When it comes to discovering information, Google has dominated the last 20 plus years.

Only with the 2023 launch of ChatGPT did the search monopoly begin to crack.

Now it seems to be such common knowledge even judges in Delaware acknowledge that search is no longer a monopoly—thwarting the breakup sought by the DOJ.

Of course, some folks are saying Google is going to win again.

From far behind in 2023, Gemini chatbot is now a leader, and Google search now often serves up AI Overviews instead of blue links. Plus just announced AI shopping tools.

Still, given Google’s cannibalizing its own core search business to stay relevant, it suggests it has got some serious competition this time around.

So what is going to happen?

Well, we may be done with the Winner Take All world.

Wouldn’t be such a bad thing.

Instead, we will have multiple sources for information and analysis, each with strengths and weaknesses. Instead of defaulting to Google.

Sure, it’s true that some of the LLMs, like ChatGPT, are far too reliant on single sources. Such as Wikipedia. But even that’s changing, driven by the LLM competition. Elon Musk’s Grok recently launched Grokipedia. Pretty rough now, but it will get better.

Plus we’ve got a full plate of LLMs with real user bases. Perplexity for research, Claude for coding and no doubt more custom LLMs coming soon to disrupt every disruptible field in the US and the world. Which is just about everything in professional services, the arts and information technology.

What does this mean for your brand’s or business’ efforts to optimize for AI discovery, reputation and ultimately brand awareness and sales?

First, it means that silos are over. Google’s total dominance of search and keywords is over. Meta’s dominance of digital product advertising is next. As the Chatbots play a bigger and bigger role in recommendations and discovery.

Second, it means you need to coordinate much more effectively across all your channels to succeed in a world where no channel dominates.

Turn into our final law to learn how to make it happen.

Leave a comment

We’ve all heard lots of things about the consumer journey, and why it’s important. But what is it?

Well, the idea was born as far back as 1898 by someone named Elias St. Elmo Lewis. They don’t make names like that anymore. An advertising pioneer, Lewis wanted to bring scientific rigor to selling products, and in those early days of industrializing, the challenge wasn’t building a brand or relentless competition. It was consumer ignorance.

New products were flooding the market, and the public needed to be educated before they could be persuaded.

Lewis developed a model to guide a prospect to a sale. The consumer journey, in other words. You’re probably familiar with Lewis’ model: AIDA (Attention, Interest, Desire, Action). As fit the media market of those days, it was linear.

• Attention: The first crucial step. If you fail here, Lewis argued, nothing else matters.

• Interest: Transform fleeting awareness into deeper curiosity about the product’s features, utility, etc.

• Desire: Go beyond the rational mind (”I get it”) and tap the power of emotions (”I want it”). It’s where brand building starts.

• Action: If all the other stages are done right, Lewis felt consumer action was a natural, inevitable consequence.

Today, more than a century later, the AIDA model is still in marketing textbooks. It aligns with behavioral psychology. It’s core.

Until today.

The chatbots blew it up.

Perplexity, Claude, Gemini, they don’t take the consumer journey. They’re not consumers. They’re predictive algorithms. They follow the beat of their own drummer.

They travel on the Bot’s Journey.

It’s not about Attention or Interest. It’s certainly not about desire. And the chatbots can’t actually do anything but, well, what they do. Analyze language and find patterns in response to queries.

So to map the bot’s journey, we need to know where they look for information and in what sequences. In other words, what happens after the first question is asked. So here’s a shot at the new AIDA.

Relevance–the bots crawl the web looking for pertinent information based on the analysis of word patterns.

Authority–the bots look for sites with authority based on links and traffic (we think). But it doesn’t work like Google sera

They select the best sources available. That doesn’t mean the New York Times. Behind a paywall. But too often it does mean Wikipedia.

Discovery–They prefer HTML to PDFs. All technical choices must be optimized to ensure the answer engines find the right material.

Depth–They favor information that responds as clearly as possible to the initial query and follow ups.

So, if you want to make sure you’re part of the Bot’s Journey.

It’s time to get RADD!

Leave a comment

For decades, the communications and marketing world has operated under a single, unchallenged theology: The “Mount Olympus” of credibility. If you landed a feature in the New York Times, the Financial Times, or the Wall Street Journal, you had reached the pinnacle. You weren’t just getting a story; you were buying a permanent seat at the top of Google’s PageRank system. Search was and is snobby. It favors the professional creative classes and the legacy media giants, leaving smaller players and even major corporate brands to fight for the scraps.

But the ground just shifted beneath our feet. Per a Britopian analysis, earned media accounts for a low of 5.7% to high of 15.4% of media citations in responses to queries related to the broad technology trends including cloud infrastructure, AI/data analytics, security/compliance and collaboration productivity.

Now drill down further. Within AI answers, the competition for “top-tier” visibility is intense. Per a Gemini analysis drawing on disparate sources, AI platforms often concentrate 34% of their journalistic citations on just one or two dominant brands.

Bottom line: we are spending millions of dollars and thousands of man-hours chasing a sliver of the digital pie that represents less than 16% of how the new world discovers information.

Summary of Citation Metrics for the

“Big Three”

Why is this happening? Because AI is hungry. Unlike the snot-nosed search engines of the past that cared about the prestige of a backlink, LLMs like ChatGPT, Perplexity, and Claude are optimized to provide specific, detailed answers to increasingly granular questions. Journalism, by its very nature and business model, cannot provide this level of deep, technical knowledge. A journalist needs a “hook” and a “narrative” to drive clicks; a robot just needs the facts to fulfill a query.

This is the reality of the “Zero Click” world. When 60% of searches end without a click because the answer is served directly by the AI, the “Blue Link” economy is effectively over. If you are still measuring success by your PR firm’s ability to get you into the Associated Press, you are measuring a ghost. The real battle for influence is happening in the other 97.4%—the technical white papers, the deep-dive blogs, and the structured data that the hungry machines are actually eating.

It is time to pivot from “prestige” to “presence.” You don’t need a journalist to validate your existence anymore; you need to feed the bot.

