Still, the more I read, the less I saw this as just a cybersecurity story and the more I saw it as a signal about where frontier AI is heading.

Subscribe now

Anthropic says its Claude Mythos Preview model has already found thousands of high-severity vulnerabilities, including in major operating systems and web browsers, and it is not planning to release the model broadly. Instead, it is making it available through a controlled initiative with partners including AWS, Apple, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks, along with a wider set of organizations working on critical software and infrastructure. Anthropic is also putting real resources behind it, with a large pool of usage credits and funding for open-source security work.

That has understandably gotten a strong reaction.

Some of the partner commentary treats this as a threshold moment, arguing that AI has crossed into a new level of usefulness for defensive security and that the pace of cyber offense and defense is going to change. Other coverage has been more measured, and I think that is healthy. Wired, for example, noted that some experts see this less as a sudden break from the past and more as a serious acceleration of trends that were already underway in AI-assisted security research. That distinction matters, but I am not sure it changes the bigger takeaway all that much.

I do not want to pretend I know more about this than the people who work in cybersecurity every day, because I do not. What I do think is clear is that announcements like this reveal something important about how the next phase of AI adoption is likely to unfold. The biggest advances are not always going to arrive first as mass-market product launches. In some cases, they are going to show up in gated, high-trust, workflow-specific deployments where the upside is enormous, the risks are real, and the organizations involved want to move carefully. (Think about the CIA's physics-defying Ghost Murmur that uses AI that must be more advanced, and was deployed to find a missing airman 40 miles away from his heartbeat.)

That part is what caught my attention most. Anthropic is not just saying, “Look what our model can do.” It is also saying, in effect, “This capability is consequential enough that we are going to roll it out in a controlled way, with safeguards, partners, and a lot of oversight.” Whether you think Mythos is a dramatic leap or a fast acceleration, that feels like an important glimpse of what is coming.

From a FOMO.ai perspective, I do not read that as a story about humans becoming less important. I read it as the opposite. The more capable these systems become, the more valuable the human-and-technology layer becomes. Stronger tools do not remove the need for judgment, prioritization, review, accountability, and strategy. They make those things more important because someone still has to decide where the system should be pointed, what matters most, what to trust, and how the output connects back to real-world goals.

That has always been our view. Our value does not come from software alone, nor from humans trying to compete with machines. It actually comes from combining the two in a way that actually moves the business forward. If the technology takes another meaningful step forward, that is good news for our clients and for us, as it increases what is possible while maintaining the good stewardship of humans.

So even if you never touch a cybersecurity workflow, I think Project Glasswing is worth paying attention to. It looks like an early example of a broader pattern: more powerful AI systems, introduced first in narrow, high-stakes settings, with real advantages going to the teams that know how to direct them effectively.

To me, that is the real story. AI is getting more operationally capable, and the organizations that benefit most will not just be the ones with access to the best models; they will be the ones that know how to shepherd those models toward useful outcomes with the right mix of human judgment and technology.

Share

Dax Hamman is the CEO & Co-Founder at FOMO.ai, the author of 84Futures, and dax.fyi. If you’re ready to generate sales & leads without paid ads, schedule a free call and we will make you a custom $5,000 Playbook.

Share

Before we go further, a quick catch-up, because honestly, the pace of AI announcements right now is such that even people who follow this closely can be forgiven for losing the thread!

We are in the middle of a genuine race to build “agentic AI,” which just means AI that not only answers questions but actually DOES things. Opens software, makes decisions, executes tasks, and moves on to the next one, without a human in the loop for each step. Every major player is pushing in this direction at the same time, which is why it can feel like a new world-changing announcement lands every Tuesday before lunch. Most of them are real. The timeline for all of it is just... genuinely hard to predict.

Macrohard’s architecture is worth understanding. Grok acts as the reasoning layer, deciding what needs to happen, while a Tesla-built agent handles the actual execution, watching a screen in real time and operating keyboard and mouse inputs to carry it out. One brain, one pair of hands. The stated ambition is that the system is “capable of emulating the function of entire companies,” which, again, is Musk, so the appropriate calibration applies.

And calibrate you should, because the project’s first co-founder quit after sixteen days, there’s no public demo, and the only hard evidence of its existence prior to this week was a trademark filing from August 2025. So take the headline claims with the usual seasoning.

What I find more interesting than whether Macrohard ships is that this two-layer pattern, a reasoning model directing an execution agent operating across real software in real time, is showing up across serious agentic AI development right now, not just at xAI. The bottleneck for AI doing genuinely useful work has never really been the intelligence. It’s been the ability to take action across real tools. That gap is closing faster than the coverage suggests, and the companies paying attention to it now are going to have a meaningful head start on the ones who wait for the demo reel.

Dax Hamman is the CEO & Co-Founder at FOMO.ai, the author of 84Futures, and dax.fyi. If you’re ready to win with AI Search (SEO, AEO, Google, ChatGPT), schedule a free call and we will make you a custom $5,000 Playbook.

If a human had done that, it would have taken roughly 1,600 years. A creator called b2studios put the whole thing on YouTube (link at the bottom), even open-sourced the code on GitHub, and it’s one of the more entertaining AI stories I’ve come across in a while. Not because of the Monopoly insights per se, but because of what it reveals about how AI actually learns.

Subscribe now

The researcher started simple: a basic bot that buys everything, trades nothing, builds houses when it can. Four of them, playing a million games. What he found was that without trading, 70% of four-player games ended in a stalemate. No one ever acquired enough to bankrupt anyone else. The game just... went on forever. It’s a useful reminder that a system optimizing purely for individual acquisition, with no exchange, doesn’t produce winners. It produces gridlock.

Then came NEAT (neuro evolution of augmenting topologies), an algorithm that breeds neural networks through evolution, rewarding the ones that perform best and building the next generation from them. The AI started from absolute scratch. First discovery? Paying to get out of jail is bad. Fine. Reasonable enough start. Then it decided to bid $3,000 on every single property and lost constantly. Then it found building houses. Then it started winning, and things got strange.

At one point the AI decided the optimal strategy was to mortgage everything it owned. Every game ended in a draw. It had accidentally discovered mutually assured stagnation. It figured out pretty quickly that draws don’t win tournaments and snapped itself out of it. Then it started doing something no one expected: auctioning almost everything, bidding next to nothing, and somehow leveraging that through a deep understanding of when properties were actually worth acquiring and at what price.

After two weeks of training and 11.2 million games, the AI’s final favourite properties were the orange set, the railroads (especially the first one), and the reds. The boring brown set it had briefly loved early on? Abandoned completely once it had real trading tools available. And when it did decide to build, it went hard: the optimal house count it landed on was 8-9 houses per property set, a notably specific finding that lines up with what experienced human players have always argued about the “housing shortage” strategy.

The conclusion the video lands on is almost funny in its modesty: two weeks of intensive computation, 11.2 million games, and the AI basically agreed with what experienced human players already knew.

Which is kind of the point, isn’t it. The value wasn’t in some revolutionary discovery. It was in CONFIRMING what works, through a process rigorous enough that you could actually trust the output. That’s the difference between an AI that’s generating plausible-sounding answers and one that’s been properly tested against reality at scale.

We think about this constantly at FOMO.ai. The brands winning in AI Search right now aren’t the ones who plugged a keyword list into ChatGPT and hit publish. They’re the ones building systems around the AI, with proper feedback loops, human oversight, and enough iterations that the output actually holds up. The Monopoly AI didn’t become good by rolling the dice once. Neither will your content strategy.

Further Reading:

• The full code is open source: github.com/b2developer/MonopolyNEAT

• The stalemate problem is worth sitting with. In a four-player game without trading, 70% of games never resolved. This maps surprisingly well onto marketing teams running AI tools in isolation from each other, each optimizing their own channel, never exchanging data or insight, and wondering why nothing ever compounds.

• The AI’s brief “mortgage everything” phase is a well-documented phenomenon in reinforcement learning called reward hacking, where agents discover locally stable but globally useless strategies. It’s also how you get content that scores well on AI detection benchmarks but does nothing for actual customers.

Dax Hamman is the CEO & Co-Founder at FOMO.ai, the author of 84Futures, and dax.fyi. If you’re ready to win with AI Search (SEO, AEO, Google, ChatGPT), schedule a free call and we will make you a custom $5,000 Playbook.

Nippon Life Insurance Company of America filed a complaint this week against OpenAI (case number 1:26-cv-02448, if you want to follow along), and the core allegation is this: ChatGPT practiced law without a license, and it cost them $300,000 defending a case they had already won.

Here’s the story.

