AI Systems
A successful tool call is not a verified claim: where MCP assurance stops
The fetch can succeed while the memo gets the meaning wrong. Check which document arrived, what its table says and how far the evidence carries your decision.
Author and subject-matter expert
Len P. van der Hof
The books
Live titles and systems.
Published work
Recent articles.
SEO, GEO and AEO
AEO meaning: answer engine optimization, not customs
The letters are a craft on one URL. Customs is a different AEO.
SEO, GEO and AEO
Before you buy an AEO tool, ask what its score measures
A page check, an observed citation and a sale are three different events. Make the vendor show which one it measured.
SEO, GEO and AEO
AEO vs GEO vs SEO: three jobs, not three synonyms
If you use the three letters as synonyms, you will buy the wrong work and measure the wrong door.
SEO, GEO and AEO
AEO vs SEO: two jobs, two numbers
You can rank and stay unnamed. You can be quoted from a URL that does not rank.
AI Systems
Agent audit trail: reconstruct the action, approval, and outcome
If the only witness is the agent, you do not have an audit trail. You have a testimony.
AI Systems
Agent runtime budget: keep retries inside the same ceiling
A cap that resets on every retry is a cap per attempt, not per task.
Venture Building
An agent scorecard turns output review into a permission decision
One good-looking brief can hide an unsupported claim. Grade the work against the job it was given, then decide what the role may do next.
AI Systems
Agent shadow mode: evaluate proposals before granting action rights
Let the agent be wrong on paper first, where a mistake costs a comparison instead of a customer.
AI Systems
The Agent Trust Ladder: how an AI role earns autonomy
A scorecard records how the role performed. The ladder decides what it may do next. The action still sets the ceiling.
AI Systems
An agentic workflow is not an automation
A script repeats a path. An agentic workflow holds a goal, uses tools, and needs a ceiling.
AI Systems
AI agent or chatbot: three tests before you mix the words
If it cannot refuse a tool call, it is not an agent you can run. It is a demo that can spend money.
AI Systems
AI Agents for Startup Strategy: Build the Strategic Intelligence Layer
Agents are infrastructure, not toys. This book installs a strategic intelligence layer you can design, govern, and run in ninety days.
AI Systems
AI fake citations: a 15-minute check for the claims that matter
A citation makes two claims: this document exists, and it says this. Check both in the source itself, not by asking the tool.
Venture Building
What does '$3 per million tokens' mean? How to read an AI bill
The input price covers what you send. The answer runs on a separate, usually dearer meter, which can also include reasoning you never see.
SEO, GEO and AEO
Answer engine optimization: the craft, not the acronym
The craft fits in the first HTML screen. Everything after it is written for humans.
SEO, GEO and AEO
How do AI answer engines choose what to cite? A reading list of primary sources only
Twelve documents, read in the original, beat a hundred summaries of them.
SEO, GEO and AEO
If the answer is in paragraph nine, you did not write a page
The first screen is the page. Everything after it is permission to stay.
Founder Performance
Attention leak vs distraction: why putting your phone away is not enough
A muted phone cannot finish an unresolved decision or give your team authority to act. Trace the demand that keeps pulling you back.
Founder Performance
Belief audit vs fact-check: inspect the claims you already act on
An unspoken assumption can steer next quarter's plan without ever being put on the table. Start by writing down what you already lean on.
Venture Building
BooksOS vs a content calendar: what decides publication?
A date reserves a slot. A gate records the standard, the person who checked it and the decision to release or hold. The calendar still has work to do.
SEO, GEO and AEO
Build a citation prompt set: ten questions to test AI source links
A source citation and a brand mention are different results. Your questions and run rules decide what either count can mean.
Founder Performance
What does calibrate mean? Compare, record, then decide: calibration is not adjustment
Read the instrument before you turn the screw. The same rule works on your own judgment.
Founder Performance
Calibration vs Brier score: what each measure misses
A low Brier average can sit beside a 90% band that runs high. A flat 70% forecast can match its hit rate exactly and still say nothing about which cases were strong.