Leave a comment

Gemini sources for Citation Metrics for the “Big Three”

• https://arxiv.org/html/2507.05301v1

• https://ppc.land/chatgpt-referrals-surge-as-google-searches-hit-zero-click-records/

• https://www.position.digital/blog/ai-seo-statistics/

• https://ahrefs.com/blog/ai-seo-statistics/

After nearly 17 years working with Google Ads and over a decade working with Meta ads, the most important lesson I’ve learned is that the best marketers don’t obsess over tools, dashboards, or the latest growth hacks.

They focus on the bread-and-butter fundamentals. They focus on the handful of metrics that actually move the needle. And most importantly, they understand their customers deeply.

Recently, I’ve been hearing a message more and more, especially from people inside the ad platforms themselves:

“You don’t need to worry about audiences anymore. Just upload great creative and let Meta find the customers.”

On paper, it sounds great. Bring good creative, a good offer, and a solid landing page, and Meta’s algorithm will figure everything else out.

But after years of running campaigns and watching brands scale (and fail), I can tell you there are three big reasons why that approach doesn’t actually work in the real world.

1. Meta Optimizes for Meta — Not for You

The first thing every advertiser needs to remember is simple: Meta’s algorithm is optimized for Meta’s business first.

Yes, you can select campaign goals like: Conversions, Leads, Purchases, Email signups

But at the end of the day, the platform’s primary incentive is to maximize ad spend and platform efficiency, not to guarantee that your brand wins.

If every advertiser is handing the algorithm full control, then the system ultimately decides which advertiser gets the best audience.

And guess what? Meta has no loyalty to your brand. If you’re going head to head with a large incumbent brand that is spending 10x your budget, guess who META gives the algo love? Not you!

2. No One Will Ever Know Your Customers Better Than You

One of the biggest common traits I see among brands that scale successfully—especially in direct-to-consumer—is that they deeply understand their customers.

Not just basic demographic targeting. I’m talking about real customer intelligence.

The brands that grow fastest are the ones that:

• Download and analyze their customer data

• Build detailed customer personas

• Study spending behavior across thousands of customers

• Segment audiences based on real purchasing patterns

Instead of thinking about a customer as: “Jane Smith in Mississippi.”

They understand something closer to: “Jane Smith, age 35, two kids, spends $120 every three months, typically purchases after seeing educational content.”

The platforms claim their algorithms build these personas automatically. But in practice, the brands doing their own analysis almost always outperform the ones that rely purely on automation.

Industry data supports this as well.

• According to Meta’s own marketing science reports, campaigns using first-party data and customer segmentation can improve return on ad spend by up to 2–3× compared to broad automated targeting alone.

• Research from eMarketer and Nielsen has shown that brands leveraging first-party customer insights improve conversion rates by roughly 20–30%.

• Meanwhile, Gartner reports that companies using customer data for personalization generate up to 40% more revenue from those activities than average competitors.

The takeaway is simple: Algorithms are powerful. But they don’t replace customer understanding.

Every practitioner I’ve seen who says: “It’s all creative now. Just let the algorithm handle everything”...ventually runs into problems. Because creativity without customer insight is just guessing.

3. The Algorithms Change Too Fast

The third issue is something that’s gotten significantly worse over the last few years:

The pace of change inside ad platforms has accelerated dramatically. Both Meta and Google update their algorithms constantly.

A campaign will perform well for a while. Then suddenly performance collapses. The feedback sounds like this:

“Everything was working great… and then two weeks ago it just died.”

When advertisers rely entirely on automation, they often lose visibility into what actually caused the change. The speed of algorithm changes means that blind trust in automation creates instability. The more control you give up, the harder it becomes to diagnose problems when things break.

The Hidden Cost of “Learning Mode”

There’s another detail most marketers overlook. Every algorithmic ad platform—including Meta and Google—relies on something called learning mode.

Learning mode essentially means: “Let us spend your money while we figure out who your customers are.”

Think about that for a moment. The platform is using your budget to discover information you should already know. You’re paying for the algorithm to learn. But if you truly understand your customers, you can start much closer to the answer from day one. Which means less wasted spending.

At the end of the day, the brands that succeed in digital advertising aren’t the ones who rely entirely on automation. They’re the ones who do the hard work most companies avoid, which is knowing their customers inside and out. Then they use platforms like Meta and Google as distribution engines, not decision makers.

Bottom line: brands that aren’t investing in customer intelligence today are probably leaving 15–20% of potential revenue on the table. In modern digital marketing, that’s a gap your competitors will happily take.

Your website, your most likely boring corporate website, is where you need to start.

I can already hear the objections.

I don’t want to mess with the site.

Do we really need to publish more content?

Nobody reads it anyway…

Stop. Right. There.

You’re fighting the last war–the site visits, user engagement and SEO battles.

We’re in the age of AI. Don’t forget law number one, the most important law.

AI is now your most important audience. And per law four, you need to write for the robots, not just people.

There’s no better, no more efficient place to do that than on your owned digital properties–your website at the core.

So what do you do exactly?

You use law six, know thy queries, to map out the 40 to 60 most important questions for your brand. You identify where you appear, what’s right, what’s wrong and what’s missing.

Then you create articles, charts and videos for the robots.

You don’t worry about narratives for humans who expect to be pulled into an article or story or video.

You answer the questions clearly and specifically.

Every single one of them.

You add pages to your website–not for the human journey but for the bot’s journey.

You make sure the pages are discoverable.

You structure your sentences around the positive sentiment terms associated with your business, brand and product.

You isolate the negatives.

Why because machines don’t read the way we do. Law four!

Then once you see the pages index in google–the LLMs still use google to find pages–you rerun your key questions.

You track the changes.

If you are thorough and publish quality, detailed answers, you will be surprised how quickly you can impact results.

It’s the golden age of AIO, as we’ve said before.

Your website is your most valuable resource.

Use it.

In the next video, we take you a step further to track just where the bots are going. So you get the right information to the LLMs in the right places.

Over the weekend I spent a lot of time digging into the hype around OpenClaw. If you spend even ten minutes on social media right now, you’ll see posts claiming things like “I built a seven-figure business in 30 minutes with OpenClaw.”