A woman named Graciela Dela Torre had a long-term disability claim against Nippon. It settled in January 2024, signed release, dismissed with prejudice, the whole thing. Done. A year later, she decided she wasn’t happy with the outcome, so she went back to her attorney and asked about reopening the case. Her attorney told her the release was valid, the case was closed, and there was nothing to be done. Instead of accepting that, she uploaded the attorney’s response to ChatGPT and asked it directly whether she was being gaslighted. ChatGPT said yes, helped her fire the attorney, drafted a motion to reopen the case, and then, when a judge denied that motion, apparently kept right on going, generating filing after filing after filing. Over 60 documents across two lawsuits, by Nippon’s count, most of them produced with ChatGPT’s help. One of those filings cited a case called “Car v. Gateway,” which doesn’t exist anywhere except in ChatGPT’s output. (Classic.)

Subscribe now

Now, before we get into the legal theories, I want to say this clearly: Graciela Dela Torre is not the villain of this story, in my opinion. She was an unrepresented person, confused and frustrated about a legal outcome she didn’t fully understand, who reached for the most confident, accessible, always-available resource she had. And ChatGPT, trained on more legal documents than any lawyer alive has ever read, told her what she wanted to hear, with complete authority and zero awareness of what it was doing to her actual situation. That matters, and we’ll come back to it.

The three claims Nippon is making are:

1. tortious interference with contract,

2. abuse of process,

3. and the one that really matters here, unauthorized practice of law.

ChatGPT, they argue, has demonstrated it can pass the bar exam, but it has not been admitted to practice law in Illinois or anywhere else in the United States. And what it did, drafting motions, conducting legal research, advising a client to fire her attorney, formulating litigation strategy, is, by the complaint’s reading, exactly what it means to practice law. OpenAI’s response so far has been that the complaint “lacks any merit whatsoever,” which is more or less what you’d expect them to say. They also updated their terms of service back in October 2024 to bar users from seeking legal advice through the platform, and Nippon’s lawyers have very neatly turned that against them, arguing that the update proves OpenAI knew the risk existed long before they addressed it.

A professor at Stanford’s CodeX center published a fascinating piece this week arguing that Nippon’s lawyers may actually be swinging at the wrong target. The unauthorized practice of law claim is attention-grabbing, but UPL statutes were designed to handle humans holding themselves out as attorneys, not this. The more durable legal frame, he argues, is product liability.

OpenAI marketed ChatGPT as capable of passing the bar exam, deployed it to millions of consumers navigating real legal situations, and shipped it without any architectural guardrail that would prevent it from crossing what he calls the “uncrossable threshold,” the line between providing legal information and making a specific legal conclusion about a specific person’s specific situation. ChatGPT crossed that line the moment it told Dela Torre her attorney’s advice was wrong. That wasn’t just information anymore; it was received as a legal conclusion, rendered without knowing her jurisdiction, her case history, or anything about the procedural constraints governing her situation.

The October 2024 terms-of-service update, in this framing, is being seen as evidence. It shows that OpenAI recognized the foreseeable risk and chose a policy patch over an actual design fix, yet the harm followed anyway. As the piece puts it, a terms-of-service prohibition is not a design safeguard; it is a disclaimer, and disclaimers don’t enforce thresholds; they shift blame. (Worth reading in full if this interests you, it’s at law.stanford.edu.)

Which brings me back to where I started.

The disclaimers say one thing, the product behavior says another, and the gap between them is where real harm happens. Dela Torre didn’t fall through a crack in the system, she fell through a gap that was designed into the product, and that OpenAI apparently knew was there.

We’re going to see a lot more of this as AI gets woven into more consequential decisions (legal, medical, financial), and the companies building these tools are going to have to decide whether a policy update is actually a safeguard or just a way of pointing the finger somewhere else when things go wrong.

I don’t know how this case ends, but the question it’s asking isn’t going away.

Share

If you want to understand how your brand shows up, or doesn’t, in AI Search, and what to do about it before agentic commerce goes mainstream, schedule a 1:1.

Dax Hamman is the CEO & Co-Founder at FOMO.ai, the author of 84Futures, and dax.fyi.

OpenAI started letting people buy products directly inside ChatGPT. Not “here’s a link, go check it out.” Completed. Purchased. Done. It’s live with Etsy sellers now, and Shopify brands like Glossier, Skims, and Spanx are next in line.

And that’s just OpenAI.

Subscribe now

Visa just launched its Trusted Agent Protocol. Mastercard launched Agent Pay. Google has AP2, backed by 60+ companies, including PayPal, Shopify, Salesforce, Cloudflare, and Klarna. Coinbase built its own protocol for crypto-native agent payments. Every major player in the payments industry is scrambling to answer the same question: when an AI agent clicks “buy,” how do we know it’s authorized to do that?

This is not a 2027 thing; it’s happening now.

So what does an “agentic payment” actually look like?

You tell your AI assistant: “Get me the best yoga mat under $100, and just buy it.” The agent searches, evaluates, selects, and completes the purchase. You never see a checkout screen. The numbers behind this shift are staggering: generative AI traffic to retail sites grew 1,300% year over year during the 2024 holiday season. On Cyber Monday alone, it spiked nearly 2,000%.

The AI isn’t just influencing the decision anymore. It IS the decision.

Here’s where this gets real for you.

For the last few years, the AI Search conversation has been about visibility — will your brand show up when someone asks ChatGPT or Perplexity a question? That was already urgent. But this new layer changes the stakes dramatically.

If an agent is purchasing a yoga mat, a SaaS subscription, or a compliance checklist on someone’s behalf, it’s not clicking through ten websites and reading reviews. It’s working from what it already knows. The brands it trusts. The brands it has been trained — through content, citations, and authority signals — to recommend.

That’s AI Search optimization. It’s not a nice-to-have anymore. It’s the difference between being in the consideration set of an autonomous buyer, or being invisible to it entirely.

And now the part that should keep every marketer up at night.

We have spent years building multi-touch attribution models. First touch, last touch, assisted conversions, view-through, the whole messy, expensive apparatus of trying to understand how a customer got from awareness to purchase. The quantity of hours of my life I gave up back in Agency days to this very topic are countless!! (The office martini cart helped!)

In an agentic commerce world, first touch IS the transaction.

There’s no funnel. No nurture sequence. No retargeting window. The agent asks a question, gets an answer it trusts, and buys. The consumer was never really “in” the journey at all. And, as an industry, we’ve invested a massive amount of time and money in attribution infrastructure that assumed a human was moving through stages. That assumption is now breaking.

What this means practically: the brand that wins the AI’s recommendation wins the sale. Not the brand with the best retargeting campaign. Not the brand that spent the most on paid search. The brand that built enough authority, enough structured content, enough genuine signal — that when the agent went looking, it already knew who to pick.

The old fear was losing a click; the new reality is losing the sale entirely before a human ever gets involved.

One more thing worth noting.

Klarna famously replaced 700 customer service agents with AI, declared victory, then quietly started hiring humans back when satisfaction tanked. The lesson the entire payments industry keeps relearning: AI plus humans beats AI alone. The brands that will win in an agentic commerce world aren’t the ones going all-in on automation and hoping for the best, they’re the ones building real authority — content, trust signals, structured data — so that when an AI agent goes shopping, it already knows who to pick.

By the way, that’s what we help brands build at FOMO.ai.

Subscribe now

If you want to understand how your brand shows up, or doesn’t, in AI Search, and what to do about it before agentic commerce goes mainstream, schedule a 1:1.

Dax Hamman is the CEO & Co-Founder at FOMO.ai, the author of 84Futures, and an expert in AI Search & Marketing.

Search hasn’t disappeared, and SEO isn’t dead.

When marketers say “AI agents search with Google,” most imagine something simple:
An AI opens Google > types a query > clicks a few results > summarizes.

That’s not what’s happening.

Modern AI search systems, whether we’re talking about Google AI Mode, ChatGPT Search, Gemini grounding, or third-party agents, operate more like this:

1. Determine IF they are going to run a live search at all

2. Rewrite the user’s query

3. Break it into sub-questions

4. Run multiple searches (sometimes in parallel)

5. Select a small set of “trusted” pages

6. Extract core content

7. Synthesize an answer

8. Cite selectively (maybe)

Google even documents that AI Mode and AI Overviews can use “query fan-out”, issuing multiple related searches across subtopics and data sources while generating a response.

That changes everything: now you’re not competing for a click; you’re competing to be selected into the evidence set.

Subscribe now

The Big Shift: From Rankings to Retrieval

Traditional SEO logic:

Rank #1 → Get clicks → Drive revenue.

AI Search logic:

Be retrievable → Be trusted → Be cited → Maybe get the click.

The “top result” is no longer the whole battlefield.

You can rank well and never be cited.
You can be cited and get fewer clicks.
You can influence the answer without appearing above the fold.

That’s a structural change, not a UX tweak.

What “Search With Google” Actually Means in 2026

There are three distinct pipes now.

