The Home Entity Guide

AI engines do not recommend businesses. They recommend entities they can verify. When someone asks ChatGPT or Perplexity who you are, or who they should hire, the engine reconciles every fact it can find about you against whatever source it trusts most. If no trusted source exists, it hedges, guesses, or recommends someone else. An AI agent acting for a buyer will not take your word for anything; it can only act on what it can check.

The fix is what Jason Barnard calls an Entity Home: one page, at /about, on a domain you own, that states the facts about you so clearly and consistently that machines treat it as the source of truth. Every profile you control repeats the same words and links back to that page. Engines cycle the loop, meet identical facts on every pass, and their confidence climbs until they stop hedging and start asserting.

You do not need to be technical to build one. Each phase below tells you why it matters, what to do, and gives you an exact prompt to paste into Claude that does the hard part for you. Work through the phases in order. The build is the easy part; the engines take months to absorb it, which is exactly why you start now.

The build, at a glance

Phase	Title	What you get
01	Claim your domain and your description	A canonical 20 to 50 word Executive Summary
02	Build your Entity Home at /about	Your Entity Home live at /about
03	Add the Person schema graph	One connected, validated schema graph
04	Wire your profiles outbound	Visible profile links that match sameAs exactly
05	Close the corroboration loop	Every bio repeating your canonical words
06	Open the doors to AI crawlers	AI crawlers explicitly welcomed in
07	Write name-first Q&A	Liftable Q&A merged into your graph
08	Prove you are alive	A /now page with an honest date
09	Earn independent corroboration	A /media page and outreach in flight
10	Monitor the answers monthly	A monthly audit habit with a log

Claim your domain and your description

What you get: A canonical 20 to 50 word Executive Summary

Why

Every fact an engine learns about you gets reconciled against one anchor, and Barnard's rule is that the anchor must live on a domain you personally own: if your entity lives on your employer's site and you leave, you leave your home behind. Before you build anything you also need the words. One Executive Summary, first sentence a semantic triple, 20 to 50 words, stating who you are, what you offer and whom you serve. Machines learn by repetition, and this is the string you will repeat.

Do this

1 Buy a domain in your own name (yourname.com, or the closest sensible variant) and register it personally, not through your company. If you already own one, you are ahead.
2 Decide the exact form of your name and commit to it everywhere: same spelling, same capitalisation, no drifting between 'Jim' and 'James'.
3 Gather your raw facts: current role, company, what it does, who you serve, and one or two other true, notable facts.
4 Paste the prompt below into Claude with your facts filled in, and pick the strongest of the three candidates it gives you.
5 Save the winner as your canonical Executive Summary. From now on it is copy-paste only: never retyped, never 'improved'.

Hand it to Claude

You are helping me write my canonical entity description, called an Executive Summary in Jason Barnard's Kalicube methodology. This is the single most important string of text in my personal brand: I will repeat it word for word on my website and on every profile I control, so that AI engines see identical facts everywhere they look.

The rules, which you must follow exactly:
1. The first sentence must be a semantic triple: "[YOUR NAME] is a [category or role]." Subject, predicate, object. No warm-up words, no adjectives before my name.
2. The whole summary must be 20 to 50 words. Count the words and show me the count.
3. It must state three things: who I am, what I offer, and whom I serve.
4. Plain factual language. No hype words (passionate, visionary, world-class, leading). Use only facts I give you below; if you are missing something important, ask me rather than inventing it.
5. Third person throughout.

My facts:
- Name (the exact form I always use): [YOUR NAME]
- Primary role or category: [e.g. founder, consultant, architect]
- Company or main project: [YOUR COMPANY], which [WHAT IT DOES, ONE LINE]
- Who I serve: [YOUR AUDIENCE, e.g. small manufacturing firms in the UK]
- One or two other true, notable facts: [OPTIONAL: previous company, book, credential]

Write three candidate Executive Summaries that each follow every rule. Under each one, show the word count and point out the semantic triple. Then recommend the strongest candidate and explain in two sentences why, judged on clarity to a machine that has never heard of me.

Check it worked. Read your chosen summary aloud as the answer to 'Who is [you]?'. If it stands alone as a complete, true answer and the word count sits between 20 and 50, it works. Save it somewhere permanent: it is now canonical.

Build your Entity Home at /about

What you get: Your Entity Home live at /about

Why

Barnard designates the /about page, not the homepage, as the Entity Home, because it is entirely dedicated to you: a homepage forces machines to juggle everything your site does at once. Treat the page as a training document for AI engines, not marketing copy. Thin sales prose builds no machine confidence; fact-dense third-person prose, with the most important content first and modular H2 sections machines can lift cleanly, does.

Do this

1 Create a page at yourdomain.com/about. Any platform works if it outputs real HTML: WordPress, Squarespace, Webflow, Framer, or a static site.
2 Run the prompt below to generate the full page content.
3 Check the output: your Executive Summary must be paragraph one, untouched, and anything marked [TODO] gets a real fact or gets deleted.
4 Publish, and resist the urge to lead with a sales pitch. This page is a reference document about you, not a landing page.
5 Link to /about from your site navigation so every page on your site reaches it.

Hand it to Claude

You are helping me build my Entity Home: the /about page on my own domain that acts as the canonical source of truth about me for search engines and AI engines. Jason Barnard describes this page as a training document for AI, not marketing copy. Write it that way: fact-dense, third person, no sales voice.