If you scratch beneath the surface, most of those posts are just clickbait. But after spending the weekend actually trying to configure and run OpenClaw myself, I came to a different conclusion:

OpenClaw is already dead. But here’s the twist: while it’s a zombie project walking toward irrelevance, it has accidentally opened the door to an entirely new industry.

The Idea Behind OpenClaw Was Brilliant

To understand why OpenClaw matters, you need to understand where it came from.

A developer originally created a project called Claudebot, which later became Open Claude (to avoid trademark issues) and eventually evolved into OpenClaw. The core idea was simple but powerful: Instead of sending all your data to the cloud, run AI locally on your own machine.

In practice, that meant something like this:

• Take a small computer like a Mac Mini

• Turn it into a personal AI server

• Connect it to your preferred large language model

• Let it manage tasks across your system: Access your calendar, Automate workflows, Process media files, Create content and distribute it across platforms

Naturally, I thought: I need to try this.

Then the Problems Started

Actually implementing OpenClaw revealed the real story. First, the governance structure is messy. The original developer eventually joined OpenAI, and while OpenAI has shown support for the project, OpenClaw itself is maintained by a decentralized community of engineers.

That’s great for experimentation. But it’s terrible for enterprise reliability.Then there were the technical headaches.

Unless you configure everything perfectly, OpenClaw breaks constantly. Installing dependencies like Node.js can fail depending on the environment, hardware configurations vary wildly, and the interface is—being generous—rough. Even experienced engineers can run into issues.In other words, OpenClaw feels exactly like what it is: A community-driven open-source experiment. Not an enterprise platform.

But OpenClaw Proved Something Huge

Despite the problems, OpenClaw revealed something that might be even more important than the product itself. There is massive pent-up demand for private AI infrastructure.

People want AI tools that: Run locally, Protect sensitive data, Avoid massive cloud compute bills, Operate directly on personal or enterprise hardware The numbers back this up.

The global private AI market is projected to grow from $11.1 billion in 2025 to $113.7 billion by 2034, expanding at nearly 30% annually. (Dimension Market Research)

Meanwhile, the AI server market is expected to explode from roughly $124 billion in 2024 to more than $850 billion by 2030. (Grand View Research)

The Industry Has Already Noticed

The moment something proves a massive market exists, the big players move in.

• Perplexity just launched a new system called Personal Computer, designed to turn a Mac into a 24/7 AI agent running locally on your machine. (The Verge)

• Hardware companies are shipping devices that run local AI models out of the box, including AI-ready NAS systems with OpenClaw preinstalled for tasks like video editing and automation. (Tom’s Hardware)

• Lenovo has introduced powerful desktop AI workstations designed to bring large-model development directly onto personal machines. (Lifewire)

• At the same time, tech giants are investing billions in specialized AI chips and infrastructure to support the next generation of AI computing. (Reuters)

What the Next 365 Days Will Look Like

OpenClaw showed the market. Now the giants will build the real products. In the next year, expect:

• Hardware vendors shipping AI-ready machines

• LLM providers releasing local AI systems

• Enterprise platforms integrating private AI servers

Imagine a future where Microsoft Copilot, or Gemini offers two modes:

Cloud Mode: Heavy processing, Shared enterprise data, Massive compute workloads

Local Mode: Sensitive documents, Video processing, Internal company workflows, Private datasets

Instead of uploading terabytes of data to the cloud and paying for compute every time, organizations will run AI directly on local infrastructure. Consumers want this. Enterprises need this. Industries like healthcare, law, finance, and consulting practically demand it.

Why OpenClaw Won’t Survive

The irony is that OpenClaw started this conversation but probably won’t win the market it created. Why? Because it has three structural disadvantages:

1. Complex setup

2. Weak user experience

3. Decentralized development

Meanwhile, the companies entering this space have: billions in capital, dedicated engineering teams, hardware integration, massive ecosystems Once companies like Microsoft, Google, Nvidia, or Perplexity release polished alternatives, OpenClaw will struggle to keep up. Not because it was a bad idea. But because it was the prototype.

The Real Legacy of OpenClaw

History is full of technologies that changed everything but never dominated their own market. Think of: Netscape in the browser wars, Friendster in social media, Palm in smartphones

They proved something was possible. Then the giants arrived. OpenClaw may end up being the same. A year from now, no one may be talking about OpenClaw itself.

But everyone will be talking about something it helped reveal: The future of AI isn’t just in the cloud. It’s on your desk.

Welcome back to SEO is Dead’s Ten Law of AI Optimization. Let’s quickly recap where we’ve been.

Law one: AI is your brand’s most important audience.

Law two: Search engines are snobby, but AI is hungry for information, deep information. It’s the greatest opportunity for brands and businesses since the launch of the internet.

Law three: to win, you need to understand how machines read.

Law four: you need to write for robots too. If you want to make it into the Chatbot answers that matter.

Law six: algorithms are moving target, and you must act now to seize the golden moment for AIO.

Now let’s get to law six. You must know yourself to know your queries.

Sounds like pretentious nonsense, doesn’t it? Against the spirit of our laws.

But here’s the rub. To draft the right answers, you need to know the right questions, and the machines won’t tell you. They are black boxes. Google gives you a lot more. Such as the volume of searches for a keyword. ChatGPT doesn’t tell you which questions are asked most frequently.

You have to take a different approach.

If you start with a clear sense of who you are or want to be–easier said than done sometimes–but there’s no way around it to optimize for AI discovery.

Here’s the process:

First, you come up with a query list. At least 40 to 60 questions.

Second, you divide the queries into branded (using your name) and non-branded (broader category questions).

Third, you ask the questions in at least three LLMs. We like Chat, Perplexity and Google AI Overviews.

Fourth, you use the best tools available to support you, identifying where you appear and which sources the LLMs cite.

Fifth, you dig in and analyze the results.

You will learn what ChatGPT knows about you and your category, and what it doesn’t know and cannot find. You’ll discover a competitor performs better in Perplexity, and you’ll be surprised to find that Google AI Overviews rely more on your website and ChatGPT favors Wikipedia.

