posts do.

So do the opposite of automation: read the actual post, reply to people by name, reference the actual thing they said, argue in good faith. Real conversation in your comments is now a trust signal the platform is actively protecting, and it costs you nothing but attention.

When in doubt, use a meme.

A model can fake a paragraph. But it can’t fake taste, timing, or a joke that actually lands. A well-chosen meme is proof there’s a human with judgment behind the account, and it’s native to how people already talk on LinkedIn.

Evan Lee does this well: he pairs a meme with a sharp point about matching your ads to your funnel instead of another 400-word “here’s what I learned” essay, and it lands harder.

You don’t need to be funny every time. You need to sound like a person, and nothing says “not a bot” like a meme you actually found funny.

The bottom line on AI slop button

The slop button reads like a threat, and for the founders who outsourced their voice to a model, it is one. But I think LinkedIn just did something useful: it turned “sounds like a real person with something to say” into the biggest distribution advantage.

The same tools that made it easy to flood the LinkedIn feed just made flooding it pointless. What’s left is something founders have plenty of and big companies can’t copy: a clear point of view, backed by numbers only you can see, in a voice that’s clearly yours.

About the author

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post LinkedIn Just Gave Every Reader a AI Slop Button to Kill Your Reach appeared first on BYVI.

Snap now blocks wholly AI-generated videos from Spotlight recommendations, and the number of people posting original content is climbing fast. For founders fighting for reach, that turns “made by a real person” into a distribution edge you can actually build a growth loop around. Here’s how to move on it before everyone else catches up.

What Snap actually changed

On July 31, 2026, Snap updated its Spotlight recommendation system so that, as the company put it, “wholly AI-generated videos will no longer be eligible for recommendation on Spotlight.” The reasoning is blunt: as low-quality, repetitive AI content floods the internet, Snap wants Spotlight to stay a place where people discover “original perspectives, personal storytelling, and the moments that people choose to create and share themselves.”

Two details matter for anyone building a company.

1. This isn’t Snap turning its back on AI.

Content that’s been enhanced or edited with Snapchat’s own AI creative tools stays eligible for recommendation, and it carries transparency indicators so viewers know what they’re looking at. The line Snap drew is between content that is wholly synthetic and content where a human is doing the creating, with AI as a tool in their hands.

2. The audience is already voting with their cameras.

Snap reports that the number of unique Spotlight contributors globally is up more than 120% compared to last year. Back in April, the company shared that daily US Spotlight creators grew more than 70% year over year in Q1 2026, and that creators who post daily grow their followers more than 10x faster than those who don’t. This is happening on a platform that reached 474 million daily active users in Q4 2025.

Put those together and you get a rare thing in distribution: a large, growing audience where the algorithm is now actively tilted toward exactly the kind of content a scrappy founder can make with a phone (if, of course, your buyers are actually on Snapchat).

One honest caveat before you commit: this audience skews young. Snapchat reaches 90% of 13- to 24-year-olds across 25+ countries, 18- to 24-year-olds are its single largest age group at 35.4%, and roughly 79% of the base is under 35.

If you sell to enterprise buyers, treat Spotlight as a brand and recruiting play rather than a demand channel. If your customers are Gen Z or younger Millennials, this is your front row.

This is a pattern, not a one-off

If Snap were alone here, I’d tell you to note it and move on. It isn’t.

YouTube got there first. In July 2025 it renamed its “repetitious content” rule to “inauthentic content” and made mass-produced, template-driven videos ineligible for monetization. YouTube was careful to say AI-assisted work still earns money, but content that’s “easily replicable at scale” with “little to no variation” doesn’t. Same line as Snap, drawn in a different place.

Pinterest went the transparency route. In April 2025 it rolled out “AI modified” labels globally and gave users a “see fewer” control for AI-generated Pins in categories like beauty and home decor. When a platform hands users a dial to turn down synthetic content, it’s telling you where demand actually sits.