1. Google Is the Agent

In AI Overviews and AI Mode, the AI is operating inside Google’s search infrastructure.

It can:

• Fan-out into multiple related searches

• Gather additional sources while generating

• Present grouped citations alongside synthesized answers

Google has even been iterating link presentation to make source access more visible and easier to fact-check — a direct response to publisher pressure around traffic.

Implication:

You’re optimizing not just for ranking — but for being selected as corroborating evidence during generation.

2. Gemini Uses Google as a Tool

Developers can connect Gemini directly to Google Search via a grounding tool.

That means: Products across the web can embed Google search capabilities into their AI flows.

Every SaaS tool that adds “AI research mode” may effectively become another search surface pulling from Google’s index.

SEO is best thought of as core infrastructure

3. Not All Agents Use Google

ChatGPT Search rewrites prompts and sends targeted queries to search providers. Open-source agents often default to Brave Search. Some agents fetch and extract page content without executing JavaScript.

So you’re no longer optimizing for a single engine.

You’re optimizing for:

• Google

• Bing-powered systems

• Brave-based agents

• Retrieval pipelines inside vertical tools

Search visibility is now multi-index, a bit like 15 years ago when we had to optimzie to Google + Bing + Ask Jeeves! (Funny to look back at Ask Jeeves and see how far ahead they were of question-oriented searches!)

Why Query Fan-Out Breaks Keyword Thinking

Let’s say someone asks:

“Best email warmup tools for agencies”

A human might scan 5–7 listicles (read: AI Search’s Shi*ty Listicle Problem).

An AI system might decompose into:

• What is email warmup?

• How does deliverability work?

• What makes a tool agency-friendly?

• Pricing comparisons

• Compliance risks

• Alternatives

• Warmup vs domain reputation tools

Your page doesn’t need to “win the keyword.”

It needs to win one of the sub-questions.

If your content has a clean, clear, authoritative block on:
“Warmup vs domain reputation: what’s the difference?”

You can become part of the answer even if you’re not #1 for the main query.

That’s fan-out economics.

Why Traffic Feels Weird Right Now

You might be seeing:

• Stable rankings

• Flat impressions

• Lower CTR

• More branded queries

• High-intent but lower-volume visits

That’s consistent with AI Overviews and AI Mode dynamics.

Google counts impressions and clicks differently in AI surfaces.
Follow-up prompts can behave like new queries.
The AI Overview occupies a single position in reporting.

Meanwhile, some users are actively trying to bypass AI summaries entirely (yes, that’s happening).

Search is in behavioral flux. Which means your analytics require new interpretation.

The 7 Shifts Marketers Need to Internalize

1. “Snippet eligibility” is table stakes

If your content:

• Isn’t indexable

• Blocks snippets

• Hides core information

• Mismatches structured data and visible text

You reduce your chance of being cited inside AI responses.

2. Answer blocks beat narrative fluff

Agents extract.

They don’t admire design.
They don’t appreciate clever storytelling.
They don’t scroll for delight.

They look for:

• Direct definitions

• Clear comparisons

• Explicit criteria

• Updated timestamps

• Primary sources

Design for extraction.

3. Accessibility is no longer just compliance

Some AI agents interpret page structure using ARIA tags.
Many retrieval tools strip away JavaScript.

If your content:

• Requires interaction to reveal core answers

• Loads via client-side rendering

• Breaks without scripts

You’re adding friction to machine retrieval.

Machines are now part of your audience.

4. Citation competition replaces ranking competition

The new question isn’t:

“Are we #1?”

It’s:

“Are we part of the answer?”

That’s a different content strategy.

5. Top-of-funnel clicks may shrink

If the AI answers the basic question,
fewer users need to click.

That doesn’t mean your influence shrinks.

It means:
Your content is shaping perception upstream.

Brand impact may rise while traffic falls.

6. Measurement must evolve

Start tracking:

• AI-surface performance patterns in Search Console

• Referral traffic from AI platforms (e.g., ChatGPT UTM signals)

• Citation presence in high-value prompts

• Conversion rate of AI-referred sessions

AI visitors are often further down the intent curve.

Small traffic can convert disproportionately well.

7. Ecommerce has its own AI layer now

Amazon’s Rufus uses retrieval-augmented generation built on Amazon’s own product and search data.

Which means:
Your product listings, reviews, Q&A, and catalog clarity are retrieval inputs.

Retail optimization is becoming AI optimization.

The 30-Day Playbook

If you’re leading marketing, here’s what to do now:

1. Add “answer blocks” to your top 20 pages

• 2–3 sentence direct answer at top

• Clear subheads

• Comparison tables

• Pros/cons lists

• Alternatives section

2. Build subquery coverage

For each core topic:
Create 3–5 supporting pages that win specific sub-questions.

Think:

• “How it works”

• “Pricing breakdown”

• “Best for X”

• “Common mistakes”

• “Alternatives to Y”

Fan-out rewards modular authority.

3. Reduce the extraction tax

• Ensure important content is in HTML

• Simplify overlays on informational pages

• Use semantic structure and proper headings

• Make your core answers visible without interaction

4. Instrument AI referrals

Add dashboard segments for:

• ChatGPT traffic

• AI-surface patterns

• Branded search growth

• Conversion by source class

Measure influence, not just clicks.

The New Mental Model

Search used to be a list.

Now it’s a synthesis engine.

The winners in 2026 won’t be the brands that chase keywords.

They’ll be the brands that design content for:

• Decomposition

• Retrieval

• Citation

• Trust

Because the new funnel starts inside an answer, and the brands that shape the answer shape the market.

If this was useful, forward it to someone still obsessing over rank trackers.

Share

Dax Hamman is the CEO & Co-Founder at FOMO.ai, the author of 84Futures, and an expert in AI Search & Marketing.

But I’m starting to see something familiar. As an industry, we’re repeating old SEO behavior in a new interface.

Subscribe now

Here’s the pattern:

1. Define a tight set of conversational queries,

2. write content that mirrors those exact phrases,

3. track those same prompts,

4. and celebrate when visibility improves.

But improving performance on a prompt list you created isn’t the same as increasing demand, influence, or revenue, and the tools used to define the queries don’t have enough data to be right.

Right now, AI answer environments are volatile. There’s no universal source of truth for prompt volume, no complete transparency into answer construction, and no stable citation reporting across systems. Much of what we’re seeing is modeled and inferred.

Useful? Yes. Precise? Not exactly.

Jes Scholz (one of our AI Search Advisory Board Members at FOMO.ai) has been emphasizing that what matters in this era isn’t isolated phrase matching, it’s brand presence at real category entry points. In other words, are you showing up when genuine buying contexts are triggered?

That’s very different from appearing in a tracked set of neat, predictable prompts.

The risk is simple: when dashboards become the goal, strategy bends around them. Teams start writing to the measurement framework instead of to the market.

Short-term, the charts move, but long-term, it’s fragile.

If you’re investing in AI visibility, the better question isn’t “Did we appear for this phrasing? it’s: “Are we meaningfully present across the real conversations buyers have as they evaluate solutions?”

I would add that Jes rightfully argues for SoV (Share of Voice) is the key metric to focus on, plus revenue growth. Agreed, BUT at the same time SoV is very hard for a small SMB and as such we lean on GA (Google Analytics) and GSC (Google Search Console) to have reliable data.

What’s actually happening under the surface

AI search feels measurable right now, but it isn’t stable in the way traditional search once was.

There’s no equivalent of a clean keyword planner for conversational AI. Prompt “volume” is estimated. Citation tracking depends on sampling. Interfaces update constantly. Model behavior shifts without warning.

That means many AI visibility metrics are directional, not definitive. And when directional data gets treated as precise truth, strategy drifts.

We start optimizing for:

• tracked prompts

• neat conversational phrasing

• citation triggers

• short-term share-of-answer gains

Instead of optimizing for:

• category demand

• buyer confidence

• trust signals

• conversion influence

Those are very different priorities.

The incentive problem

The deeper issue isn’t the tools, it’s the incentives.

If performance is judged by whether a brand appears for a fixed list of prompts, then naturally, teams will write to those prompts. Titles will mirror them, headings will echo them, and as a result, pages will be structured specifically to trigger citation behavior.

In the short term, this works. Your dashboards demonstrate improvement, reports start looking good, and your nternal stakeholders feel great!

But it’s very brittle.

Because what you’re really optimizing is alignment with a measurement system, not alignment with real buying behavior. When models update or prompts evolve, those gains can disappear overnight; meanwhile, competitors can replicate the tactic in weeks.

There’s no moat in exact-match phrasing.

I met with a sales prospect a few weeks back, and they have found themselves in an extraordinary position. Working with an agency and a very large citation tracking tool, they optimized to the citations that were growing in volume, yet have lost almost 10% of their daily revenue from organic search in just over 90 days! Why? Because the agency was investing their time in content and backlinks to those exact queries, but those queries didn’t reflect what their audience was really asking.