Founder Performance
Calibration techniques for founders: five practices that fit a week
You do not need a nine-skill curriculum to stop treating a loud sentence as a probability.
Venture Building
Case study vs evidence: what one success can tell you
The featured customer gives you a story. The unfeatured attempts are where a success rate would come from.
SEO, GEO and AEO
ChatGPT mentions vs citations: a name is not a source
If you cannot click through to the claim with your name on it, you were mentioned. You were not cited.
AI Systems
CHORUS and Chorus AIDLC: which one do you mean?
One is a way of working you write down. The other is software you install.
Venture Building
Cialdini in the funnel: six levers, one honesty test
Reciprocity is a real sample. A fake crowd is still a fake crowd. Name the lever, then keep it.
SEO, GEO and AEO
How to measure ChatGPT citations: a weekly log, not a screenshot
One good answer proves it happened once. A dated row per question, per run, shows how often.
SEO, GEO and AEO
Citation-readiness audit: eligible is not cited
If you cannot fail the page in fifteen minutes, you do not have an audit. You have a mood.
Venture Building
The cobra story is unverified. Hanoi's rat bounty is documented.
Pay for rat tails and you get rat tails. Whether you get fewer rats is a separate question.
SEO, GEO and AEO
Content architecture that answer engines can lift
One claim per URL, named entities, and structured data a machine can quote without inventing you.
SEO, GEO and AEO
Can your page answer the next question in AI search?
A broad answer gets the reader oriented. The condition, exception or comparison that follows often decides what they do.
SEO, GEO and AEO
A conversion system that stays honest when you automate it
A conversion system is allowed to remember. It is not allowed to invent the reason to buy.
Venture Building
Correction vs quiet edit: who needs to hear what changed?
Updating a page does not update a customer's plans. Follow the earlier claim to the people and decisions it reached.
Venture Building
Corroboration is not a source count
Five links can carry one upstream result. The question is whether distinct paths support the same claim.
AI Systems
Cost per route vs the monthly AI bill: meter where the money goes
A flat monthly bill can hide a route that doubled its spend. Attribute each charge, separate volume from cost per request, and give the route an owner.
Founder Performance
Diagnostic evidence vs fitting evidence: what changes a belief?
A warm reply may be real and still tell you little about a sales target. The useful question is whether campaigns that miss the target show the same sign.
Founder Performance
Do 90% of startups fail? How to fact-check a viral statistic in ten minutes
The 90% figure does exist somewhere. It just measures something other than what the post says.
Venture Building
Do you need a literary agent for nonfiction? What agents do, what they cost, and when to publish yourself
An agent is a door to large publishers. If you are not going through that door, you need a production system instead.
SEO, GEO and AEO
Does the language of your question change who AI cites?
An English-only test leaves your Dutch questions unchecked. Compare the cited pages, not just the language of the answer.
Venture Building
Downstream effects: why the real cost of a decision lands later, and somewhere else
Every choice starts a river. Most teams only watch the first bend. This is the operator definition of downstream effects, and a method you can run this week.
Venture Building
Downstream effects or affects? The one-line rule, and the two exceptions
Affect is what a decision does. Effect is what it leaves behind.
Venture Building
Downstream impact, consequence, implication, outcome, effect: five words, five jobs
Pick the word by the question it answers: what came out, how hard it hit, who has to live with it, what it means next.
AI Systems
Dutch costs 20% more GPT tokens than English. We measured it on 78 article pairs
Same content, same author, two languages. The Dutch version was 2% longer in words and 20% longer in tokens.
Founder Performance
Emotional granularity vs venting: precision before action
Apprehension about a customer, resentment about a missed handoff and relief that a conversation is still possible can all hide inside one word: 'stressed'. Separate them before you choose a response.
AI Systems
Evaluate the reasoning, not the fluency
A fluent answer is not a graded one. Score the trace before you trust the sentence.
Venture Building
Fake urgency: how to tell honest scarcity from a false deadline
Honest scarcity and fake urgency can look identical on the page. The difference shows when the timer reaches zero.