TikTok built the plumbing for all of it. In 2024 it became the first video sharing platform to implement C2PA Content Credentials, the open standard that attaches provenance metadata so AI-generated uploads get labeled automatically.

That standard is backed by Adobe, Google, Microsoft, OpenAI, and Sony, which tells you provenance labeling is becoming infrastructure, not a feature.

Why this is an opening for founders

Most coverage of these changes frames them as a crackdown. I’d flip it. Every one of these moves lowers the value of the thing your best-funded competitors can buy at scale (synthetic volume) and raises the value of the thing you already have (a real person with a real story and a real product).

That’s a redistribution of reach toward the small and the specific. And the buyers are primed for it: a Snap study with MAGNA found people respond most positively to creators who prioritize “keeping it real” over polished perfection, and 49% say they’re interested in buying from authentic creator content.

Here’s where I’d focus.

Founder-led content just got a bigger multiplier.

The platforms are rewarding a face, a voice, and a point of view, which is the one asset a solo founder can produce that a content farm can’t fake at scale. Scrub Daddy turned a founder and a smiling sponge into hundreds of millions of TikTok views by leaning entirely on personality and behind-the-scenes reality. You don’t need a studio. You need to show up as yourself, consistently, on a surface that’s now weighted in your favor.

“Boring” behind-the-scenes footage is now premium inventory.

The candid, unpolished, shot-on-a-phone moment is precisely what these algorithms are being tuned to surface. Duolingo built a following in the millions on unscripted, chaotic clips of its mascot filmed around the office, not on rendered ad spots. For a startup, the build itself (packing orders, debugging live, talking to your first customers) is the content, and it reads as authentic because it is.

Speed and reaction beat production value.

Authentic content wins on timeliness, and small teams are structurally faster than big ones. Ryanair runs one of the most-watched brand accounts in short-form by posting same-day, self-deprecating reactions that a large brand’s approval chain could never ship in time. A two-person startup can react to a trend before lunch. That’s a genuine edge now that the feed favors native, of-the-moment posts.

Distribution is getting cheaper exactly where paid is getting more expensive.

As synthetic volume gets throttled, organic reach for real creators becomes more valuable, and Snap’s own numbers (daily posters growing followers 10x faster) show the compounding is real. Gymshark built a nine-figure brand largely on athlete and customer content rather than ad spend, and that playbook gets stronger every time a platform tilts toward human-made posts. If your CAC is climbing, an authenticity-weighted feed is a discount you can go claim.

@charles__johnson OG Gymshark Addict #gymshark #fyp #davidlaid #OGGymshark #gym ♬ original sound – c0nd1nh0

Provenance becomes a trust signal you can market with.

With C2PA labels spreading, “made by a human” is turning into a verifiable badge rather than a vibe. Liquid Death built a billion-dollar brand on real people and a real point of view, and in a labeled-content world that kind of provenance stops being a nice-to-have and starts being a filter customers use to decide who to trust. Being demonstrably human is becoming a moat.

Where this goes

The last decade of social taught founders that reach was something you rented from an algorithm you couldn’t influence and mostly had to pay. What’s happening now is quieter and better: the algorithms are re-pricing reach toward the one thing money can’t manufacture, which is a real person making something real. And that’s a market correction in your favor.

About the Author

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post Snapchat Spotlight Banned AI Video. Here’s the Founder Play appeared first on BYVI.

Pull up your own landing page and read the headline out loud. If the pitch is some version of “the fastest way to [do the thing],” “a beautiful, intuitive interface for your workflow,” “automate your expense reports,” “all your tools in one place,” or “powerful, no-code, and easy to learn”, stop on it for a second. What those lines have in common is that every one of them describes a task a person repeats inside your software. And doing a repeatable task inside software is exactly what an AI just learned to copy by watching.

On July 21, 2026, Anthropic shipped a feature in Claude Cowork called Record a skill.