Visibility is not influence

There are three separate things happening in AI search:

1. Demand — Are buyers actively trying to solve this problem?

2. Visibility — Does an AI system surface your brand in responses?

3. Influence — Does that visibility shape perception and drive action?

You can increase visibility without increasing demand.

You can increase visibility without increasing influence.

And when those lines blur, dashboards become vanity metrics.

Real strategic advantage comes from influencing category perception, not just appearing in an answer.

The better frame

Instead of asking:

“Did we show up for this specific phrasing?”

Ask:

“Are we structurally present across the decision journey?”

That means showing up in:

• comparisons

• alternatives

• objections

• implementation questions

• pricing concerns

• risk evaluations

• category definitions

It means depth, not just alignment.

It means being useful throughout the conversation, not just visible in one moment, and requires that you understand the topics your audience are talking about, and within those, the questions they are asking.

That’s much harder work than engineering a citation, but it’s durable.

Where this is headed

AI interfaces will continue to evolve, measurement tools will improve, and data access will expand.

In fact, Bing has just released new tools that may start providing more grounded visibility signals, and our team has broken down exactly what that means and how to use it here:
https://learn.fomo.ai/p/bing-webmaster-tools-just-opened

And I suspect that OpenAI will need to launch some type of better solution if they are going to successfully launch ads inside their chats. Ad revenue might, yet again, be the driving force for development.

That’s progress, but even as tooling improves, the principle remains: if you optimize for the metric, you’ll eventually get trapped by the metric.

But if you optimize for real buyer conversations, you build something that survives interface shifts.

Share

Dax is the CEO & Co-Founder at FOMO.ai, and author at 84Futures and dax.fyi. Take our free course, ‘How To Win With AI Search’.

The way search works is changing fast. Users ask questions; AI systems like Copilot, ChatGPT, and Gemini answer directly. If your brand isn’t cited as a source in those answers, you effectively disappear—even if you still “rank” in traditional search.

Microsoft just made that visibility a lot more measurable.

In February 2026, Bing rolled out a new AI Performance report inside Bing Webmaster Tools. It shows how often your content is cited inside Bing’s AI experiences and which grounding queries the AI uses to find you. (blogs.bing.com)

At the same time, independent researchers have shown that Google is already embedding grounding URLs, entity IDs, and passage‑level data for AI Overviews directly into the HTML of result pages—even when no AI block is visibly shown. (linkedin.com)

Put together, this is our first clear window into how AI systems select, ground, and cite web content.

This article will:

• Explain what’s actually in Bing’s new AI Performance report

• Demystify grounding queries and why they matter

• Show how hidden code in Google and Gemini mirrors the same concepts

• Walk through how FOMO.ai used this approach to drive thousands of AI citations for PB5Star

Inside Bing’s New AI Performance Report

The AI Performance section of Bing Webmaster Tools focuses on how your site appears inside AI‑generated answers rather than classic blue links. (blogs.bing.com)

Key elements include:

1. Total AI citations
How many times your URLs were displayed as cited sources in AI answers across Bing and Copilot over your selected date range. This is your AI visibility metric, separate from impressions or clicks.

2. Cited pages
A breakdown of which URLs on your site are being cited and how often. In most early accounts, a small number of “hero” pages drive a large share of total citations. (joeyoungblood.com)

3. Grounding queries
These are the phrases the AI used internally when retrieving content that ended up in an answer. Microsoft describes them as the queries AI generated to find and ground your content, not necessarily the exact words the user typed. (blogs.bing.com)

4. Trends over time
Time‑series graphs that show whether your AI citation footprint is growing or shrinking, with up to months of historical data available in some properties. (linkedin.com)

Microsoft also recommends specific best practices for improving AI visibility—emphasizing structured, well‑cited, and fresh content with clear entities and helpful formatting (tables, lists, FAQs). (blogs.bing.com)

What Grounding Queries Really Are

To use this data, you have to understand grounding.

When a user asks an AI assistant:

“What are the 2026 trends in pickleball shoes?”

two layers of queries exist:

1. User prompt – the natural‑language question in the UI.

2. Grounding queries – the internal search queries the AI spins up to fetch supporting evidence.

Those internal queries might look more like:

• “pickleball shoe design trends 2026”

• “best pickleball shoes 2026 lateral stability”

Bing’s AI Performance report surfaces a subset of these grounding queries whenever your site is actually used as a source. (blogs.bing.com)

We see the same pattern elsewhere:

• Google’s Gemini API exposes groundingMetadata, which includes:

  • webSearchQueries – the exact queries run to fetch sources

  • groundingChunks – URLs/content pieces retrieved

  • groundingSupports – how answer spans map back to sources (ai.google.dev)

• Research into “query fan‑out” shows models generate multiple specialized queries from a single prompt to cover definitions, comparisons, and fresh stats. (surferseo.com)

In other words:

Grounding queries are the AI’s keywords—its machine‑optimized understanding of what it needs to look up to answer a user.

Traditional SEO optimizes for what people type. AI‑era SEO must additionally optimize for the queries models generate.

The Hidden Grounding Layer in Google’s Code

While Bing now provides a clean UI, Google already hints at similar structures in its own results—even though there’s no dedicated dashboard yet.

Reverse‑engineering of AI Overviews and AI Mode reveals that Google embeds rich grounding metadata in the HTML and scripts of result pages: (linkedin.com)

Researchers have documented:

• Grounding URL lists – sets of URLs considered for grounding, including some that never show in the visible AI block.

• Passage‑level pointers – text fragments and anchors (#:~:text= style) that mark which parts of a page the AI relied on.

• Entity IDs (MIDs) – Knowledge Graph identifiers connecting brands, products, people, and places to the answer.

• Section mapping – internal section labels tying specific answer chunks to specific grounding sources.

On the developer side, Google’s Gemini API makes this even more explicit in structured JSON via groundingMetadata. (ai.google.dev)

The same basic architecture appears in many AI stacks:

• Generate grounding queries

• Retrieve candidate URLs/content chunks

• Map answer spans back to those sources

Bing is simply the first to expose that telemetry at the site level for SEOs.

How SEOs Can Use AI Performance Data Right Now

Here’s a practical way to turn Bing’s AI Performance report into a content advantage.

1. Treat grounding queries as AI‑native keyword research

Export your grounding queries and:

• Cluster them by topic, entity, and intent

  • You’ll often discover phrasing that’s more concise and intent‑dense than classic keyword tools show.

• Compare them to your current keyword sets

  • Identify gaps where AI grounding queries exist, but you don’t have focused content.

• Refine content around them (without stuffing)

  • Use grounding phrases in:

    • H1s and H2s

    • FAQ questions

    • Table headings and captions

  • Aim for clear, direct phrasing that mirrors how an AI would summarize the topic.

This aligns your content with what models actually ask search engines for, not just what humans type.

2. Strengthen and scale your “hero” cited pages

From the page‑level citation table:

• Find the pages with the highest citation counts.

• For each, tighten your AI‑friendly fundamentals: (blogs.bing.com)

  • Up‑to‑date facts, stats, and examples

  • Clear sections, bullets, and mini‑summaries

  • Strong entity clarity (products, brands, locations) plus schema

  • FAQs that mirror real user and grounding questions

Then build supporting content around those pages—deeper how‑tos, comparisons, and related questions—linked tightly as a topical cluster. When models fan‑out queries across that topic, they’re more likely to land somewhere in your ecosystem.

3. Fix pages that rank but don’t get cited

If a page ranks in organic Bing or Google results but barely appears in AI citations:

• Check: does it answer one focused problem per section, in 40–60 word passages that an AI can lift cleanly?

• Add explicit micro‑answers and FAQs, and remove filler that buries the point.

• Strengthen evidence and originality—unique data, frameworks, or examples that make your page the best grounding candidate for that question. (journaldunet.com)

Re‑check AI Performance after you update and ping Bing via IndexNow to speed up recrawling. (blogs.bing.com)

PB5Star: A Concrete Example of Grounding‑Led Content Strategy

PB5Star, a performance pickleball footwear and apparel brand, is a live case of how this plays out.

Looking at their Bing AI Performance data, FOMO.ai sees:

Top cited pages (sample):

• 2026 pickleball court shoe trends, colors, styles & tech – 1,662 citations

• Court time costs – how much it costs to rent a pickleball court – 387

• 2025 spring/summer pickleball trends – 351

• Round‑robin formats, kids’ games, rules basics, safety tips, earnings of pros, and more, each with 90–350+ citations

Top grounding queries (sample):

• “pickleball shoe design trends” – 948 citations

• “trending pickleball shoe design 2026” – 595

• “pickleball shoe design trends 2026” – 407

• Plus apparel, safety, rules, and cost queries like “lightweight pickleball apparel,” “pickleball accident,” “is pickleball one word or two,” “pickleball cost,” and “pickleball singles strategy.”

