Search Just Became an Answer Engine. Here's What Solo Knowledge Entrepreneurs Should Do in the Next Thirty Days.

At I/O 2026 this week, Google announced what it is calling the biggest change to its search box in over 25 years. The default output of Google Search is no longer a list of blue links. It is a conversational answer, generated on the spot by Gemini 3.5 Flash, with the option to keep asking follow-up questions inside the same workspace.

One statistic from the early studies tells you everything you need to know about what just happened: when users are shown an AI-generated answer, they click through to the original websites only about 8% of the time. (That figure comes from research on Google's AI Overviews, deployed since 2023; the new AI Mode default is expected to match or exceed it.)

That is not a traffic dip. That is a structural change to the way the internet hands attention to the people who create knowledge for a living.

If you run a one-person knowledge business — coaching, courses, teaching, writing — where the model is "people find me through search, then I educate them, then I sell to them," the front end of that funnel is now exposed to losing up to 92% of its volume.

And there is a quieter problem underneath the traffic one. The AI now answering most of those queries is being trained on a vast volume of content produced, increasingly, by other AIs. The result is what I have come to call the Dead Sea Effect — content that is technically competent, broadly distributed, and experientially empty. It looks correct. It reads fluently. And it leaves no mark on the reader, because no human ever truly authored it. The search system is becoming a closed loop of machine-generated answers summarising machine-generated articles. The people most positioned to break that loop are the solo operators who still have lived authority to put back into their work.

I spent the morning after the announcement working through the implications with three AI systems — Claude, Gemini, and ChatGPT. The plan that emerged came from synthesising what each one saw and what each one missed. Here is what I am doing, and what I think anyone in a similar position should consider.

What Actually Changed in Google Search at I/O 2026

The technical announcement is dense, but the practical change is simple. Google Search has stopped being a directory of websites that compete for ranking. It has become an answer engine that consumes those websites as raw material and serves the answer directly to the user.

In the old model, the user typed a question, Google returned a ranked list, and the user clicked through to a website to read the answer. In the new model, the user types a question, Google reads the relevant websites silently, and returns the answer directly inside the search interface. The user never has to leave Google to get what they came for.

The cascading effects are the part most people are missing.

The first cascade is that informational content collapses. Articles written to rank for queries like "how to improve focus," "what is the best journaling habit," "how to read more books in a year" — these were already in a long, slow decline. They have now fallen off a cliff. The AI summary will absorb them and render them invisible in a single query.

The second cascade is that transactional and branded search is largely protected. When somebody searches for your specific name, your specific course, your specific product, the AI is not going to invent a summary that replaces a purchase decision. It will route the user to you. So if a visitor already knows you exist, the search system still works. If they do not yet know you exist, the search system is no longer the discovery mechanism it was last week.

The third cascade is the part nobody is saying out loud yet. The kind of business that gets devastated by this shift is the high-volume, thin-authority content operation — affiliate-heavy blogs, generic productivity sites, listicle factories, anywhere the model is "rank for many keywords, monetise the trickle." The kind of business that gets strengthened by this shift is the one with named methodologies, proprietary frameworks, identifiable expert authorship, and a layered product ladder that does not depend on cold traffic to function.

If you have spent years building a real teaching system, a real diagnostic instrument, a real coaching practice with documented outcomes — this shift is not the threat to your business that the headlines suggest. It is the moment your accumulated work becomes disproportionately valuable. The market is about to viscerally feel the difference between content that exists and authority that has been earned.

What Survives in the AI Answer Engine Era

Three things survive intact, and one new thing emerges.

The first survivor is owned channels. Your email list, your newsletter, your private community, your podcast audience — anything where the relationship with the human is direct and not mediated by a search engine — becomes structurally more valuable. If most cold organic search traffic is disappearing, every email subscriber is worth a multiple of what they were worth a month ago. The work you have already done to build an audience is suddenly the most important asset you own.

The second survivor is named, proprietary methodology. AI answer engines preferentially cite specific, named systems with clear authorship over generic concepts. If you teach "productivity," you are competing with a thousand summaries. If you teach a specific named system — your framework, your protocol, your method, your diagnostic instrument — the AI has to attribute the term to you when it appears in answers, because nobody else owns that vocabulary.

My own example. I have spent 21 years running Speed Reading Simplified, and the methodological frame I teach inside that business is what I call High-Performance Reading — HPR. Generic "speed reading" content gets summarised away. HPR gets cited, because it is a specific named system with specific named components and specific authorship. That is the difference. And it is available to any solo knowledge entrepreneur willing to do the work of naming, defining, and documenting what they actually teach.