When you stop, Claude studies the recording and turns it into a reusable skill it can run again on command. No prompt, no script, no instructions to hand-author. You show it once and it does it forever.

OpenAI shipped a near-identical feature, Record & Replay, in Codex 33 days earlier. When two rival labs converge on the same idea inside a single product cycle, it’s not just a feature anymore. It’s a whole new direction.

And this direction points straight at a whole category of software: products whose value was mostly a nice front-end, a library of templates, or a bundle of integrations wrapped around a task a person repeats.

That is the most replicable thing you can own. And copying it just went from “a rival team and a 6-month head start” to “one afternoon with the recorder running.”

This is already happening, not coming

When Anthropic shipped its Claude Cowork plugins in late January 2026, roughly $285 billion was wiped off software, financial-services and asset-management stocks in a single session, a rout Jefferies’ trading desk nicknamed the “SaaSpocalypse.” By spring, analysts were tallying the damage across enterprise software in the trillions.

So investors were doing blunt math: agents that handle work at the task level mean companies need fewer seats, and per-seat licensing is exactly how most software makes its money.

Then it got specific. 3 days before Anthropic launched Claude Design in April, its chief product officer, Mike Krieger, quietly resigned from Figma’s board.

Anthropic is running the oldest predatory playbook in Big Tech (Save this).

Here is what actually happened.

Anthropic's own Chief Product Officer, Mike Krieger, was sitting on Figma's board and he resigned on April 14, 2026.

Three days later, Anthropic launched Claude Design,… pic.twitter.com/UjBFf16nCK

— Milk Road AI (@MilkRoadAI) July 5, 2026

Claude Design went live on April 17. Figma’s stock fell about 7% the same day. If you use Claude Design, you know that the tool takes a plain-language brief and produces prototypes, slide decks, and landing pages, then hands a website design straight to Claude Code as a build-ready bundle.

For a designer whose job was execution, that’s a large chunk of the work, gone. And investors didn’t read it as a one-company story: Adobe, Wix, and GoDaddy all slid in the same window. The market repriced the entire design-to-publish stack.

To be fair to the other side of the trade: Figma still owns an estimated 80–90% of professional UI/UX design, Claude Design is a token-hungry research preview, and LLMs remain shaky the moment you edit individual visual elements.

Wedbush called the sell-off an “Armageddon scenario” that’s “far from reality,” and enterprises won’t rip out billions in existing software overnight. Anthropic itself is hedging – partnering with Salesforce, Intuit, and DocuSign rather than only replacing them. The point isn’t that software dies. It’s that the interface stopped being the thing worth defending.

Figma made the point clearest of all: it shook hands with the company supposedly eating its lunch. In February 2026, the two shipped a feature called Code to Canvas that pulls interfaces built in Claude Code straight into Figma as fully editable design layers. You can now type “Send this to Figma,” and running AI-generated code becomes something a designer can drag, tweak, and refine.

It’s a shrewd hedge: rather than pretend agentic coding isn’t happening, Figma made itself the place that work lands. But it also captures the tightrope every incumbent is walking.

As CNBC put it, the risk is that Figma is building a better on-ramp to a highway it no longer controls.

Why a UI is no longer a moat

For 15 years, a great interface was a legitimate edge. Being the tool that made a painful task feel easy could carry a company to a real business, because the alternative (a competitor rebuilding your UX and re-earning your users’ habits) was slow and expensive.

Whatever your users do over and over inside your product is precisely what these recorders capture best. If the only thing standing between a user and the outcome is your screens, an agent can increasingly deliver the outcome and skip your screens.

So the strategic question flips from “do we have the best interface?” to “do we have anything a recording can’t capture?”

That reframing is the whole game. Below is where I’d look.

What actually defends you now

Six things a screen recording can’t capture. Build for these.

1. Proprietary data that compounds.

Not “we have data”, everyone says that. Data you uniquely generate and that gets better the more your product is used: repayment outcomes for a lender, real-world delivery times for a logistics tool, conversion results tied to specific decisions.