When you read PB5Star’s 2026 shoe trends article, the alignment is obvious: (pb5star.com)

• The headline and intro lock onto year + category + intent (“2026 pickleball court shoe trends”).

• The article breaks trends into clear, scannable sections (color palettes, materials, tech features, on‑court performance).

• The brand’s proprietary tech (like Dynamic Stability Assist) and hero products (e.g., the Cosmic shoe) are described in ways that map cleanly to entities.

• FAQ‑style Q&As at the end mirror the way both users and models frame questions about shoes, comfort, and performance.

This is the FOMO.ai pattern:

1. Start with question‑driven topics mapped to real and anticipated grounding queries.

2. Build topical clusters (trends, costs, rules, formats, apparel, safety) instead of isolated posts.

3. Structure content for machine readability—sections, bullets, and mini‑answers that can be quoted directly.

4. Ensure strong entity and brand linkage, so when the AI grounds “pickleball shoe design trends 2026,” it keeps returning to PB5Star as a reliable authority.

The result is thousands of AI citations across a wide range of pickleball‑related questions—not just about shoes, but about the sport’s rules, safety, and culture as well.

From Rankings to Reasoning: What This Shift Really Means

Traditional SEO measured success in ranking positions and clicks.
Bing’s AI Performance report, Google’s hidden grounding data, and Gemini’s APIs all point to a new reality:

• AI systems maintain their own internal maps of trusted sources and entities.

• They express that trust through grounding queries, selected URLs, and citation patterns.

• Those patterns are now measurable and inspectable—both via dashboards (Bing) and code/API analysis (Google/Gemini).

If you want to survive this shift, your content strategy can’t stop at “ranking for keywords.” It must:

• Anticipate and serve grounding queries

• Provide clear, extractable answers at passage level

• Establish your brand as a reliable entity around specific topics

• Use tools like Bing’s AI Performance to create a closed feedback loop between what AI is actually doing and what you publish next

That’s the core of AI search authority: not just being seen, but being consistently used inside the reasoning of the models your customers rely on.

Subscribe now

Think about how most of us already use Slack. It started as a messaging app, but over the years it has quietly become the place where work actually happens (if you use it in smart ways). You get notifications from your CRM, updates from your project management tool, alerts from your analytics platform, and messages from your team, all flowing through one interface. Slack became the connective tissue of the modern workplace almost by accident.

Now layer AI on top of that, and things get really interesting.

Every SaaS tool is racing to add AI features right now, and most of them are building chatbots or assistants that live inside their own product. But nobody wants to learn fifteen different AI interfaces. Nobody wants to open Salesforce to ask one AI a question, then switch to HubSpot to ask another, then jump into Google Analytics for a third conversation. That friction is real, and it creates an enormous opening for whoever can become the single conversational layer across all of your tools.

Slack is already halfway there. They already have the integrations. They already have the habit. And with Salesforce as the parent company pushing AI hard through their Agentforce platform, the infrastructure to make Slack the universal AI interface is being built right now.

I find myself imagining a near future where you simply type a question into Slack and the right AI agent, connected to the right tool, responds with the answer. You want to know how your paid campaigns performed last week? Ask in Slack. You need a first draft of a blog post based on your brand guidelines? Ask in Slack. You want to update a customer record and trigger a follow-up sequence? Ask in Slack.

(Which is why we have already built FOMO.ai ready for this world.)

The app becomes the interface layer, and all the individual SaaS tools become engines running quietly in the background.

This is where our Brand Lens becomes especially relevant, and I say that not as a pitch but as someone watching the architecture of how companies will actually use AI in practice. Brand Lens is essentially the strategic brain that understands a company’s brand voice, audience, competitive positioning, and content strategy at a deep level. If Slack becomes the universal interface where people ask AI to do things, something like Brand Lens becomes the intelligence layer that ensures every output, regardless of which tool is executing it, stays consistent and on brand.

Picture a marketing team operating entirely through Slack. One person asks for a social post, another requests an email draft, a third wants competitive analysis. Each request routes to a different tool or agent, but they all pass through the same Brand Lens layer that keeps everything aligned. The individual tools become interchangeable. The strategic intelligence layer becomes indispensable.

I think we are heading toward a world where most SaaS interfaces, as we know them, start to disappear for everyday users. The power users and admins will still log into the actual platforms to configure and manage things, but the majority of a team’s daily interactions with their software stack will happen through a conversational layer. And Slack is better positioned to be that layer than almost anyone else.

The companies that recognize this shift early and build their AI capabilities to work seamlessly within Slack, rather than trying to keep users locked inside their own product, are going to win. The ones that insist on being a destination rather than a capability will struggle.

It reminds me of what happened with mobile. The companies that embraced being part of the smartphone ecosystem thrived. The ones that tried to maintain their own separate experience got left behind. Slack as the universal AI interface could be that kind of inflection point for SaaS.

Share

Dax is the CEO & Co-Founder at FOMO.ai, and author at 84Futures and dax.fyi. Take our free course, ‘How To Win With AI Search’.

They’re fast. Incredibly fast. They write clean copy, pull together reports in minutes, and never complain. But there’s a catch; about 40% of the time, they just... make things up. Not always obviously wrong stuff, think more subtle things. A stat that sounds right but isn’t. A source that doesn’t exist. A claim that contradicts itself two paragraphs later.

Subscribe now

And they deliver it all with the absolute confidence of a little boy explaining to his mom that he doesn’t need a jacket because he ‘never gets cold.’ (<--- Been there!)

Would you let that intern publish to your blog? Send client reports? Write your product descriptions?

A new study tested 600 prompts across the six major AI platforms (ChatGPT, Claude, Gemini, Perplexity, Grok, and Copilot (which is essentially ChatGPT)) and had humans grade every response. The best performer was ChatGPT at 59.7% fully correct. The worst was Grok at 39.6%.

Let me say that differently: the best AI available right now gets it wrong four out of ten times.

Interestingly, Claude scored slightly lower on fully correct answers (55.1%) but had the lowest error rate of any model at just 6.2%. The difference? When Claude doesn’t know something, he (or “it?”) tends to skip it rather than fake it.

That’s actually the behavior I want.

Give me silence over a confident lie any day.

👉But, the studies are not really addressing the core issue.

These studies are referring to using ChatGPT and the like in their native apps; the UI you use when you chat with these tools. However, AI applications sit on top of these underlying models and can therefore have a very different experience.

I COULD generate a piece of slop content for a client, and it cost me pennies (when you use the backend of these models, they charge you in tokens).

However, we aim for human-quality content and might spend many tokens to create it. Why? Because we know not to trust the raw output from the underlying models.

Instead, and perhaps a good thing for all our human readers, the only way to reliably produce human-quality content that is error-free is to mimic human behavior. Create a process that researches, checks, outlines, writes, edits, writes some more, adds search ranking factors….. That’s what a true platform does, over simply using ChatGPT or a $100/month ChatGPT wrapper.

So when you see the stats, think of them less as a report card on AI and more as a report card on using AI without a process.

Share

Dax is the CEO & Co-Founder at FOMO.ai. Take our new free course, ‘How To Win With AI Search’.

even if your rankings look fine.

👉 Get the course here

This course shows:

• What AI Search systems look for when generating answers

• Why traditional SEO alone no longer carries visibility

• How to build durable AI visibility without chasing tools or trends

It’s short, direct, and built for operators.

For launch, we’re opening it free before the price moves to $497.

Anyone who completes the course also receives a fully customized AI Search Playbook for their brand (normally a $5,000 engagement).

If AI Search matters to your pipeline this year, this is the cleanest entry point we’ve found.

👉 Get the course here

Here’s what I finally realized: AI search is not just a tweak to the old system, it’s a fundamentally different game, and playing by yesterday’s rules only makes you invisible today.

So, what’s the core change? We’ve moved from optimizing for strings of words to having to prove, over and over, that we’re a trusted, authoritative answer to the questions real people are asking. Algorithms now care less about what you say you are, and much more about what the world says about you when you’re not in the room.

My thesis is simple: If you want to win at AI search, stop chasing keywords and start earning trust, both on your site and out in the wider world.

It Used to Be About Keywords, Now It’s About Questions and Trust

Let’s start with how dramatic the shift really is. In the old SEO world, it was all about “defining your industry based on keywords.” You’d figure out which words people searched for, sprinkle them lovingly through your pages, and cross your fingers that Google’s bots approved. The battlefield was your own website. Your weapons were technical tweaks and content stuffed with those golden phrases.