You’ll discover which answers you shape, which answers you don’t, and where the opportunities and challenges are in educating the Chatbots and, ultimately, your consumers or customers.

In the next law, we’ll explain why this approach to Artificial Intelligence Optimization–AIO–is urgent.

If you miss the window to shape the questions, you may not get another.

How AI Is Coming for Service Industries You’d Never Expect

I was watching a ski video with my kids the other night. They’re at that age where skiing is still half excitement, half terror. So I like them to see proper technique before we hit the mountain.

The instructor in the video was from the French Alps. Calm. Technical. Clearly world-class. And at one point he stopped and said something that caught me off guard:

“Please support real ski instructors. Don’t rely on AI-generated videos to learn how to ski.”

I said what! AI-generated ski videos? So I searched.

And sure enough — YouTube is now full of AI-stitched ski tutorials. Generic voiceovers. Stock mountain clips. Sloppy animations explaining turns and braking. It’s what we’d call AI slop. Low effort. Mass-produced. The sport itself has never been more popular. So we can see why people who don’t know anything about skiing would try to monetize via AI videos.

Skiing Is Growing. So Is the Pressure.

Since the 2022 Winter Olympics, winter sports participation has seen a measurable bump. The National Ski Areas Association reported one of the highest skier-visit seasons on record shortly after the Games — more than 65 million visits in the U.S.

Demand is strong. At the same time, costs are the highest they’ve ever been.

Lift tickets at major operators like Vail Resorts and Alterra Mountain Company can exceed $200 to $250 per day at peak. A private two-hour lesson can easily range from $400 to $800 in many markets. Group lessons often approach $200, and that’s before you factor in rentals, lodging, food, and travel.

So what happens? People look for substitutes. They show up on the mountain, skis on for the first time, phone out, watching a video about how to wedge stop.

I spoke to multiple instructors in the Northeast this season. They all said the same thing: skiing is busy and they’re working, but more people are trying YouTube first.

That’s the pattern. Demand is up. Costs are up. AI instruction is up. That’s not unique to skiing. That’s the blueprint for how AI enters service industries.

Now Fast Forward Five Years

Imagine this.

You arrive at the mountain. Instead of booking a human instructor, you rent a humanoid robot powered by something like Optimus. It skis with you. It stabilizes you through turns. It picks you up when you fall. It analyzes your posture in real time and adjusts your movements with mechanical precision no human coach could match.

Or imagine you’re wearing Meta-powered AR glasses from Meta Platforms. As you approach a slope, the glasses tell you the grade, the snow condition, and the recommended turn radius for someone at your ability level. They alert you to brake in fifteen feet. You see a holographic instructor in front of you demonstrating the move before you execute it.

You record your run. The system auto-generates a breakdown explaining that you leaned back seven percent too far in your last three turns.

That world is not science fiction. It’s inevitable. The hardware will catch up. When it does, what happens to ski instructors?

The Barbell Effect

AI does not eliminate industries. It eliminates the middle.

The mediocre instructor who simply repeats standard drills without personalization becomes replaceable. The elite instructor who embraces AI becomes harder to compete with than ever.

This is the pattern I’m seeing across service industries. If you’re average and analog, AI compresses your value. If you’re exceptional and augmented, AI multiplies it.

How a Ski Instructor Becomes 10x More Valuable

If I were a top ski instructor today, I wouldn’t resist AI. I’d build on top of it.

First, I would create real instructional content — not AI slop, but high-quality lessons where I am the expert on camera. I would use AI to enhance production with overlays, angle diagrams, slow-motion analysis, and mountain mapping, but the authority would remain human. As low-quality tutorials flood the internet, credibility becomes more valuable, not less.

Second, I would implement intelligent pre-lesson diagnostics. Before a student even arrives, they would complete a structured intake that captures athletic background, prior skiing experience, fear level, injury history, and specific goals. That data would feed into a system that helps design a customized lesson plan in advance. Instead of guessing on the lift, I’d be diagnosing from the first turn.

Third, and most powerful, I would use AR recording tools to generate post-session performance reports. Imagine recording thirty minutes of instruction and then instantly producing a personalized breakdown with annotated clips, specific corrections, and drills to practice before the next session. The lesson would no longer end when the skis come off. It would continue at home.

That doesn’t replace instruction. It upgrades it.

Will Robots Replace Ski Instructors?

Eventually, robots will exist on slopes. Eventually, AR overlays will teach beginners the basics. Entry-level instruction will likely become partially automated.

But technology analyzes mechanics. It does not read fear in a ten-year-old’s eyes at the top of a black diamond. It does not calibrate encouragement in real time. It does not understand when a student needs confidence more than correction.

Confidence, judgment, and emotional intelligence and deep expertise in coaching humans are not easily automated.

When you combine human emotional intelligence with AI performance analytics, you don’t eliminate the instructor. You evolve the role into something stronger.

The Bigger Pattern

This isn’t about skiing. It’s about every service industry that assumes it’s insulated.

Personal trainers, tennis coaches, golf pros, physical therapists, and many consultants are all on a similar trajectory. AI first enters through lower-cost digital content. Consumers experiment. Hardware improves. Performance analytics mature. The lower tier gets commoditized. The top tier gets amplified.

The Future Belongs to the Augmented Ski Expert

If you are great at what you do and you embrace AI, your value increases. If you ignore it, your value compresses. Five years from now, you might see a robot skiing.

But the human who understands how to combine craft, judgment, and intelligence — that’s the one people will still book, and likely pay more for. Same slopes. Same snow. Different economics.

SEO & AIO: The Rise of “Zero-Click” Primacy

New industry benchmarks released yesterday reveal that Google AI Overviews (AIO) now appear in over 25% of all searches, with Health Care reaching a staggering 49%. The report confirms that traditional SEO is no longer about the click, but about “Brand Visibility” within the summary.

• Key Stat: AI referral traffic is growing 1% month-over-month, with ChatGPT driving nearly 87% of all AI-based referrals to websites.

GEO: “Earning the Citation” replaces “Gaming the Link”

The shift toward Generative Engine Optimization (GEO) has hit a tipping point. New strategies emphasize “Modular Content Design”—writing in self-contained, high-density paragraphs that AI can easily “clip” and cite.