The third survivor is depth of lived authority. Generic informational content is exactly what the AI was trained on. It cannot produce something genuinely new from material that already exists everywhere. But it cannot replicate specific lived teaching experience, specific student transformation stories, specific case examples from your actual practice. The more your content is dense with the verifiable particulars only you can write, the less the AI can substitute for you.

The new thing that emerges alongside all three is what some people are starting to call generative engine optimisation — the discipline of structuring content so AI systems cite it as a source rather than absorbing it anonymously. This is different from traditional SEO. The audience you are writing for is no longer just human readers and search algorithms. It now includes the AI systems that summarise content for those readers. The structural choices you make in how you organise and mark up your site directly affect whether the AI treats your work as a referenceable source or as undifferentiated training data.

The Reframe: From Traffic-First to Authority-First

The single most useful mental shift I have made is this: my website is no longer primarily a discovery tool. It is a credentialing and conversion tool.

In the old model, the website's job was to attract cold visitors via search, warm them through educational content, and eventually convert them into customers. In the new model, the website's job is to be the authoritative destination for people who have already heard of me through other channels — newsletter, word of mouth, podcast mention, social share — and need to verify quickly that I am credible, that my methodology is real, and that the next step is worth taking.

Almost every page on a knowledge business website should now be evaluated against that new job description. Does the page do credentialing work? Does it stake a position only I can stake? Does it surface specific evidence — student outcomes, methodology details, lived case examples — that no AI could fabricate? Does it convert efficiently for the warmer visitor who arrived from a deliberate channel rather than stumbling in from a cold search?

If a page on your site does not pass those questions, it is doing the old job in a world where the old job no longer exists.

The 30-Day Action Plan for Knowledge Entrepreneurs

I have organised this into three tiers. Tier 1 is what I am doing in the next thirty days. Tier 2 is what I am doing in days 30 to 60. Tier 3 is what I am doing in days 60 to 90. If you only have bandwidth for the first tier, that is fine — the first tier alone puts you significantly ahead of most of your competitive field.

The temptation in a fast-moving news cycle is to chase every tactic at once. Resist it. You are one person. The plan only works if you can actually execute it.

One reassurance before we get into the specifics. The Tier 1 work touches schema markup, entity disambiguation, and structured data — terms that sound like they belong to a specialist trade. They do not. You do not need to become a technical SEO specialist to do this competently. You need a few hours, a willingness to copy and adapt structured templates, and a course platform that lets you inject custom code blocks. That is the entire skill stack.

Tier 1 — The Next 30 Days (Schema & Canonical Content)

Deploy structured schema markup on your site. Schema.org is a 15-year-old standard, Google actively uses it, and AI answer engines demonstrably parse it. This is not speculative. Every hour spent on schema is an hour spent on infrastructure that is certain to be read by the systems now driving discovery. The specific schema types that matter for a knowledge business are Organization (your brand as an entity), Person (you as a named expert with authority signals), Course (each of your offerings, with prices and outcomes), Book (any published or forthcoming titles), and FAQ (for high-conversion landing pages).

If your site is on a hosted course platform like Kajabi, you can usually inject this via a Custom Code block in the site header. The schema does not need to be visible to human visitors. It needs to be readable by the AI systems that now summarise your work.

Audit your brand for entity confusion. This is the unglamorous one, and most people will skip it. When an AI system encounters your brand name, your acronym, your personal name, does it know who you are? If your brand uses an acronym that collides with a well-known generic term, the AI will route users away from you to the generic meaning. If you use both a professional name and a personal name in different contexts, the AI may treat them as two separate entities and split your authority signals across both.

The fix is mechanical. In your schema markup, use the alternateName property to explicitly map every name variation to a single entity. Use the sameAs property to link your various profiles, social accounts, author pages, and platforms to that same entity. The goal is that every AI system encountering your work treats it as one connected body of authority rather than as fragments.

Write the single most important canonical page on your site. Pick the one named methodology, framework, or system that is the conceptual anchor of your entire teaching. Write it as a definitional reference page — not a marketing page. Define the term. Explain the components. Show the structure. Cite the lived authority that produced it. This page is what the AI will reach for when somebody asks what your methodology is, and the way it is structured directly determines whether you become the cited source.

That is it for the first thirty days. Three concrete deliverables. Each one moves you measurably forward, and each one would be valuable even if you stopped there.