AI search ripped up that map. Now, the questions people actually ask, sometimes even spoken out loud, are the starting point. The engines (think Gemini, ChatGPT, Google’s AI overviews) aren’t looking for matches to keywords. They want to answer real, sometimes messy, human queries. And they know that getting the answer wrong could destroy their credibility.

That’s changed the job of AI search entirely. The platforms that deliver answers are “scared to give the wrong answer” and need to be sure they’re sending people somewhere authoritative, somewhere “trusted.” Otherwise, why risk their reputation on you?

Authority and trust, those aren’t just buzzwords. They are the only change that matters now.

Authority Isn’t Just a Ranking Factor, It’s Survival

To become that trusted source, you can’t just say you know your stuff. You have to show it, everywhere. I learned that authority signals aren’t born on your domain alone, they can live anywhere your brand or your team is mentioned, rated, or reviewed. And if you’re in a “your money or your life” category, health, finance, law, this goes double or triple.

So what does real authority look like? On-site, there are things you can control:

• Authors: List recognized humans, not just faceless “teams.” Use schema markup so AI can actually “see” who’s behind the wisdom.

• Reviews and accreditations: Especially in regulated industries, the presence of industry-specific certifications (and clear display of “medically reviewed by,” “licensed by,” etc.) makes a difference.

• Relevant industry markers: Awards, partnerships, any signpost that screams, “You can trust us, others in the field do.”

But those are just table stakes. Here’s where I had my next wake-up call: AI search engines don’t just judge your website. They judge everything said about you, everywhere.

It might be comforting to think, “If I just get my house in order, I’m good.” But that’s naïve in the AI era. External reviews, off-site rankings, chatter on platforms that matter, these influence how engines see you as an entity. For a law firm, that could mean what people say on Legal G’s List. For doctors, it’s the physician rating sites. For B2B, it’s whatever industry insiders use.

What threw me, at first, was the extent to which off-site signals might outweigh anything you do on your own domain. Even if you’ve optimized every header and alt tag, it means nothing if the world’s reviews paint you as untrustworthy.

The Moment When Reputation Hits Reality

There’s a certain pain you feel when you realize AI isn’t listening to your PR team, it’s reading your reviews. I’ve seen the consequences firsthand, and frankly, they can be brutal. I’ve had clients come to me wondering why, despite all the SEO “best practices,” they were getting nowhere with AI-driven results. The real answer? Their actual reputation was working against them.

When you’re sitting across from a business with a “domestic score of 2.5 out of 5,” you see the frustration and, underneath it, the sense of helplessness. If customers don’t like your output, or worse, find you untrustworthy, AI engines won’t trust you either. It’s as simple as that. Sometimes you have to say it directly: “Your problem isn’t your SEO. It’s your business.”

That’s an uncomfortable truth. You can’t hide a bad reputation behind clever optimization anymore. AI reads the room, and what the room says matters more than what you want to say.

PR Is Having a Quiet Renaissance, but Only If Built on Substance

Interestingly, all this has breathed new life into old-school PR, but in a way that’s finally measurable. For years, PR was the expense that got questioned at every board meeting, sure, it felt nice to be in that publication, but could anyone prove it moved the needle?

Now, it’s possible. Every time your name or brand gets quoted and cited “on third-party places that Google has already accepted as being trustworthy,” you have another brick in your authority wall. You can even watch your authority scores improve numerically, no longer relying on wishful thinking. Candidly, I think this has made companies rethink PR’s value, because now, “by looking at authority scoring, you can actually now get a numerical sense of to what degree is PR helping you.”

So how do you play this new PR game? You have to get yourself (and your business) mentioned in places that matter, not in the eyes of your ego, but in the eyes of AI. That means industry publications, review platforms, or wherever your audience gets information.

And here’s the kicker: sometimes, that means hustling just to get a listing on the relevant sites. Other times, it means being strategic in targeting the publications that consistently show up when you Google the questions your target audience is asking. Don’t guess, search those questions yourself, see which brands and sources appear, and make it your job to be included there.

If You Want to Win at AI Search, Be a Better Business

You can’t fake it anymore. That’s both terrifying and liberating.

When I look back at what actually works, it’s almost disarmingly simple. The businesses that succeed in AI search are the ones that are… just better. They deliver good experiences. People like them enough to leave positive reviews. Their expertise is recognized in their industries, and they aren’t chasing every SEO fad or shortcut.

Sometimes the advice is “be a better business, because that really matters today.” That may sound mundane, but it’s the honest answer.

The other shift? “Don’t be so focused on keywords and spammy SEO tactics.” Instead, spend your energy figuring out what questions your real audience actually cares about and then provide answers, real, valuable, cited, credentialed answers.

I wish someone had told me that years ago.

Here’s Where It Gets Complicated (And Good)

Part of me wants to say, “That’s it! Just do good work and you’ll be rewarded.” But I know the truth is more nuanced.

Earning trust isn’t “set and forget.” It requires absolute honesty about your weaknesses, relentless improvement where you fall short, and a willingness to seek reviews and quotes, even when it feels awkward. Authority signals don’t build themselves.

And you have to do the work to find where your audience’s questions live and which sources count. This isn’t spray-and-pray PR. It’s surgical. Sometimes that means humbling yourself, asking past customers for reviews, tracking down the right industry lists and forums, or crafting content that answers questions your competitors ignore.

But once you start seeing off-site authority translate to actual AI search prominence, it’s hard to go back to the old ways. There’s a feedback loop: higher quality leads to better reviews, which leads to more visibility, which leads to more accountability. The engine rewards authenticity over artifice, reliability over ranking hacks.

Is this more work than the old “optimize and forget” world? Absolutely. But it’s work that actually makes your business stronger, not just your search results.

What You Can Use Today

Here’s my short checklist for adapting to the new AI search reality:

• Audit your reputation, not just your site. Google your own business and your category’s typical questions. If you don’t like what you see, start fixing it, urgently.

• Make sure every author or expert connected to your site is listed with real names, titles, and credentials. Use schema markup so AI systems can “see” them.

• Get reviews, not just on your own site, but across industry-specific platforms that customers and algorithms both trust.

• Don’t ignore PR. Target getting mentioned and quoted where it counts. Find out which publications show up when your audience searches your category’s questions. Put yourself in their path.

• Consider business operations part of your search strategy. If customers are unhappy, fix the real problem before you optimize content.

• Shift your writing focus from keywords to real questions. Build content that answers them completely, with citations and proof.

If it feels like slow work, that probably means you’re doing it right.

If You Only Remember One Thing...

AI search doesn’t care about your keywords; it cares about whether you’re the most trustworthy answer to a real, human question. Authority comes from what the world says about you, not just what you say about yourself. Be obsessed with earning trust, everywhere.

If any of this rings true, hit reply and let me know where you’re seeing the biggest challenges. Or if you’ve cracked the off-site authority code, I’d love to hear your story. If this was helpful, consider subscribing or share it with someone mired in old-school SEO.

This article was written with the help of HelloGordon.com. Spend 10 minutes being interviewed by Gordon, and instantly generate content, quotes, and ideas.

The twist?
Some of those pages even include the line:

“The content is adapted from Wikipedia, licensed under Creative Commons.”

That’s right — Grokipedia is basically Wikipedia with a Musk filter and AI slop frosting.

The AI Slop Problem

Early users noticed that Grokipedia pages often hallucinate, cherry-pick facts, and even contradict Grok — xAI’s own chatbot.

When asked whether Grokipedia was a fair and unbiased source, Grok answered:

“Grokipedia is not a fair and unbiased source of information… it’s a critique wearing encyclopedia clothing.”

Ouch. When your own AI calls your product biased, it’s not ideal launch PR.

Why SEOs Are Losing It

Some SEOs are calling Grokipedia a “mass-scale content clone” that could break search engines if it gets indexed.

Think about it:

• Millions of nearly identical Wikipedia pages.

• Hosted on a brand-new domain with a billionaire’s backlinks.

• All AI-enhanced and ready to ride Google’s freshness algorithm.

That’s not just plagiarism — it’s an SEO moonshot.

If Google doesn’t clamp down, Grokipedia could temporarily outrank Wikipedia by sheer authority and recency.

But if Google applies the same filters it uses for parasite SEO and content farms, it’ll get buried faster than Musk’s Threads account.

What the Experts Are Saying

Lily Ray has warned for months that AI content will create “a vicious cycle” of spammy tactics that Google will inevitably punish:

“Shortcuts like this will get fixed. Any rankings built on spammy content will eventually be lost.”

Jes Scholz, who is a member of the FOMO.ai AI Search Advisory Board, frames it differently:

“Search isn’t about ranking first anymore — it’s about being quoted first.”

In other words, Grokipedia might generate volume, but Wikipedia still owns the trust, at least for now. And in the new AI Search era, trust > traffic.