An agent can watch your screen and copy the workflow but it cannot reconstruct a dataset built from a hundred thousand real outcomes it never saw. The moat that keeps a Bloomberg Terminal or Stripe’s Radar fraud models hard to touch.

2. Be the system of record.

Agents are getting very good at reading data and taking actions, but someone still has to be the authoritative ledger: the place data actually lives, where changes are committed with permissions, audit trails, and integrity guarantees.

If you’re a thin view on top of someone else’s source of truth, you’re the most replaceable layer in the stack. If you’re where the data is born and held, you’re infrastructure, and the agents end up routing through you.

You must own the write layer, not just the read.

3. Real network effects.

Value that grows with each user in a way no one can clone by copying your front-end: marketplaces, collaboration where the graph itself is the product, data network effects where one customer’s usage improves the product for the next.

Slack is the clearest case. Anyone can rebuild the chat interface in a weekend, but no one can rebuild the years of messages, channels, and integrations your whole company has wired into it.

4. Be the accountable party in high-stakes work.

In regulated or consequential domains (health, finance, legal, safety) someone has to be licensed, insured, auditable, and liable. An agent can draft the filing, but it can’t be the certified entity that signs it and carries the risk. If your business sits where trust and accountability are structurally required, you own a position agents can’t take by watching a screen.

It’s the moat under a Gusto, a Carta, or a Vanta, where someone has to be the accountable party of record when payroll, a cap table, or a compliance audit is wrong.

5. Own distribution and brand.

When capability commoditizes, distribution becomes the scarce asset. The company users think of first, trust with their data, and already have a relationship with wins even when the underlying task gets automated.

Founders under-invest here because it feels less defensible than tech, but right now it’s the opposite. Owning the customer relationship and the channel is one of the few edges an agent can’t reproduce from a demo. HubSpot and Canva didn’t win on features alone, they both won on being the name people already reach for.

6. Be the tool the agent calls

Stop standing in the agent’s path, become the thing it calls. Expose your product as a tool, skill, or MCP server that Claude and Codex route to for a specialized capability they can’t do well themselves.

Be the best node in an agentic workflow rather than a UI the agent bypasses. So become the Twilio of your niche, the thing agents route through, not around.

And move off per-seat pricing, which is the exact model the market is punishing. Price the outcome or the usage, so your revenue grows with value delivered rather than with headcount that agents are about to shrink.

The bottom line

Prompting isn’t dead, and neither is software. But the ground shifted, and it shifted under a specific kind of company: the one whose edge was a lovely interface for a task people repeat. If that’s you, this is the year to build something durable underneath it.

The defensible positions are the ones an agent can’t cheaply reproduce by watching: proprietary data that compounds, being the system of record, real network effects, structural accountability, distribution, and being the capability the agents depend on rather than the layer they route around. Those are the real product, and no recording can reach them.

So build that. Build the company agents route through, lean on, and depend on. The most capable tools the industry has ever shipped are now aimed squarely at the grunt work, which frees you to pour your energy into the things that were always going to decide whether you win.

The founders who spend the next few years building something a recording can’t capture are the ones this era belongs to.

ABOUT THE AUTHOR

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post Claude Now Learns Any Task by Watching and It’s Coming for Your Software appeared first on BYVI.

At Stripe Press, Klarna and Stripe took the stage to make a case that might surprise most e-commerce teams: the biggest threat to your sales in an AI-driven world isn’t a bad checkout experience – it’s never being found in the first place.

The talk, centered on agentic commerce, offered a ground-level look at an infrastructure shift that’s already in motion. AI traffic to e-commerce sites is up 393% year-on-year according to Adobe’s Q1 2026 report. But the opportunity that number represents is being quietly squandered, because the product catalogs agents are being sent to read were never built for machines.