Founder Performance
Familiar vs true: why repeated claims feel right
Three places you heard a claim may all lead back to one person. Trace the route before you count the sources.
Founder Performance
Write the five lines before you act
An entry after the outcome is a memoir. An entry before it is a record.
Founder Performance
Flow is a condition of work, not a brand mood
Boredom is too easy. Anxiety is too hard. Flow sits between, and it has to be built.
Founder Performance
A holiday is not a refill protocol
Infrastructure has a spec, slack, and an owner. A mood after two weeks off is none of those.
Founder Performance
Founder systems you can run when willpower is gone
Willpower is not the instrument. A weekly cadence that does not ask you to want it more is.
AI Systems
Frontier, workhorse, or local: three classes, not a religious war
The forum fight is cloud versus local. The operator question is which class owns this task.
SEO, GEO and AEO
GEO examples: six before-and-after edits that make a page easier for AI to quote
Most GEO is plain editing. Here is what it looks like on one page.
Venture Building
Goodhart's law: when a KPI stops measuring the outcome
A measure tells you how things are going. A target tells people which number to produce. Once a KPI is both, check whether it still tracks the outcome it stands for.
AI Systems
Governed agent memory: what may persist, and who may write it
An agent's memory is a database that the agent writes itself. Give it the rules you would give any other database.
SEO, GEO and AEO
GPTBot vs OAI-SearchBot vs ChatGPT-User: three bots, three jobs
One vendor name hides three decisions: training, Search, and a fetch that a person started.
Venture Building
A go-to-market system that survives the first ten customers
The deck is a prior. The first customers are the reality-test. Write one sourced claim per name.
SEO, GEO and AEO
Hiring an AEO agency? Seven questions to ask before you sign
No agency can guarantee a citation. You can buy the work, and a record of what changed.
AI Systems
HITL in AI workflows: put the named person on the irreversible step
If the model can ship while the reviewer is on a plane, you did not apply HITL. You labelled a spectator.
Venture Building
HOLD is a research result, not an unfinished job
You can finish the handoff without clearing the action. A useful HOLD says what is blocked, who owns the next check, and what would reopen the decision.
SEO, GEO and AEO
How Answer Engines Choose Sources
Citation visibility is not won with an AI hack. It is earned by combining technical eligibility, clear answers, original value and evidence that survives scrutiny.
AI Systems
How many AI agents can one person supervise?
Starting another agent is a one-time decision. Reading its work, checking its handoffs and handling its exceptions come back every week.
SEO, GEO and AEO
How to get cited by ChatGPT
Open the Search door. Write a sentence a machine can steal. Then check if your name appeared.
Founder Performance
How to run a pre-mortem in 20 minutes
Asking "what could go wrong?" turns every doubt into a challenge to the plan. Declaring "it has failed" makes explaining the failure the task. The result is a list of risks, not proof of a better decision.
AI Systems
When nobody approves in time, your AI agent must not act
An expired request records silence. A rejected request records a decision. Both must block the action.
AI Systems
Human in the loop meaning: a named person who can fail the work
If nobody can fail the output against written criteria, the loop is decorative.
AI Systems
Human on the loop: watching is not a fail gate
If the work can ship while the watcher is on the gallery, the loop is already the wrong loop.
SEO, GEO and AEO
Is there an ideal word count for AI citations?
Your page has a length. The passage that answers the question has to make sense without the rest of it.
Founder Performance
Identity fusion vs commitment: how founders tell the difference
A long week proves little. The question is whether a setback stays a business problem and whether other parts of your life still carry weight.
AI Systems
Introducing Undominated.ai: the independent AI inference price index
Best first. Then price. The board that names every model something else beats without a trade-off.
SEO, GEO and AEO
Is SEO dead? No. Answer engines turned it into the entry ticket
The AI answer still shops in the search index. Fewer people walk through the door afterwards.
SEO, GEO and AEO
Do AI crawlers run JavaScript? Check the HTML they receive
Your browser paints the page after its scripts run. A crawler may read only the HTML your server sends first. Check that response before you assume a bot can read your answer.