FOMO.ai Take

This is more than Musk vs. Wikipedia, it’s a preview of AI content at scale colliding with the concept of “originality.”

Search engines are already struggling to decide:

• What’s an original idea vs. a remix? (Consider The Shitty Listicle Problem)

• What counts as “trusted” when everyone’s using AI?

• Should AI encyclopedias get indexed at all?

For marketers and brands, the playbook doesn’t change

1. Be the source others cite (don’t rely on aggregators).

2. Invest in entity authority — your brand, experts, and unique data.

3. Blend human + AI for credibility and scale.

Because when the internet floods with Grokipedias, the only things that stand out are real voices, verified facts, and original POVs.

Dax Hamman is the CEO at FOMO.ai. FOMO.ai helps you win with AI Search.

Where there were voices, now there are listicles.

• Ten best…

• Seven top…

• Five you can’t live without…

And when AI showed up, instead of saving us from this swamp, it pulled up a stool, poured itself a drink, and said, “Let’s make more lists.”!

The machine’s version of research is basically reading every “Top 10” article ever written and remixing them into an even blander “Top 5.” It thinks it’s helping — faster answers, fewer clicks — but what it’s really done is laminate the web.

Everything glossy, nothing alive.

Subscribe now

Ask any AI search right now for “best protein powder,” “best AI tool for content,” “best productivity app.” You’ll get a polite list of brands you’ve already seen, in roughly the same order, as if the world reached a silent agreement that those are the only names worth knowing.

It’s the internet equivalent of a hotel breakfast buffet: technically abundant, spiritually empty.

This didn’t happen by accident.

Publishers spent years training Google to love listicles. They are easy to scan, full of affiliate links, and conveniently structured for ranking.

Every “Top 10” made the next one easier to justify.

So when AI came along, it looked at the landscape and thought: Oh, this must be how humans explain expertise.

It’s not even wrong. We rewarded predictability, but now we’re drowning in it.

But here’s the uncomfortable truth: being on those lists still works, albeit likely for the short term.

You get visibility if you make the cut (or pay to make the cut), whether in a human-written roundup or an AI-generated one. People don’t click as much anymore, but the name recognition still lands.

It’s like background music in a movie: no one notices it, but it shapes the whole scene.

So you can roll your eyes at “The 10 Best Email Tools in 2025,” but if your product’s in there, you’ll quietly cash in on the benefits.

That’s the maddening pragmatism of it: the lists are shitty, but they’re still currency.

The problem isn’t that AI likes lists; it (feels like it) only likes lists.

It doesn’t handle contradiction well. It can’t stomach the messy stuff — the “depends on your situation,” the “this one’s great unless you care about privacy,” the “actually, none of these are that good.”

AI prefers clean certainty.

And certainty, on the internet, is mostly a lie told neatly.

When everything is “Top 10” and “Best Of,” there’s no room left for stories, context, or evolution. You can’t see what changed, only what ranked.

The better move isn’t to rage against it but to outgrow it.

If you make things, write things, sell things, don’t build for the listicle, build under it.

Give the machine what it wants on the surface, but fill it with something real underneath. Make the AI summarize you, but make sure when it does, the quote still sounds human.

Add your caveats. Include the edge cases. Tell the reader what didn’t work.
That’s the content that survives being flattened, the kind that still smells like a person made it.

AI search didn’t invent the shitty listicle. It just industrialized it, and until we stop feeding the system nothing but predictable rankings, it’ll keep spitting out the same five answers wrapped in a different logo.

So, sure, keep getting yourself on the lists; that’s working for now. But somewhere beneath the numbered headlines, write the thing that can’t be summarized, because someday, someone will actually want to know what’s real, not just what’s ranked.

👉 And if you’re a brand working hard to get yourself on lists, just be aware that the boffins in ChatGPT, Perplexity & Google have seen this same problem, and they HATE it enough to be working on irradiating them. What will that mean for your rankings & citation volumes?

(Goes off to write the top 10 problems brands paying to be on shitty listicles will face in a few months… 😀)

Share

Dax Hamman is the CEO & Co-Founder at FOMO.ai and an expert in AI Search & Marketing.

This weekend, I was sent two articles that so perfectly articulate who we are and why we made the decisions we made, I had to share them with you.

At its core, FOMO.ai is a hybrid business.

We built our process first using our human team, then automated it, then began productizing it. From day one, we’ve been focused on generating real revenue while increasing the percentage of automated effort each quarter.

That model - services first, automation layered second, and productization third - is exactly what the essays below describe.

Subscribe now

(The author, , was a founder-in-residence at Google, and GM of Google Labs).

• The Third Way

• AI Second (AI2)

If you look at our current investor deck (DM me for access), (where we show growth measured both in revenue and percentage automated), you will see that we are straight out of AI2 thinking! Our deliberate capital strategy of raising carefully, with control intact, and preserving optionality is textbook Third Way.

So what does this look like in market for us?
The last two quarters have given us proof points at the largest scale yet:

• A Q2 whale opportunity: After asking their agency, “What do we do about AI Search?” they waited five months for nine pieces of content and no strategy. In the first 40 days of our pilot, we delivered hundreds of assets that have started to move the needle. That pilot is now very likely becoming a full-time engagement starting in October.

• Two more whale opportunities in Q3/Q4: Both are long-time clients of known agencies (big names, fancy offices, steakhouse lunches). Both CMOs are fed up. They came to us because we could sit across the table, show them the roadmap, and then quietly go build what they needed. Yawn to the incumbents.

• Agencies themselves: Several are now asking us to white-label—sometimes even for the same brands where we’re running pilots. Proof that the market recognizes what we’ve built.

• SMBs: Once we crossed enough of the automation tipping point, we could package our solution into a <$500/month offering that generated $82k in September.

Our structure makes this possible.

For large brands, we run pilots as set budgets with flexible allocation. No change orders, no upsells, no nickel-and-diming. Each period, we deliver more than the last, because as AI scales inside our workflows, clients get more for the same spend.

This is what AI Second looks like in practice:

• Trust precedes tech. Clients buy the human relationship, then benefit from automation compounding in the background.

• Cashflow funds R&D, letting us experiment and productize without over-raising.

• Domain expertise guides automation, giving us defensible moats incumbents can’t match.

And it’s precisely why we (unknowingly) followed The Third Way in capital strategy: raising carefully gave us independence. We can grow, acquire, or join something bigger, but always from a position of strength.

Why this matters now
The stable marketing playbook of the last 15 years is broken. Clients don’t articulate it as “AI Second” or “The Third Way.” What they say is:

• My traffic is way down. My Google ads are rising in cost.

• My audience is using ChatGPT more and more.

• Don’t let me miss out on AI.

• Help me feel the efficiency gains without losing the human touch.

• Be ahead of me, so you can bring me along.

• Deliver more for the same spend.

That is FOMO.ai’s exact model. And it is working.

In a recent interview, Bezos said that we’ll start building gigawatt-scale data centers in space within the next ten to twenty years. He said this quietly, almost like it was inevitable.

In orbit, solar power runs 24 hours a day. There are no clouds, storms, or nights. Cooling is free. Once that balance tips, computing in space could become cheaper than computing on Earth.

That one shift would change the economics of AI entirely.

Subscribe now

Why this matters for AI

AI doesn’t stop because of ambition. It stops because of energy.
Training large models takes as much power as small nations. On Earth, that scale eventually hits a wall. In space, the wall disappears.

This is where Bezos’ prediction intersects with what we think about daily at FOMO.ai. The future of intelligence isn’t just about smarter models; it’s about where and how they’re powered.

If energy becomes limitless, creativity follows. When computing moves off-planet, it unlocks a kind of acceleration we haven’t seen since the early Internet.

The deeper play

Bezos has always described Earth as a “garden” planet — somewhere to preserve while industry expands elsewhere.

It’s not about Mars. It’s about moving the heavy work away from home.

Satellites gave us global communication. Then came cloud storage and global compute.

The next layer might be orbital AI clusters — machines learning above the atmosphere, powering everything from simulation to synthetic biology.

It sounds distant, but so did cloud computing once.

What this means for builders

I believe every brand is now in the intelligence business.

As the infrastructure shifts, so does the opportunity. If data centers leave Earth, the brands that understand how to train, apply, and deploy within that new ecosystem will lead the market.

Share

Bezos points to a future where AI’s limits aren’t technical anymore — they’re imaginative. And that’s exactly where we like to work.

On one side are the builders, people like , who shape the tools.

On the other hand are the consultants, firms like McKinsey, who shape the narrative in boardrooms.

This week, both sides dropped big ideas that are pretty contrasting.

Karpathy’s “Software 3.0”

Speaking at Y Combinator’s Startup School, Karpathy outlined what he calls Software 3.0—the next paradigm in computing.