The Invisible Catalog Problem

This was the statistic that opened the discussion:

That’s not a rounding error. It means the majority of product pages that an AI shopping agent visits today are returning incomplete data. And incomplete data means the agent can’t confidently recommend the product. From the agent’s perspective, those items might as well not exist.

To make the point concrete, the speakers walked through a case study of a major activewear brand. The brand had 160,000 products in its catalog. Of those, only 10,000 (just 6%) were actually discoverable by AI agents.

“For that reason, the AI agents are not able to confidently compare, rank and recommend this product.”

Without GTINs, agents couldn’t build complete product cards. Without product cards, they couldn’t make recommendations. The brand’s catalog was effectively invisible at the most important new layer of commerce.

The framing was direct and simple: “If you can’t be found in this layer, your product doesn’t ever have a chance of being purchased at all.”

3 Attributes That Unlock AI Discoverability

The talk distilled the discoverability problem down to 3 catalog attributes that every merchant needs to get right.

1. GTIN (Global Trade Item Number)

GTINs are the connective tissue of agentic commerce. They’re the identifier that lets an AI agent match your product listing against the same item sold by other retailers – enabling comparison, ranking, and confident recommendation.

2. Shipping Cost

Agents don’t rank on product price alone. They rank on total price. If your shipping cost isn’t part of your feed, the agent is presenting an incomplete picture to the consumer, and your product will lose out to one where the full economics are visible.

“If you’re not giving a full picture to the agent and the consumer, your product won’t be picked.”

3. Delivery Time

When a consumer tells an agent “I need these running shoes by Friday,” the agent filters its results to products where delivery by that date is confirmed. It doesn’t guess or check elsewhere. If your delivery window isn’t explicitly stated in your product feed, your item is excluded from that search entirely — no matter how good the price or how relevant the product.

All 3 attributes are prerequisites, not nice-to-haves. The agents doing the searching require structured, machine-readable data to do their job – and merchants who haven’t built for that are already falling behind.

Klarna’s Discovery Layer

Behind the scenes, Klarna has built the infrastructure designed to power agentic product discovery at scale: 100 million unique products, 400 million merchant offers, 4,300 subcategories, and one million merchant partners — all surfaced through a single API that’s normalized across merchants, LLM-optimized, and protocol-agnostic.

The pitch is straightforward: agents query the graph, get structured product data in real time, and can reliably compare and rank items across different sellers. As the speaker described it: “This graph really is what allows your products to be found.”

Rethinking How Payments Work in Agentic Flows

If the first half of the talk was about discovery, the second was about what happens after an agent finds the right product, and why the payment handoff is more complicated than it looks.

In traditional e-commerce, the buyer and the payment processor are in the same session. In agentic commerce, they’re not. The agent acts on behalf of the buyer, but the merchant still needs to process the charge. Alison from Stripe framed this as a fundamental separation:

“Payment intent is being separated from payment processing.”

To solve this, Stripe created Shared Payment Tokens (SPTs). The concept works like a sealed package: the agent creates a token scoped to a specific seller, for a specific timeframe, containing the buyer’s payment credentials. That package is handed to the merchant, who opens it, reviews it, and decides whether to process it or decline.

“You’re able to scope it to a specific seller, a specific time, and then you issue that to the seller and the seller at that time when they receive it, can open it up and they can say, I like what I see, I’m going to process as payment.”

Crucially, SPTs work with existing infrastructure — existing network tokens, device tokens, Klarna tokens. When Stripe launched SPTs, they supported cards only; Klarna support has since been added, with expansion to 100+ payment methods underway through Stripe’s growing ecosystem.

What This Means for Merchants Already on Stripe

For merchants already using Stripe and Klarna, the practical lift is close to zero. Agentic flows reuse existing Stripe components. Existing fraud configurations, Radar risk signals, and payment method settings carry over automatically. As Alison put it: “If you have enabled Klarna, it is already enabled in your agentic flows.”