• The Strategy: For B2B and tech brands, GEO is now driven by earned media. AI engines are prioritizing journalistic sources (25% of citations) and recent content (published within the last 7 days) to ground their answers.

AEO & GEO: Multi-Modal Optimization

A major 2026 guide on Answer Engine Optimization (AEO) highlights that AI assistants are no longer just reading text; they are “viewing” content. Optimization now requires Visual AEO, where images, feature comparison tables, and instructional diagrams are structured specifically for AI “vision” to extract and present as the definitive answer.

• Trend: The “Answer-First” format (placing a concise 40-60 word answer at the very top of a page) is now the required standard for capturing the “Position Zero” voice response.

  

The News: Google officially made Gemini 3 the global default for AI Overviews and introduced Gemini 3.1 Flash-Lite. They also rolled out “AI Mode,” allowing users to jump directly from a search summary into a continuous back-and-forth conversation.

• AIO Impact: AI Overviews (AIO) are now more conversational. The “rank + click” model is being replaced by “cite + converse,” where being the source of a follow-up answer is as valuable as the initial summary.

2. Apple Unleashes M5 Chips with “Neural Accelerators”

The News: Apple announced the MacBook Pro with M5 Pro and M5 Max chips, featuring a dedicated Neural Accelerator in every GPU core.

• GEO/AIO Relevance: This hardware push signals a shift toward Local GEO. As more AI processing happens on-device rather than in the cloud, brands will need to optimize for “Edge AI” models that prioritize local data and privacy-compliant content structures.

3. Rise of “Agentic Commerce” & LLM-first Content

The News: 25% of content marketers now identifying LLMs as their primary audience, the focus has shifted to “Agentic SEO.”

• AEO/GEO Strategy: Tactics are moving away from keywords toward Answer Blocks and llms.txt files—a new (though experimental) standard used to tell AI agents exactly how to parse and recommend your products during a “buying agent” session.

Watch Video

When search engine optimization first got going, it wasn’t hard if you put the work in.

If you published high quality content, Google recognized it. Like we did for GE, Credit Suisse, Dell, Pfizer and many others.

Some of our work is still sticking to the side of search, like a barnacle on a ship.

Try a search for How Loud is a Wind Turbine. The infographic you see. We created it for GE more than a decade ago.

Back then, media outlets were hungry for content they could monetize via ads. Business Insider even syndicated our investment insights from Credit Suisse.

It was the golden age of SEO.

But like all good things, it came to an end, and it became much harder to rank in search queries. Both because Google changed its algorithm again and again and because new media became more important. Namely, video and social media.

Suddenly, you needed paid support to get noticed. Very convenient for Google and Meta.

We’ve been stuck in this increasingly inefficient and expensive loop until ChatGPT burst onto the scene.

Information discovery changed, and we’re now in the golden moment for AIO.

If you follow the laws of AIO, if you know your objective, purpose, mission, if you know and analyze your queries, you have the opportunity of a lifetime to influence the AI answers that matter to you.

But this window will close.

Not tomorrow. Probably not in 2026. But the algorithms behind ChatGPT, Perplexity, Google AI Overviews, they will change.

They will have to.

Right now, even subscribers, paying customers, are being subsidized.

A GPT pro account costs $200/month and…Sam Altman confessed to TechCrunch it’s still a money loser.

The LLMs will have to make money, and when they do, they are going to integrate advertising, and they are going to stop ingesting content at the same pace and in the same way.

When that time comes, if you’re not ALREADY in the right answers in the right way, it may get as hard to break into the chatbot responses…as it is to land the first page in Google search.

Now is the time to accept the information discovery has changed and move fast.

In the next law, we tell you how to get started.

At the Mobile World Congress (MWC) in Barcelona, Samsung and Lenovo unveiled the next generation of “Agentic AI” hardware. Samsung’s Galaxy S26 showcase highlighted “Proactive Intelligence,” where the OS anticipates user needs rather than waiting for prompts. For marketers, this signals a shift toward Assistive Agent Optimization (AAO), where the goal is to be the preferred service provider for an autonomous AI agent making a booking or purchase on a user’s behalf.

Google AIO Expansion & “Citation Volatility”

New data released yesterday shows Google’s AI Overviews (AIO) now trigger in approximately 25% to 50% of searches across high-impact industries like Healthcare and Finance. However, a “citation drop” was noted where AI results are becoming more synthesized, often citing fewer external links than in 2025. This puts immense pressure on GEO tactics—specifically “citation engineering”—to ensure a brand remains one of the few clickable references.

The “Dark SEO Funnel” and Vanity Metrics

A leading industry report published yesterday warns that traditional “organic traffic” is officially becoming a vanity metric. With AI Referral Traffic growing at 1% month-over-month (dominated 88% by ChatGPT), the focus is shifting to AI Share of Voice. Marketers are being urged to track “Brand Mentions” within LLM responses rather than just SERP positions, as users increasingly receive full answers without ever visiting a website.

Consumer Adoption: The 50% Milestone

New research from the SSRS Opinion Panel confirms that as of March 2026, 52% of Americans now use AI platforms weekly. This mainstream saturation means that AEO (optimizing for voice and conversational queries) is no longer a niche tactic but a requirement, as “natural language” searches are now 2–3x longer and more specific than the keyword-based searches of the past decade.

Watch Video

Dependency on LLMs is Death

One of the most useful ways to think about where AI startups are headed is to go backward for a moment.

I remember clearly when Dropbox decided to move off third-party cloud infrastructure and start running its own servers. At the time, a lot of very smart people in Silicon Valley said Dropbox was making a mistake. Their argument was that Dropbox wasn’t really a storage company; it was a UX company. Storage was a commodity. The interface was the value.That argument never made sense to me.

Files still have to live somewhere. And if that “somewhere” is owned by a third party, like Amazon Web Services, then that third party controls pricing, terms, access, and ultimately your margins. They can change the rules whenever they want and even compete with you down the road.