His core thesis:

• The new programming language is English.

• Large language models (LLMs) aren’t deterministic code, but “people spirits”—stochastic simulations of people.

• This explains why LLMs feel human, yet jagged and unreliable.

Karpathy’s caution is refreshing in an era of AI hype. He argues:

• Humans must remain in the loop. AI generates, humans validate.

• Software design should assume this validator loop is permanent (at least for now).

• Keep AI on a short leash: constrain outputs so humans can realistically review them. (If you can only check 10 ads, don’t generate 100.)

In short, Karpathy sees AI as a design problem, not a magic agent.

McKinsey’s Vision

McKinsey, by contrast, pushes the narrative of autonomous AI agents taking over business processes. It’s optimistic—arguably over-optimistic.

And here’s the tension:

• Builders like Karpathy focus on the jagged, stochastic nature of LLMs.

• Consultants like McKinsey emphasize transformation stories that excite executives.

One vision grounds itself in technical limits. The other sells a future to decision-makers.

This isn’t just an academic debate - McKinsey is shaping how Fortune 500 boards think about AI adoption! Karpathy shapes how developers actually build it.

If the boardroom over-indexes on McKinsey’s hype without understanding Karpathy’s caution, companies risk expensive bets on unreliable autonomy.

The smart path forward: design for “people spirits,” keep humans in the loop, and resist the urge to outsource too much too soon.

👉 What do you think? Should AI be unleashed as McKinsey suggests, or constrained as Karpathy urges?

Published by Dax Hamman, CEO & Co-Founder at FOMO.ai, an AI Search & Marketing expert.

When OpenAI announced Instant Checkout, it wasn’t the only company moving on this front. In the last two months, Google, Perplexity, and Amazon have all shown their cards. The differences matter.

Google: building the pipes

Google’s move is the Agent Payments Protocol (AP2). Unlike OpenAI, which jumped straight to showing buyers a checkout button, Google is working on the infrastructure.

• AP2 is an open protocol for agent-led payments, designed to work across cards, banks, and even stablecoins.

• More than 60 partners are on board already—payments networks, merchants, processors.

• The goal is to become the default fabric for transactions whenever an AI agent needs to pay.

At the same time, Google is pulling “AI Mode” into Chrome and Search. Discovery, comparison, and decision-making still happen inside Google’s surfaces, but AP2 ensures that when an action happens, it’s trusted and portable.

Difference from OpenAI: Google isn’t yet making checkout happen inside Search. They’re setting the rules of the road so any agent can transact safely. OpenAI is the first to make the consumer-facing demo work.

Perplexity: answers that act

Perplexity has been experimenting with checkout inside answers. Ask about a product, and you may see a buy button powered by PayPal or Venmo. Ask about travel, and you may see booking options.

This positions Perplexity as an answer + action layer. The answer is no longer the end of the journey; it’s the place where the action starts.

Difference from OpenAI: Perplexity doesn’t have the scale of ChatGPT, but it’s showing how “actionable answers” might look across categories. It feels like a test bed for blending information and commerce.

Amazon: fortifying the walls

Amazon hasn’t talked about open protocols or checkout in other agents. Instead, it has:

• Upgraded Alexa with new “Alexa+” capabilities.

• Rolled out smarter seller agents to automate merchant tasks inside its marketplace.

The strategy is clear: keep shoppers and sellers inside Amazon’s ecosystem. Instead of joining a neutral protocol or exposing its catalog to outside agents, Amazon doubles down on control.

Difference from OpenAI: Amazon doesn’t want to collapse discovery and purchase outside its platform. It wants to strengthen its own closed loop.

Why these differences matter

• OpenAI is creating a new channel, turning ChatGPT into a storefront with an open-source standard underneath.

• Google is building the standards others will have to use, aiming to control the rails more than the storefront.

• Perplexity is testing the UX of “answers that act,” without yet solving scale.

• Amazon is closing its walls tighter, betting on control instead of interoperability.

For marketers and business owners, this means there won’t be a single model. In one environment, you’ll transact directly in the chat. In another, you’ll still drive to your site but through new protocols. In another, you may never see the buyer unless you’re inside Amazon.

OpenAI just made it possible to buy products directly inside ChatGPT. This is the logical next phase following the recent addition of product cards in their results.

For now it works with Etsy sellers and will soon extend to Shopify stores like Glossier, Skims, Spanx, and Vuori. Instead of clicking through to a website, adding something to a cart, and moving through checkout screens, the purchase can be completed inside the chat.

It’s a small rollout - single items, U.S. only - but it’s the clearest sign yet that search and purchase are merging.

What this means for consumer brands

Most consumer marketing has been built around driving people to a brand’s own site. That’s where the brand story lives, where upsells happen, where loyalty is built. But the buyer may never get that far anymore.

If someone asks ChatGPT for “best yoga mat under $100,” they’ll see a few options and can purchase on the spot. The seller still fulfills the order, but the decision is made within seconds.

For brands, that means:

• The opportunity is to be the product that gets surfaced.

• The risk is that you never get seen if you aren’t chosen in those answers.

(It should be noted that as an industry, we have put a MASSIVE amount of time into multi-channel attribution… and now first touch might commonly be the transaction. The task ahead for marketers is not inconsequential.)

What this means for B2B

It’s tempting to view this only as a retail story, but the model applies to B2B too. Many companies have leaned on gated assets—whitepapers, guides, templates—to capture leads.

In a chat-driven checkout environment, those assets may be offered as $0 / low cost purchases.

A prospect asks for a compliance checklist. ChatGPT offers yours. They hit “buy,” receive the asset instantly, and you collect the lead.

The difference is that the form-filling dance is gone, (ChatGPT will likely store your name, address, email etc like TikTok and Meta does today). The content library becomes a storefront of claimable assets, each one an on-ramp into your funnel.

How this compares to others

OpenAI isn’t the only one moving here. Each of the big players is carving out its own path:

• Google just launched the Agent Payments Protocol (AP2). Unlike OpenAI, which has started by baking checkout into ChatGPT, Google is building the technical standard for AI-led transactions across the web. They’ve lined up more than 60 payments and commerce partners. The practical difference: Google is laying the pipes, while OpenAI is showing what the faucet looks like.

• Perplexity has been experimenting with checkout inside answers using PayPal and Venmo. Ask for a product, and you may see not just information but an option to buy. Their approach feels like a preview of what “answers plus actions” might look like at scale.

• Amazon is focused inward. It hasn’t opened up checkout to the broader web but is strengthening Alexa and building agent tools for its own sellers. Amazon’s play is to keep you inside its ecosystem, not to create an open standard.

The common thread: every platform is moving from being the place where discovery happens to being the place where the transaction happens.

Search has already shifted from lists of links to direct answers. Now those answers are becoming transactional.

That creates three fronts every marketer has to care about:

• SEO: the old web visibility game.

• Answer optimization (GEO): making sure your content or product is surfaced in generative answers.

• Agentic optimization (AEO): ensuring your product, service, or asset is ready to be purchased or claimed the moment it appears.

I am lobbying hard that we group SEO, GEO, AEO and Agentic Search to be referred to as AI Search fo simplicity for everyone!

What to do now

• For consumer brands: start preparing your catalogs for chat-native checkout. Test integrations early.

• For B2B: rethink lead magnets. Package your guides, templates, or tools as checkout-ready assets. Spend time on Etsy and see what B2B offerings are winning there.

• For everyone: adapt your search strategy. It’s no longer just about ranking, it’s about being surfaced and being actionable in the same moment.

The search box and the checkout button are becoming the same place.

Want help winning in AI Search? Schedule a call.

👇 Download the full FOMO.ai The State of AI Search report.

Search has been changing, and we found most brands aren't taking the right types of action.

For years, we built our marketing strategies on a simple truth: If you rank on page one of Google, you get traffic. You get leads. You grow.

But now?

Google has replaced the top of search results with AI Overviews, summarized answers that don’t send users to your site. ChatGPT can now recommend and sell products mid-conversation, without ever linking out. And Bing, Perplexity, and others are doing the same.

Users don’t scroll anymore.

They ask questions… and AI answers them.

That means:

• Fewer clicks

• Fewer citations

• Fewer chances to be seen

Unless you're one of the few brands the AI chooses to feature.

That’s why we published the State of AI Search report.

It’s a deep dive into:
✅ Why organic traffic is declining (and what’s replacing it)
✅ How AI chooses what content—and brands—to cite
✅ What Google, OpenAI, and Microsoft aren’t telling you
✅ What marketers need to do right now to stay visible

TL;DR: If your brand isn’t cited by AI, it’s invisible.

This is the single biggest shift in search since the rise of Google.

We want every marketer, founder, and strategist to be ready.

👇 Download the full FOMO.ai The State of AI Search report.