Product catalogs can be fed into the discovery layer via CSV upload or import APIs, with the option to plug in existing commerce stack integrations rather than rebuilding from scratch.

In Agentic Commerce, a Failed Transaction Has No Recovery Play

One of the more striking moments in the talk was the discussion of what happens when an agentic transaction fails, and why it’s categorically different from a regular abandoned cart.

In standard e-commerce, a failed checkout is painful but recoverable. You can retarget, send a follow-up email, offer a discount. The consumer is still reachable. In an agentic flow, that path largely disappears. As the speaker explained:

“That’s very different when there’s a failed experience in an agentic flow, because that’s something that a consumer might not recover from, at least not quickly.”

Worse, a failed transaction doesn’t just cost the merchant a sale, it damages the consumer’s trust in the agent itself. “They might not trust that agent very quickly again.” This changes what agents optimize for. Rather than pure conversion, they weight for reliability in roughly this order: consumer trust, payment flexibility, post-purchase liability, and then approval certainty.

Account Linking Challenge in Agentic Commerce

The talk closed on one of the more existential concerns merchants raised: how do you maintain a direct relationship with your customer when an agent is in the middle?

Personalization, loyalty programs, purchase history, recommendations — all of it depends on knowing who the customer is. When an agent intermediates the transaction, that connection can break. As Alison noted:

“That’s probably one of the number one pieces of pushback we hear from merchants when we’re talking about agentic commerce – how do I maintain that relationship with my customer?”

Account linking is the proposed answer, connecting the consumer’s identity across agentic flows so merchants can preserve personalization and relationship data. “Solving the account linking challenge is what will allow them to really bring that rich experience and the personalization into agentic flows.”

The Takeaway

Agentic commerce isn’t on the horizon, it’s already scaling. The 393% growth in AI traffic is being driven by agents that are actively shopping, comparing, and recommending right now. The merchants who will win in this environment are the ones who’ve done the foundational work: clean GTINs, complete shipping data, accurate delivery windows, and product feeds built for machines to read.

The checkout experience still matters. But by the time a consumer reaches it, the hardest work is already done — or it isn’t. In agentic commerce, the sale is won or lost in the catalog, long before anyone clicks buy.

About the Author

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post In Agentic Commerce, the Sale Is Lost Before the Cart appeared first on BYVI.

My team and I went through a period when traction felt painfully hard. We were building. We were trying. Very few people saw us. Then a journalist came out of nowhere and wrote about us. That one story brought partnerships and clients. It also gave us the push to keep going.

That experience changed how I think about visibility. It is one of the reasons I built BYVI. I wanted to make press coverage accessible to all startups regardless of their funding stage, especially the early-stage and bootstrapped founders who usually get ignored until they raise money.

Now I see a similar opening with AI search.

Your advantage as a bootstrapped founder is your deep, practical understanding of your niche. A lot of you built the solution for the problem you experienced firsthand. And that first hand expertise is a major asset for AI search.

AI Search Changed How B2B Buyers Ask for Help

B2B buyers do not search the way they used to. They are not typing one neat keyword and hoping for the best. They explain the whole situation.

This was clear when I helped a friend who runs a recruiting agency write an article about hiring software developers. We targeted early-stage founders specifically, rather than general hiring managers, and I included a section dedicated entirely to hiring a founding engineer.

It reached the first page of Google in just a couple of days. Despite the search query being incredibly competitive, AI summaries picked it as a top source, putting us right next to YC and YouTube.

Soon after, my friend called to say a lead had come in referred directly by ChatGPT. The AI had suggested Linkus Group to someone looking for a founding engineer. This shows that when your content speaks directly to your ideal client, you can train AI to recognize your value and recommend you as the solution.

That is why old keyword thinking feels too small now. Winning content speaks to a real person in a real moment, reflecting the exact language of your ICP. It’s like sitting beside them and saying, ‘Hey, I get the mess you’re in – let me help.’