Venture Building
Kahneman for marketing: System 1 is not a permission slip
Losses weigh more than matching gains in prospect theory. That is a description. It is not a brief to frighten the buyer.
Venture Building
Len or a fractional CAIO: which work do you need?
One role owns AI across the company. The other closes one decision or installs one system. Pick by the size of the problem.
Founder Performance
LEVEL vs CALIBRATE: check the claim, or check your own judgment
One asks whether this claim can carry this decision. The other asks whether your 70% really means 70%.
AI Systems
Literary agent or AI agent: two different jobs
One is a human representative. The other is a worker with a ceiling. They do not replace each other.
AI Systems
LLM integration is an operating-design problem
The scarce work is ownership, permission, and review, not the model name on the invoice.
Venture Building
Marketing nudges are choice architecture, not a growth hack
The fruit at eye height is a decision. So is the pre-ticked box. Name it, or someone else already did.
AI Systems
MCP explained for founders
Care about the permission and the log. The logo on the protocol is the least interesting part.
AI Systems
MCP is not USB-C: where the favourite analogy breaks, and why that matters
The plug is standard. What the model reads through it is not.
AI Systems
MCP meaning: Model Context Protocol, not a product
The full form is the least interesting part. The permission boundary is the operating decision.
AI Systems
MCP OAuth is not tool authorization: a valid token is not a yes
OAuth tells your server the token is genuine and meant for it. Your code decides whether this call may approve this expense.
AI Systems
MCP vs function calling: two layers of the same tool call
In most MCP setups the model still makes a function call. MCP changes where the tool lives, not how the model asks for it.
AI Systems
Deprecation protocol vs surprise sunset: retire AI models on your schedule
A retirement notice gives you an outside deadline. A deprecation protocol names what takes over, what must pass and when you will move.
AI Systems
Model registry vs vendor list: know which models run
A contract can clear a supplier while a production model remains unknown. Compare actual calls with the register before you trust it.
AI Systems
Why multi-agent LLM systems fail: the 14 MAST failure modes
An agent's mistake and the workflow that let it through are two different problems. MAST gives you a vocabulary for both.
Founder Performance
Narrative defense vs founder story: when your own story decides for you
A record of what happened does not settle why it worked or what you should do next. Check each claim before your company story becomes a rule.
SEO, GEO and AEO
OAI-SearchBot allowed but still blocked? Check the path after robots.txt
robots.txt grants permission. Proof of delivery is a verified request and a body that contains the article.
AI Systems
Is one AI model enough? When a single model works, and when you need a portfolio
The number of models is the wrong question. The right one: did anyone write down why each model runs, and what happens when it stops?
Founder Performance
The 90% test: why 'I'm sure' is right far less often than it feels
A range you are 90% sure of should miss about once in ten. In one experiment it missed nearly half the time.
AI Systems
What an agent may never do
HITL is who can fail the output. The doctrine is what must be refused when no one is watching.
SEO, GEO and AEO
Person entity SEO: a name an engine can attach
A byline gives an engine a string. A coherent author graph gives that string somewhere to resolve.
AI Systems
Model picking audit vs benchmark screenshot: the rules your stack already runs on
A chart can suggest a candidate. The audit shows why each model got its job, and which reasons nobody can defend.
Founder Performance
The planning fallacy: a plan is not a forecast
Your plan says what you intend to do. Your record of similar projects says how long that kind of work has taken.
Venture Building
Positioning is a decision, not a tagline
If the homepage still fits the people you refused, it is a tagline.
Venture Building
Pricing psychology: a price is never a naked number
If the only lever is 'make it feel cheaper,' you do not have psychology. You have a trick looking for a page.
AI Systems
Privacy tier vs vendor promise: which data may reach a cloud LLM?
A vendor's privacy page describes the vendor. A privacy tier describes your data and decides which model hosts may receive it.
AI Systems
Prove you can stop an agent: interrupt the run and revoke access
An instruction to stop is a request. A stop is a control that someone outside the agent can operate.