Dropbox understood that early. Storage wasn’t a feature. It was core to the product. And that decision is a big reason Dropbox is still around despite pressure from Google Drive and Microsoft OneDrive. They weren’t just defending a UI. They were defending control.

That same lesson is repeating itself now, just at a different layer of the stack.

The LLM Platform Trap

I think in 2026 we’re going to see a lot of AI-first startups struggle, not because AI demand slows down, but because too many of them are built on top of platforms they don’t control.

Take a hypothetical construction software company. General contractors love it. It automates billing, compliance, and change orders. Real value. But if 90 percent of that value is coming from calls to OpenAI, Gemini, or another LLM provider, then who actually controls the business?

The LLM provider controls pricing. They control performance. They control output quality. They control whether features stay available. And at some point, they can decide they don’t need your application at all because they can offer the same functionality directly.

You didn’t just build a product. You trained the platform that might replace you.

This is why VCs are now asking a harder question than “is this AI-powered?” They’re asking: if the underlying platform changes its economics or strategy, what happens to you?

Brand and distribution help. Having tens of thousands of users helps. But those things don’t always protect you when the intelligence layer you depend on is owned by someone else.

Flipping the Pyramid

What worked five years ago doesn’t work anymore. Back then, companies built the LLM first and trusted that value would emerge later. That made sense when the model itself was the breakthrough. Today, that playbook is mostly gone.

If you’re building now, you have to flip the pyramid. You build value first. You own a workflow deeply. You control data that’s proprietary and hard to replicate. You become embedded in how someone actually runs their business. Only after that do you start pulling more of the intelligence stack in-house.

Yes, that eventually means training or owning more of your own models. That’s expensive. But capital follows value. If you’ve proven that you control something essential, investors will fund the infrastructure required to defend it. Anything else is just renting intelligence.

The AI Layer

This brings me to something I’m hearing more and more in conversations with operators and advisors: the idea of owning your AI layer.

For years, companies outsourced core systems to platforms like Salesforce or SAP. Those were systems of record. They stored data. They ran workflows. That was fine. AI is different.

Once AI starts influencing pricing, underwriting, customer interactions, forecasting, or operations, it stops being a passive system and starts becoming the way your company thinks. That’s not something you want living entirely inside someone else’s platform.

You can use an LLM provider for compute and modeling power. That’s infrastructure. That should be swappable. What matters is that the intelligence layer—the workflows, logic, prompts, proprietary data, and decision frameworks, lives inside your organization. You should be able to change LLM providers without rewriting your business.

If switching models breaks your core workflows, then you don’t own your AI layer. Someone else does. This is the same realization Dropbox had years ago. Storage wasn’t just a backend detail. It was mission-critical. So they owned it.

AI is moving into that same category now. Not as a feature or plugin. But as core infrastructure. And the companies that survive the next wave won’t be the ones with the flashiest demos. They’ll be the ones that decided early what they had to own and built their businesses around defending it.

Search engine tracking tools report that Google search ranking volatility remained elevated over the weekend and into early March. There was still chatter in the SEO community about shifts in rankings and fluctuations across niches.

What’s behind it?
Volatility usually means the algorithm is still adjusting, or signals like quality evaluation and relevance weighting are shifting — potentially tied to recent updates.

So, if your client or the company you work is raising noise about SEO flux, just make sure to remind them that Google is probably doing a significant SEO update that is also taking into consideration AIO rankings.

February SEO Recap Shows Core Trends

Recent digest summaries highlight broader SEO forces shaping search this month:

• AI bots as discovery players — Web discovery is increasingly influenced by AI crawling and bots that behave like search systems.

• E-E-A-T and AI evaluation signals remain critical after the recent core update, with authenticity and expertise weighted heavily.

Enterprise AI Optimization Strategy Expands

OpenAI continued expanding enterprise AI deployments in secure environments, reinforcing:

• AI models are increasingly embedded into business workflows.

• Organizations are building internal AI assistants trained on proprietary content.

This shift has AIO implications: Companies must optimize internal documentation for AIO retrieval.

In a direct challenge to Google Search Console, Bing Webmaster Tools has introduced “AI Performance Insights.” This tool allows webmasters to see exactly which parts of their content are being cited in AI-generated summaries and provides a “mention rate” metric.

2. Google DeepMind released Nano Banana 2

Gemini 3.1 Flash Image. This model is designed for high-speed, high-fidelity image generation and editing. This further pushes the innovation Google is doing in creative arts and continues to challenge work done by humans vs AI. In my opinion, we will reach an inflection point this year where AI generated creative assets will outperform human generated content. Resulting in significant disruption for the industry.

3. AIO Strategy Tip of the Day

Don’t write for the click; write for the citation. With 44% of AI citations coming from the first 30% of a text, ensure your most valuable “answer” content is at the very top of your pages to win the AIO game.

1 - Profound raised $96m on a $1.2 billion+ valuation. Read their direct statement here. My take - the demand for AIO / SEO tools is exploding as more people realize that LLM traffic is overtaking search traffic from Google / Bing.

2 - Bing goes all in on AI / Chat GPT Integration. Similar to the 1990s browser wars, Microsoft is embedding Copilot / Chat GPT into everything (Edge, Office 365, Windows, etc). Don’t be surprised if Microsoft becomes the market leader in AI search volume just by it’s distribution power alone.

3 - 2.5 billion prompts are happening daily on Chat GPT alone. Yet, 96% less clicks happen from an LLM to a reference website than a Google search. Show’s the scale of LLM growth, but also the competitive challenge brands Fighting for traffic from the LLMs makes AIO a critical element for any marketing strategy.

This week The Information highlighted a sharp selloff in enterprise software after new AI tools — including Anthropic’s automation products — spooked investors. Names like Salesforce, ServiceNow, Snowflake, and MongoDB dropped as much as 15% as the narrative took hold: if AI can build software and automate workflows, why keep paying millions for legacy SaaS?

Having lived through the dot-com era and the mobile transition, I’ve learned that moments like this rarely kill all incumbents. They expose which ones adapt fast enough to be even stronger.

Where the fear comes from

There are really two pressures hitting at once..