For bootstrappers, this matters even more because the upside of organic search is so big. Data-Mania reports B2B SEO can deliver around 748% ROI, while paid search sits around 36%. BrightEdge data points to 2× more revenue from organic SEO than any other channel for B2B companies. When cash is tight, that kind of return matters.

The Old SEO Playbook Is Quietly Hurting B2B Brands

A lot of founders are still getting old advice. Hire an SEO agency. Pick keywords. Publish as much as possible. Hope something ranks.

I think that playbook is breaking fast.

But the bigger problem is trust. Content is your reputation on the line. If a buyer lands on your site and reads AI slop of fluff, they will feel it right away. People can see through generic fluff. Buyers are tired of it, and AI search is getting better at filtering it out.

Google can also react badly when a site gets flooded with generic AI content. Pages can struggle to get indexed. That is a dangerous move for bootstrapped startups that cannot afford to have their domain flagged.

Velocity still matters, of course. Publishing 16 or more blog posts a month has been linked to 4.5× more leads. But quality decides whether that velocity helps you or hurts you. Better to be thoughtful and insightful than post fluff every day.

Bootstrappers Have the Raw Material and Speed AI Wants

This is why I believe bootstrappers will win.

You are usually closer to the pain than a larger company is. You are on the sales calls. You are handling objections. You are reading support messages. You are hearing the same concerns again and again. You can hear what people are saying, and if you pay attention, you can hear what they’re not saying too. Reading between those lines often comes with experience.

The Power of Industry Vocabulary

Since AI looks for depth, founders should lead with their firsthand experience and niche expertise to signal authority. The founder with 10 years of real industry knowledge has a major edge. They use the right words, what’s often called industry vocabulary. They understand the real objections. And they can convert all this firsthand industry knowledge into very impactful content.

The Bootstrapper Speed Advantage

And then there is speed.

A big company often needs rounds of approval before a piece goes live. A bootstrap founder can publish while the pain is still fresh. If you live in a work-life blend, and most founders do, your brain is already connecting the dots all day. Use that.

• After a call, write down the objection.
• After a customer question, turn the answer into a page.
• After a demo, capture the explanation that made the buyer lean in.

AI search favors that kind of freshness when it is paired with real expertise. I have seen clean, focused content get picked up by LLMs within a couple of days when the formatting is simple and the answer is easy to find.

How to Format Content for AI Discoverability

When I say formatting, I don’t mean H2s and H3s.

AI Rewards Proof, Consistency, and Human Detail

AI Search Evaluates B2B Brands Beyond Your Website

One of the biggest mistakes I see is founders thinking AI search only looks at their website.

It pulls from much more than that.

LLMs love Reddit. They crawl for community sentiment. They pull from G2, review platforms, testimonials, niche publications, product documentation, help centers, and public conversations around your category. They want proof. They want context. They want signs that real people trust you.

My founder friend acquired clients because AI search models cited his helpful Reddit comments. Those comments compared the solution to competitors in a neutral, useful way. That says a lot about where authority comes from now.

Leveraging Case Studies to Influence AI Search Results

Your case studies matter here too. AI looks for proof, and case studies are a strong signal. When you write one, describe the client in detail. Talk about their stage, their problem, and the situation they were in. That context helps AI connect your solution with a similar buyer later.

Why Inconsistent Content Hurts AI Search Visibility

Because AI synthesizes your entire digital footprint, message consistency is non-negotiable.

This is another reason I like bootstrappers in this market. Smaller teams can stay aligned more easily. There are fewer layers. There is less polishing. There is more truth.

Build AI Search Authority Through Niche Publications

In traditional SEO, boosting your website’s authority is often as simple as securing backlinks from sites with a high Domain Rating (DR), and historically, search engine algorithms haven’t cared whether you earned that link organically or paid for it

AI search models, however, are fundamentally different. They prioritize genuine, contextual trust signals from independent publications rather than just looking at link metrics. Because of this, authentic earned editorial coverage is far more effective for boosting your AI search visibility than paid press releases or sponsored links.