Founder Performance
The Psychology of Founder Resilience: Capacity Is Infrastructure
Resilience is not what you are. It is what you run. This book turns capacity into operator infrastructure you can execute on your worst Tuesday.
AI Systems
RAG freshness: last month's PDF is not a lookup problem
The system can retrieve clause 12.3 perfectly and still miss the signed amendment.
AI Systems
RAG vs fine-tuning vs agents: three jobs, not a ladder
A vector store is not a trained model. A trained model is not a worker. A worker without a ceiling is a demo.
SEO, GEO and AEO
Impressions but no clicks: when the answer was the whole visit
A definition can finish its job on the results page. A buying decision usually still has work left, and that work is the reason to visit.
Founder Performance
Cognitive reappraisal vs positive thinking: keep the facts
The board's request is a fact. Your verdict on what it means is still a guess. Separate the two before you respond.
Founder Performance
Recovery is not a day off: match the break to the load
A free Saturday can still belong to the company if you keep answering it. The question is what today's work spent and whether that capacity has room to return.
Venture Building
Red-team sheet vs devil's advocate: test the objection
A spoken objection can end when the meeting ends. A red-team sheet gives it an evidence check, an owner and a recorded verdict.
Founder Performance
When to rely on an expert, verify, or ask another
A specialist can have the right answer about a system and still lack the facts that decide your launch date. Draw the boundary around the claim, then pick the next step: rely, verify, or ask another.
Founder Performance
What resilience research found: staying level is not bouncing back
Staying roughly level and recovering after a sustained drop are different paths. That distinction changes how you read a difficult week.
Founder Performance
Resilience vs grit: what founders need beyond persistence
Finishing the work shows that you persisted. Before the next promise, check what the push spent, whether the approach still fits, and what you restored.
AI Systems
How to retire a production agent (stopping it is not enough)
A stopped agent is not a retired agent. The next schedule, a copied key or a monthly plan can keep it working and billing.
Founder Performance
Reversible vs irreversible decisions: classify the door before you decide
Bezos split decisions into one-way and two-way doors. Most founder decisions sit somewhere in between, and some doors close while you still think they are open.
Venture Building
What is a roster? From the Monday rota to a team of AI agents
The word started as a Dutch gridiron. The idea still fits your staff, your agencies and your AI agents.
AI Systems
ROSTER vs CHORUS: AI agents for one person, or for a team?
Count the people who can say no to the work. One name means ROSTER. Several names mean you need CHORUS as well.
Venture Building
Seat card vs job description: what an AI agent needs in writing
A title names the territory. The card tells the agent when to start, what to deliver and when to hand the work back.
Venture Building
Second-order thinking audit: the first-order metric is green
If you cannot name the later bill in the meeting, you do not have second-order thinking. You have optimism with a lag.
SEO, GEO and AEO
utm_source=chatgpt.com: how to see ChatGPT and other AI traffic in GA4
GA4 can count the people who clicked. It cannot count the answers that never sent anyone.
AI Systems
SENSE: a 90-day strategic intelligence layer
Stop collecting briefings. Start changing Monday.
Venture Building
When credible sources disagree: diagnose before choosing sides
Before you back the source with the bigger name, check whether the claims actually conflict. Then name the check that could help resolve whatever disagreement remains.
Venture Building
A source tier is a role, not a grade
The case study stays the same. Its support changes when the sentence in your memo changes.
AI Systems
STACK: version the agent OS like you version the app
Rules in five tools. MCP without contracts. Skills nobody owns. That is not a model problem.
AI Systems
STACK vs AGENTS.md: where instructions end and controls begin
An instruction file is a request. A control is a refusal.
Founder Performance
Stoic systems: judgment as infrastructure, not a mood
A quote wall is not decision infrastructure. Judgment under pressure needs a protocol you can run and review.
Founder Performance
Being stoic is not Stoicism: the philosophy was never about hiding your feelings
The everyday word praises a straight face. The philosophy asked what you concluded behind it.
Founder Performance
Stop rule vs one more tab: when to stop researching
Another tab may help. First name the finding in it that could change what you do.