On one side, competition just got cheaper. AI compresses build time. Teams that once needed 1,000 engineers might need 50. Startups can ship credible alternatives without years of roadmap or heavy infrastructure. The old “complexity moat” shrinks fast.

On the other side, customers start questioning what they even need. If internal teams can spin up tools with AI, why keep every expensive contract? Why not build something yourself?

So incumbent SaaS players feel squeezed from both directions:

• AI-native startups moving faster

• Customers experimenting in-house

That’s enough to compress multiples. The market reaction isn’t irrational. But it’s also not new.

We’ve seen this pattern before

Every platform shift starts with “the incumbents are dead.” History usually proves otherwise. Let me give you two personal examples I lived through

Microsoft and the browser wars

When Netscape took off, it looked like the browser would commoditize Windows and Office. Microsoft didn’t fight the internet. They went internet-first and shipped Internet Explorer everywhere, using distribution as leverage.

As Jack Welch put it:

“There’s no advantage to being a big company if you can’t use your bigness as an advantage.”

Microsoft used its bigness and won. I recall this time fondly at Microsoft. We wanted to win and everyone felt energized to beat Netscape.

You know who did not adopt fast enough, incumbents like IBM. Facebook and mobile

Facebook faced the same moment when attention moved to phones. Desktop was dying. Mobile-first startups were native. Companies like SNAP were a real threat.

Zuckerberg reset the company overnight: everything mobile-first. If it didn’t work on mobile, it didn’t ship.

Then they doubled down: rebuilt the app, fixed ads, bought Instagram and WhatsApp.

Mobile didn’t shrink Facebook. It made it bigger.

In both cases, the winners didn’t defend the old model. They reorganized around the new one. Companies like MySpace, Friendster and Bebo failed to adapt to this new model of mobile first and they slowly died away.

AI is that kind of shift

AI isn’t just another feature. It’s a new interface layer. It changes how software gets built and customized. But it doesn’t eliminate systems of record.

Enterprises still want reliability, compliance, integrations, and support. They still want someone accountable when things break. A clever AI tool doesn’t replace that overnight.

If anything, it makes the core platforms more valuable.

The winners will go AI-first

The companies that survive won’t protect legacy contracts. They’ll do what Microsoft and Facebook did: commit fully.

They’ll say: we are AI-first.

Take Salesforce. If they embrace AI deeply, the product gets better, not worse: faster implementations, easier customization, automated workflows

Plus everything enterprises already trust: scale, security, support.

That combination is hard for a small startup to replicate.

The same logic applies to ServiceNow, Workday, and any platform already sitting at the center of mission-critical workflows. AI becomes an accelerant, not a replacement.

Systems of record still win

My bet is simple.

The winners of this cycle will be the companies that already own the data and the workflows — the systems of record — and then aggressively layer AI across everything.

Not the ones fighting the shift. The ones embracing it.

Every cycle starts with panic. Most end with the adaptable getting bigger. AI doesn’t look like the end of enterprise software. It looks like its next upgrade.

Share

We all have an intuitive sense of how to write for people. Whether it’s a scientific seminar, a business presentation or a news article, there needs to be a hook to draw you in. Then a beginning, a middle and end.

You should still and must write that way for people.

But per law four, machines don’t read the way we do.

So should you write for these machines the way you write for humans?

The answer seems pretty clear—hell no!

When you write for the bots, you aren’t trying to launch substack and build a subscriber base.

Your goal should be a lot simpler.

Modify existing new content or create more to:

–Directly answer questions.

–Demonstrate expertise through meaningful details and well thought out perspectives.

–Provide deep content to answer likely follow up queries.

–Separate positive and negative ideas to avoid confusing LLMs, e.g., on the one hand and on the other comparisons.

–Publish regularly, demonstrating relevance and engagement

If you take these steps, your content will be bot-friendly and vastly increase the chances you end up in generative engines’ answers.

There is, of course, a catch.

You need to know which questions to answer. Unfortunately, unlike google search data, the chatbots are black boxes…they don’t offer any data on which questions are asked most frequently…

In our next law, we explain how to get around the lack of data and determine the right questions to ask and answer.

Leave a comment

I’ve Been Building on WordPress Since the Beginning

When it comes to WordPress, I’m an OG. Right from the beginning, back in the mid-2000s, I started building WordPress sites and have been doing so ever since. I’ve used WordPress for early-stage startups just trying to get something live. I’ve built large e-commerce sites doing millions of dollars a month using WooCommerce / Wordpress. I’ve also worked with Fortune 500 brands where WordPress quietly powered massive content platforms behind the scenes.

So when I hear people confidently say that AI and vibe coding are about to kill WordPress, I pay attention—but I also feel pretty comfortable saying they’re missing something important.

WordPress has lived through a lot of “this will kill it” moments already.

WordPress Was the Original Startup OS

Back in the early 2010s, many iconic companies began their lives on Wordpress. Groupon is one of the best examples - they launched the deal platform on Wordpress and were able to get significant traction. Founders weren’t optimizing for architectural purity. They were optimizing for speed, distribution, and the ability to publish and test ideas quickly.

Heck, when the Brand Project venture team helped launched Freshly, we built it on Wordpress. We scaled it to a significant level before the growth justified a custom tech stack.

WordPress made this possible because it solved the unsexy but critical problems: publishing workflows, user permissions, extensibility, and the ability to bolt on new functionality without rebuilding everything from scratch.

That value hasn’t gone away.

The Vibe Coding Argument (and Why It Sounds Right)

The modern argument is straightforward. With AI and vibe coding, you can prompt your way into a CMS. You don’t need WordPress anymore. You can generate pages, layouts, content, and logic dynamically.

On the surface, this makes a lot of sense. I’ve experimented with this myself. I’ve worked with engineers who suggested skipping WordPress entirely and just vibe coding everything once the designs were ready. And for the first part of the project, it feels incredible. You move fast. You see results immediately. It feels like the future.

But then you hit reality.

The Last 10% Is Where Systems Live

Once you move past static pages, you’re not avoiding complexity—you’re recreating it. Content approvals. User roles. Integrations. Edge cases. Maintenance. Updates. Suddenly, you’re building your own CMS, whether you intended to or not.