However, not all earned media serves the same purpose. While mentions in mainstream outlets like TechCrunch are great for attracting investors, they aren’t always the best path to customer acquisition.

Own Your Own Content Engine

If you are a B2B founder, regardless of whether you are bootstrapped, you should bring content writing in-house and own your own content engine.

You can absolutely get outside help with setting up the right process, this is where I’d come in. But the insights and knowledge source has to come from inside the company.

Pull from sales calls. Pull from closed-lost calls. Pull from support tickets. Pull from FAQ threads. Pull from onboarding friction. Pull from product docs. Pull from your co-founders and your internal experts. You get the idea.

Publish Content Where AI Crawlers Already Are

While your website still matters, because traditional Google indexing acts as a major gateway into AI search, your strategy must extend to the the platforms that AI algorithms crawl.

I’m a huge fan of LinkedIn. I love that its long-form articles often surface in AI search results to reach decision-makers directly. Surprisingly, even articles with very few likes appear in AI overviews and get cited by LLMs, which means AI doesn’t prioritize virality.

And I also see YouTube appearing frequently in AI overviews. I believe now is the ideal time for founders and their teams to start creating video content to increase brand visibility and shape their own narrative.

Final Thoughts

This is one of the best times for bootstrapped B2B founders to win attention.

Bootstrapped startups rarely have the massive budgets needed to buy growth, hire expensive sales teams, or secure headlines in major publications. But AI search has opened an entirely new lane. Today, helpful, specific, and deeply human content can put you directly in front of buyers without a massive spend.

The key is staying consistent long enough for your efforts to compound, which is exactly where most founders give up. AI models are constantly learning from what they find. Every useful page, thoughtful comment, detailed case study, and niche collaboration adds another trust signal. Over time, those signals synthesize into your brand’s truth.

So keep going my dear bootstrapped fellows. Channel your frontline experience into content that offers real depth and proof. Share the exact details your buyers actually need. Always stay close to your customers and actively control your narrative before someone else does.

If you pair your real-world experience with authentic proof, you will become impossible for AI to ignore. And once that flywheel starts spinning, the exact right clients will find you at exactly the right time.

FAQs

How can a lean bootstrapped team maintain the content velocity required for AI search?

You do not need a massive team. You need a ruthless system. Repurpose everything you create. Publishing 16 or more content pieces monthly drives roughly 4.5x more leads. Hit that volume by slicing one deep founder insight into multiple formats to survive.

Should bootstrapped startups abandon paid search entirely in favor of AI-driven B2B SEO?

Yes, especially early on. Paid search is a brutal game against VC-funded giants. B2B SEO delivers a massive 748% ROI compared to just 36% for paid search. Build your organic engine first. Buying ads will burn through your cash too quickly.

How does LinkedIn factor into an AI search strategy for early-stage B2B brands?

AI models constantly scrape social platforms for brand sentiment. LinkedIn is a goldmine for this. With an impressive visitor-to-lead conversion rate of about 2.74%, your organic posts feed the crawlers while directly converting the actual decision-makers already in your network.

Does AI search make traditional B2B email marketing obsolete for bootstrapped founders?

Not at all. AI search brings them to your doorstep, but email keeps them in your ecosystem. Email remains highly efficient, delivering around 261% ROI. Capture high-intent traffic with search, then convert these leads into email subscribers for nurturing.

About the author

Lidia Vijga

Lidia Vijga is the Editor in Chief at BYVI, founder-driven media that spotlights early-stage startups at the moment most press skips, when coverage can actually open doors. She’s also a content engineer who builds in-house growth engines and trains AI agents that help companies scale content and win brand visibility across traditional and AI search. Previously, she co-founded DeckLinks, which gave her firsthand experience building and scaling tech products.

The post AI Search for B2B Brands: Why Bootstrappers Will Win appeared first on BYVI.