SEO, GEO and AEO
Technical GEO: the plumbing that decides whether AI can use your page
The best paragraph on the page is worth nothing to a system that never receives it.
AI Systems
The Agentic Author: Stop Writing Books. Build a Production System.
You are not writing a book. You are building a knowledge-production system that happens to output books. BooksOS makes that system auditable.
AI Systems
The Agentic Codebase: Version the Agent OS Like You Version the App
Prompts in chat are not architecture. Scattered rules are not governance. STACK turns your repository into a maintainable agent operating system.
AI Systems
The Multi-Agent Organization: Close the Coordination Gap With CHORUS
Tools store state. Protocols decide. CHORUS is the operating layer between having agents and running them as one team.
SEO, GEO and AEO
ChatGPT, Perplexity, Google AI Overviews: three engines, three doors to your page
Opening one door does nothing for the other two.
Venture Building
Tool sprawl vs a tool matrix: give each agent role only the tools its job needs
Three specialist names can hide one shared permission problem. A tool matrix makes you justify each role's access, operation by operation.
AI Systems
TRACE vs RAG: one fetches the text, the other checks what it proves
An answer can quote the right document and still draw the wrong conclusion.
Venture Building
Trigger map vs cron job: a schedule is one row, not the map
A weekly brief can start at nine. A client email, an arriving draft and a repeated form submission still need routes of their own.
Venture Building
Two editions, not a translation: write from one spine
Keep the argument and its proof. Rebuild the prose, examples and voice for the second edition's readers.
Venture Building
Upstream vs downstream: one river picture that explains oil, supply chains, software, and decisions
Every flow has a source and an end. The two words only tell you where you stand.
AI Systems
Using AI agents effectively: one job, a ceiling, a log
A second tool without a ceiling is not leverage. It is a second way to spend.
Venture Building
Weekly agent review vs standup: one decision per role, not a status update
A list of completed tasks leaves the next decision to you. Review the week's evidence and write down what each role should do next, what it may do alone and what needs a separate session.
Founder Performance
What is a calibration session, and what should it produce?
HR calibration aims for consistent ratings. Forecasting calibration puts confidence to a later test. In both, agreement alone does not show that the session worked.
Venture Building
What are downstream effects?
The first-order metric can be green. The cost still has to land somewhere.
Founder Performance
What calibration means: matching confidence to evidence
Confidence is a report, not a mood. A loud claim with a thin source is a calibration failure, not conviction.
Founder Performance
What does 'likely' mean? Ask five people and you get five numbers
A shared word is not a shared number. The CIA learned that in 1951, and most teams still have not.
SEO, GEO and AEO
GEO meaning: generative engine optimization, not geography
The globe is the wrong object. GEO is whether a machine can quote you without inventing you.
Founder Performance
What is a confidence grade? Why "high confidence" is not 90%
"High confidence" tells you about the evidence behind a claim. It does not tell you the odds, or what to do next.
Venture Building
What is a deal architect?
The visible number is rarely the deal. The options after signature are.
Venture Building
What is a dominated AI model? Better and cheaper is only half the test
Another model is better and no dearer. Before you switch, ask the question the label skips: can it still do your job?
AI Systems
What is a knowledge graph? Named nodes, typed edges, a walk you can audit
If you cannot name the edge, you have a poster. If you cannot open the node, you have a rumour.
AI Systems
What is a model cascade? Escalate with a stop rule
A cascade without a stop rule is spending until something sounds fluent.
AI Systems
What is a personal health agent? Data architecture, not a clinician
If it answers as a doctor, it is already the wrong worker. If it cannot show the row it used, it is a vibe.
SEO, GEO and AEO
The first screen an answer engine can quote
GEO is the system. AEO is the page. If the first screen cannot be quoted, the rest of the page will not save it.
AI Systems
What is agent orchestration? Coordination, not a graph demo
If ownership or review disappears at a boundary, the canvas is not coordinating the work.
AI Systems
What is agent strategy? Selection, sequence, exposure, exit
A well-run agent can still be the wrong worker to deploy first.