Tier 2 — Days 30 to 60 (Course Schema & Conversions)

Add Course and Book schema to your specific offerings. Each course, each book, each diagnostic, each programme should have its own structured markup describing what it is, what it costs, who it is for, and what outcome it produces. This makes your specific products legible to AI systems that are increasingly being asked "what is the best course on X" or "where can I learn Y from someone who actually teaches it."

Upgrade your highest-converting landing page with FAQ schema and a tighter conversion path. AI answer engines love FAQ structures because they map naturally onto the question-and-answer format the AI is generating. A well-structured FAQ block on a key page does double work: it serves the human visitor and it gets directly cited in AI summaries.

Publish one position-staking article in your voice. Not an informational piece — those are exactly what the AI will absorb and bury. A position-staking piece is one only you could write: a claim, an argument, a frame, a lived observation that draws on specific authority and contains something that does not already exist in generic form on the open web. This kind of writing builds authority with the humans who land on your site through non-search channels, and it gets cited by AI systems precisely because the material is new.

Tier 3 — Days 60 to 90 (Methodology & llms.txt)

Build a second canonical methodology page. If the first one anchored your central framework, the second one should anchor the most important supporting system — your diagnostic, your protocol, your sequence, whatever the second-most-important named asset in your business is. Two canonical reference pages, well-structured, semantically connected, with cross-linked schema, is a moat. Most of your competitors will never build even one.

Reposition your email capture as the front-of-site offer. If your newsletter or lead magnet currently sits in a sidebar or footer as an afterthought, it needs to be promoted to the central call to action on the highest-traffic pages. The newsletter is now your most important owned-channel asset. The page architecture should reflect that.

Consider an llms.txt file. This one needs an honest caveat. The llms.txt standard — a plain-text Markdown file at your site root that tells AI systems what your site contains in their preferred format — is an emerging proposal, not yet a confirmed protocol. No major AI crawler has formally committed to honouring it. But the marginal effort to deploy one is small, the potential payoff if the standard takes hold is large, and the act of writing it forces you to articulate your site's value proposition in machine-readable form, which is a useful exercise regardless of whether the file itself ever gets read by a crawler. Worth doing as a low-cost hedge. Not worth elaborate engineering to chase.

The Dead Sea Effect: When AI Writes the Answers AI Reads

There is a second-order effect of all this that the SEO commentary is mostly missing.

The Google announcement does not just change how readers find content. It changes who is writing it. Every solo operator I know who has experimented with AI-assisted writing in the last two years has had a version of the same experience. The AI produces something. It is fluent. It is competent. It is publishable. And it is somehow inert — readable but unmemorable, accurate but unfelt, technically defensible but missing the thing that makes you trust the person on the other side of the words.

This is the Dead Sea Effect. And it is about to get much worse, because the AI that summarises content for Google's new search experience is increasingly summarising content that other AIs produced. The summaries of summaries of summaries flatten everything that made the source material worth reading in the first place. Two years from now, the open web will be flooded with content that is structurally optimised for AI consumption, written by AI, summarised by AI, and consumed by humans who can no longer tell why the answers feel so empty.

The defence against this is not better prompts. The defence is the deliberate insertion of human experience into content production at the point where it cannot be delegated. I have been building a framework around exactly that problem. It is called the Authored Intelligence Framework — AuI. The philosophical centre of it is what I call the Experience Firewall: before any AI assists with anything, the author writes 200 to 300 words of Anchor Truth from lived experience — a specific case, a specific failure, a specific observation that only the human author could have produced. Everything the AI does afterward is amplification of that signal, not substitution for it, and this is just one of twelve levels of my framework.

Try this on your next article. Before you open a chat window with any AI tool, write three hundred words about something that actually happened to you, in your work, related to the topic at hand. Not a hypothetical. Not a position. A specific moment, in specific detail, that nobody else could have written. Then bring that text into the AI session as the anchor for whatever it helps you produce. The output will feel different. So will the writing of it. That is the Experience Firewall in operation, and once you have felt the difference once, you cannot un-feel it.

I want to share one specific moment from this week that illustrates why the Firewall matters. I was talking with Google's AI Search about this change, and I used my brand acronym — SRS, short for Speed Reading Simplified — without explaining what it stood for. I wanted to see if the AI was connected enough to Gemini's broader knowledge about me and my work to recognise the brand. It was not. It interpreted SRS as Spaced Repetition System and built an entire response on the wrong foundation, never asking if this was the correct understanding, and continuing through follow-up questions without ever surfacing its initial misinterpretation. By the time I stopped it, and finally let it know what SRS stood for, I had spent ten minutes reading and deciphering Google's response on an answer to a question I had never meant to ask.