Vibe coding gets you about 90% of the way there. The remaining 10% is where production systems actually live. That’s the part where I’ve found myself repeatedly re-prompting AI to fix something subtle, only to realize it doesn’t understand the broader system context. Eventually, you open the code and fix it yourself. At that point, you’re back in the world of custom software—except now you’re maintaining a stack that only partially exists in prompts.

That’s not something vibe coding magically replaces. And the fact that WordPress powers roughly 46% of all websites on the internet isn’t an accident. It’s the result of years of battle-tested reliability, ecosystem depth, and trust at massive scale.

AI Solves WordPress’s Hardest Problem

Here’s where AI genuinely changes the game in WordPress’s favor.

The hardest part of WordPress was never installation. It was everything after that. Building dozens of pages. Designing headers, footers, and navigation. Translating brand guidelines into real layouts. Filling a site with meaningful content instead of placeholders. That’s where projects slow down, budgets get blown, and momentum dies.

AI is exceptional at this layer.

If AI can generate page structures, layouts, copy, and design elements inside WordPress, then WordPress doesn’t get replaced—it gets amplified. The biggest bottleneck in adoption disappears. Suddenly, people aren’t asking whether WordPress is worth the effort. They’re asking why they’d build on anything else.

The Bottom Line

Is AI going to kill WordPress? No.

AI is removing the friction that used to hold WordPress back. WordPress becomes the stable operating system—the system of record—while AI becomes the interface that helps you build faster, cleaner, and with fewer mistakes.

People love to say AI is the end of software, the end of SaaS, the end of platforms.

This isn’t the end of WordPress.

It’s what happens when a platform that already won distribution, trust, and ecosystem suddenly gets supercharged. And in my experience, that’s usually the beginning of a much longer run—not the end.

Share

For nearly three decades, Google Search was one of the greatest business models ever built—not because it monetized everything, but because it didn’t have to.

Most people don’t realize this, but roughly 80% of all Google searches never showed an ad. Only about 20% of searches had any commercial intent at all. And even when ads were shown, the overall click-through rate hovered around 3%. That means Google only made money on a tiny fraction of total searches.

In almost any other business, that math would be terrifying. For Google, it was magic.

Why? Because the cost of serving a search was absurdly low. A traditional Google search—pre-AI—was estimated to cost fractions of a penny. Google wasn’t generating intelligence in real time. It was returning ranked links from an already-indexed corpus, using an algorithm it had amortized over decades. The marginal cost was so close to zero that it barely mattered whether a search monetized or not.

That’s the first tail. And Google was ruthless about protecting it.

In the early days, Google made two very deliberate choices that most people forget. First, it showed only ten blue links—no images, no rich media, no fluff. Why? Speed and bandwidth efficiency. Every extra asset slowed page load and increased cost. Google optimized for instant answers over visual richness long before “page speed” was fashionable.

Second, Google didn’t rely on off-the-shelf infrastructure. It built custom servers to power crawling, indexing, and search results. Not because it was sexy—but because it let them squeeze every dollar of efficiency out of the system. This wasn’t just good engineering; it was economic warfare. Google understood that if you control cost at massive scale, you can afford an enormous non-monetized long tail.

That’s why the model worked.

Google could support billions of free searches because the tail was cheap. And when monetization did happen—insurance, legal, auto, travel—it was extraordinarily valuable. A few dollars here, ten or fifteen dollars there, multiplied across scale. Low cost, rare but lucrative monetization, infinite leverage.

Now enter AI.

When Google introduced AI Overviews, everything changed. Suddenly, search wasn’t just retrieval—it was generation. Every query required real-time intelligence powered by Gemini, layered with context, safety, and reasoning. That shift didn’t just change UX. It blew up the cost structure.

It’s widely estimated that a search with an AI Overview costs 5–10× more than a traditional Google search. At the same time, Google is knowingly cannibalizing some of its own ad real estate by placing AI answers above commercial links.

That tells you something important.

Google is not doing this casually. It’s doing it defensively. Users are being trained—by LLMs—to want answers, not links. And Google knows that if it remains just a link engine while others become answer engines, it loses relevance. So it’s accepting higher compute costs and short-term revenue pressure to protect long-term dominance.

Which brings us to the second tail.

LLMs live in a fundamentally different economic world.

Unlike classic search, every LLM interaction has real, material cost. A short answer is cheap. Anything meaningful—long reasoning, file uploads, document summaries, images, multi-step agent workflows—gets expensive fast. Compared to classic Google search, LLM queries are often 10× to 100× more expensive per interaction.

There is no free long tail here. The long tail of Google is dead!

Every question costs money. Every follow-up compounds it. Intelligence isn’t pre-indexed—it’s generated on demand, every time.

That’s the second tail.

And while companies like Google and Microsoft can afford to live here—at least for now—most others cannot. What’s interesting is that they’re already showing you their hand.

The smart LLM companies don’t want to be everything to everyone. They don’t want to subsidize consumer search, image generation, and casual curiosity forever. Instead, they’re choosing which tail to dominate.

Look at what’s happening in practice. Claude launched Claude Code. Then Claude Co-Work. That’s not accidental. It’s a signal. Anthropic is telling the market exactly who they’re for: developers building applications and professionals optimizing work. That’s a tail where intelligence directly translates to dollars, where users will pay premium pricing, and where efficiency compounds into productivity gains.

That’s a much better business than trying to be a universal answer engine for free consumer queries.

By focusing on work, Claude can extract higher ARPU, justify higher compute spend, and avoid bleeding money on low-value interactions. It’s a lane—and it’s a smart one.

This is the future.

You’re going to see more LLMs pick a lane. Some will own enterprise workflows. Some will dominate developer tooling. Some will specialize in creative production. Others will focus on vertical intelligence. Very few will try to be universal—and even fewer will survive if they do.

So yes, we are living through a tale of two tails.

The first tail was built on links, custom hardware, ruthless efficiency, and near-zero marginal cost. The second tail is built on choosing a lane, focused use cases and compute that must be justified on every interaction.

Share