AI Systems
What is AI team dynamics? Mixed teams, human accountability
If the founder is the universal exception handler, you do not have dynamics. You have a queue with extra tools.
AI Systems
What is an AI agent? A definition founders can use
If it cannot refuse a tool call, it is not an agent you can run. It is a demo that can spend money.
AI Systems
What is an AI agent charter? Seven fields or it is a demo
If you cannot fill the seven fields, do not hand it a key.
AI Systems
What is an evidence ledger?
The ledger is the product. The reference list is a rendering.
AI Systems
An MCP connector is the plug. Permission is a separate ceiling.
The connector is a plug. The server is the tool. The ceiling is still the product.
AI Systems
What is BooksOS? Seven gates, not a writing CMS
Agents draft. Gates decide. A fluent chapter that cannot survive a fact-check is inventory you will throw away.
AI Systems
What is CHORUS? Coordination, not a choir
Org charts name reporting lines. CHORUS names coordination lines.
SEO, GEO and AEO
What is citeturn0search0? The ChatGPT leftover that shows an answer was copied without its sources
The code is harmless. The missing source is the problem.
Founder Performance
What is founder resilience? Capacity, not a mood
The company can keep moving while you are its hidden single point of failure. That is what this word is for.
AI Systems
What is GRAIN? The inspectable retrieval loop
RAG names the architecture. GRAIN names the honesty check.
AI Systems
What is hybrid search? Two indexes, one inspectable hit
Keywords catch the name. Vectors catch the paraphrase. You still have to open the hit.
Founder Performance
What is LEVEL? A five-step check before you act on a claim
The same evidence can be good enough for a small test and too weak for a big commitment. LEVEL is how you tell the two apart.
AI Systems
What is LLM grounding? The sentence has to open a folder
A real source ID can decorate a false claim. Grounding begins after the citation resolves.
SEO, GEO and AEO
What is llms.txt? A manifesto, not a ranking switch
A voluntary map guides the client that opens it. It has no power over one that never asks.
AI Systems
What is model routing? A policy, not a default
If you cannot name which model owns which job, you do not have a routing policy. You have a default.
AI Systems
What is RAG? When AI looks up a source before it answers
A fluent AI answer can still be unsourced. RAG is the method that has to show the page it used.
AI Systems
What is ReasonKit? Structured decisions through MCP
An AI assistant can sound certain. ReasonKit makes it show which claims were checked, and stop when the evidence is missing.
AI Systems
What is ROSTER? How one person manages a team of AI agents
A role is defined by what it may not do. Six written decisions turn an AI agent into a specialist you can grade, fix or retire.
AI Systems
What is SENSE? Five moves, not another unread summary
A digest proves that a system can produce text. SENSE asks where that text changes the work.
AI Systems
What is STACK? Five layers, not a vendor diagram
The agent OS should be versioned like the app. If the rules live in five tools, you have debt, not a model problem.
AI Systems
What is TRACE? Five steps to check research before you act on it
Three links can be one source. TRACE is how you find out before the decision does.
SEO, GEO and AEO
When ChatGPT is wrong about you: a correction routine
Arguing with the chat changes one conversation. Fixing the public record, then retesting, is the part you can act on and measure.
AI Systems
When not to use an AI agent: five signs, and what to use instead
An agent earns its cost only when the next step depends on what it finds. Most tasks never get there.
AI Systems
When to use RAG: named passages, not a vector store
The furniture is optional. The named passage is not. If you cannot open the file, you did not ship retrieval.
SEO, GEO and AEO
Why does ChatGPT cite my competitor and not me?
The link may lead to your competitor's own site or to a review that mentions them. Open it before you decide what to change.
SEO, GEO and AEO
Why am I not mentioned in ChatGPT?
If the crawler never saw a sentence it can steal with your name on it, the model has nothing to name.
AI Systems
Why multi-agent systems fail: the coordination gap
Capability arrived in an afternoon. The protocol never did. That gap is the whole story.
AI Systems
Why your RAG answer is wrong: find the broken step before you change the model
The model is the last thing to blame, not the first.