That ten minutes is the cost of an AI session without an Experience Firewall at the front. The Firewall would have forced me to write down what I was actually doing, what SRS specifically meant in this context, and what the question I was actually asking was — before any AI engaged with any of it. The skill that protects you against an AI building entire conversations on the wrong foundation is the skill of being precise about your own situation at the front of the work. Anchor Truth before amplification. Signal before tool.

A Final Word About Being One Person

The full plan I have just described is real work. Schema markup, canonical pages, entity disambiguation, position-staking content, owned-channel reinforcement — and you are also writing the book, running the coaching, building the courses, doing the marketing, and operating the business that pays for all of this.

So I want to be honest about the constraint.

You cannot do everything in the plan in the next thirty days. You can do three things. The three Tier 1 items I described are the highest-leverage 20% of the full strategy, and they are enough on their own to put you ahead of almost every competitor still building for the search engine that existed last week.

I have been running Speed Reading Simplified as a one-person operation for 21 years. I know what this is. The patient, deliberate, brick-by-brick way of working is exactly what this moment rewards. The competitors who panic will spread themselves across forty tactics and execute none of them well. The competitors who treat this as a structural change requiring measured infrastructure work will compound their authority steadily and quietly, and in twelve months the gap will be visible.

The plan is not "do everything." The plan is "do the highest-leverage three things, then the next three, then the next three." Keep moving. Do not chase headlines. Build the machine-readable layer of your existing authority. The positioning you have already built will carry the rest — and the human signal you keep at the centre of every piece of work that goes out under your name is what makes the positioning real.

You are not behind on this. You are exactly on time.

Where I Am Building The Authored Intelligence (AuI) Framework in Public

If the Dead Sea Effect describes content you have produced and felt uneasy about, the framework that addresses it has a home. CreateGlint is where I am building the Authored Intelligence Framework in public — open methodology, working documents, the live experiments, and the questions I am still figuring out.

Two ways to stay close to the work if you want to.

The newsletter is the broader signal: weekly notes on what the answer engine era is doing to solo knowledge businesses, what the framework is teaching me as I build it, and what is worth doing about both. You can join here: https://www.createglint.com/the-pulse.

The group coaching cohort is the narrower one. Later this year I am opening a small, intentional group of practitioners who want to build AuI into their own workflow under my direct guidance. The cohort will be limited. The framework is twelve stages across four phases, and it is not the kind of thing that scales to a thousand-person course — it scales to a room small enough that I can know every participant's work. The waitlist is the only pre-launch access point: https://www.createglint.com/aui.

Neither replaces the work of building the machine-readable layer of your own authority. That work is yours, and the next thirty days are when it starts.

Before You Go — Four Questions

You just spent time reading about one of the most significant structural changes to hit knowledge businesses in a decade. If you made it this far, you are not someone who skims for reassurance — you are someone who reads to understand what to actually do.

That tells me something about where you might be right now. Maybe you are evaluating how exposed your current setup is. Maybe you already have a named methodology and are thinking about what this means for how you present it. Maybe you are at the beginning and this article clarified something you have been sensing for a while.

I want to know where you actually are.

The four questions below were built using Typeform — the tool I use across all of my sites when I want to have a real conversation with a reader rather than send them a generic survey form. If you have never used Typeform before, you will notice the experience feels different: one question at a time, a conversational pace, no wall of form fields. That is by design, and it is one of the reasons I recommend it to anyone building an audience-based business who wants to genuinely understand what their readers need. You can explore Typeform here — https://typeform.cello.so/eEidFwVRfR5 — and I will mention upfront that if you sign up through that link, I receive a small commission. I believe in being transparent about that rather than pretending the recommendation is purely altruistic.

But the form below is not here to sell you Typeform. It is here because the articles I write next, the resources I build, and the products I consider creating — they all come from answers like the ones you are about to give. Your response goes directly to me. No spam. No sales sequence. Just one solopreneur genuinely trying to understand what would be most useful to the people reading his work.

Four questions. Less than two minutes. And if anything I might build next sounds interesting — you will see it in question three.

This article was authored, not generated. Built under AuI™ — the Authored Intelligence Framework. [See the framework →]