Structure, which you must follow exactly:
1. H1: "About [YOUR NAME]"
2. First paragraph: my Executive Summary, reproduced word for word exactly as I paste it below. Do not edit, reorder or improve it.
3. Then modular H2 sections, each a self-contained factual paragraph or two:
   - "Who [YOUR NAME] is today" (current role and focus)
   - "What [YOUR NAME] does" (the work, concretely)
   - "Who [YOUR NAME] serves" (the audience and the problems solved)
   - "Companies and projects" (each one named, with one factual line and the relationship stated plainly: "[YOUR NAME] is the founder of [COMPANY]")
   - "Background" (prior roles, education, location, in brief)
4. Every claim must come from the facts I give you. Where you need a fact I have not given, write [TODO: fact needed] instead of guessing. Never invent dates, numbers, clients or awards.
5. Write for a human reader, but remember the primary audience is a machine deciding whether it understands who I am. Prefer plain declarative sentences that state facts about "[YOUR NAME]" by name, rather than long runs of pronouns.

My Executive Summary (use verbatim as paragraph one):
[PASTE YOUR EXECUTIVE SUMMARY FROM PHASE 1]

My facts:
- Current role(s): [FILL IN]
- Companies or projects and what each one does: [FILL IN]
- Audience and the typical problems I solve: [FILL IN]
- Prior work history: [FILL IN]
- Education or credentials: [FILL IN]
- Location: [FILL IN]
- Anything else true and relevant: [FILL IN]

Output the complete page as clean HTML I can paste into my site: one h1, h2 section headings, p paragraphs, no styling, no scripts.

Check it worked. Open yourdomain.com/about, right click, and choose View Page Source. Your Executive Summary should appear word for word in the raw HTML, and a search for '<h1' should find exactly one heading.

Add the Person schema graph

What you get: One connected, validated schema graph

Why

The prose teaches machines that can read; the schema removes all doubt. A single connected JSON-LD graph with stable @ids gives every fact a permanent machine-readable address, states your description in exactly your canonical words, and declares your relationships (you founded this company, this company was founded by you) in both directions. Persistence matters more than mere validity: an @id is a promise you keep for the life of the domain.

Do this

1 Fill in the prompt below with your domain, facts and real profile URLs.
2 Paste the generated script block into the head of your /about page. Most site builders have a 'custom code' or 'code injection' setting for individual pages.
3 Paste the second, smaller AboutPage block into the head of your homepage: it declares on the homepage itself that the site is about you, pointing at the same Person @id on /about.
4 Validate both pages at validator.schema.org and fix anything they flag.
5 Write down your Person @id (https://yourdomain.com/about#person) somewhere permanent. It must never change, through any redesign, ever: changing it resets your history with the knowledge graph.

Hand it to Claude

You are generating the schema.org JSON-LD for my Entity Home page at https://[YOURDOMAIN.COM]/about. Follow these rules exactly; they matter more than brevity:

1. Output ONE script block for the /about page containing a single connected @graph. Never emit disconnected schema snippets on the same page.
2. Every node gets a stable @id. Use these and never vary them:
   - Person: "https://[YOURDOMAIN.COM]/about#person"
   - WebSite: "https://[YOURDOMAIN.COM]/#website"
   - ProfilePage: "https://[YOURDOMAIN.COM]/about#webpage"
   - Homepage AboutPage: "https://[YOURDOMAIN.COM]/#webpage"
   - Each organisation: "https://[YOURDOMAIN.COM]/about#org-[short-name]"
3. Relationships must be @id references (for example "founder": {"@id": "https://[YOURDOMAIN.COM]/about#person"}), never a second full copy of the node.
4. The Person node must include: name, givenName, familyName, description, disambiguatingDescription, url (the /about page), image (an ImageObject with its own @id, url and caption), jobTitle, worksFor (@id reference), founder (@id references), homeLocation (a Place naming my city and country), alumniOf (my school or university), knowsAbout (my topics), sameAs, and mainEntityOfPage.
5. The description property must be my Executive Summary below, character for character. Do not rephrase it.
6. The ProfilePage node must declare mainEntity and about pointing at the Person @id, plus datePublished and dateModified taken only from the dates I give you below.
7. Each company gets an Organization node with name, url, a one-line factual description, and founder: {"@id": "https://[YOURDOMAIN.COM]/about#person"} so the founder link exists in BOTH directions.
8. sameAs may contain ONLY the profile URLs I paste below. If a platform is missing from my list, leave it out. Never add a plausible-looking URL I did not give you.
9. Use only values I give you; where something is missing, output [TODO] rather than inventing anything, including dates.
10. ALSO output a SECOND, small script block for my HOMEPAGE: an "AboutPage" node with @id "https://[YOURDOMAIN.COM]/#webpage", isPartOf referencing the WebSite @id, and both about and mainEntity set to {"@id": "https://[YOURDOMAIN.COM]/about#person"}. This block lives on a different page, so it is the one permitted exception to rule 1: it declares, on the homepage itself, that the site is about me.

My details:
- Domain: https://[YOURDOMAIN.COM]
- Full name: [YOUR NAME] (given name: [FIRST], family name: [LAST])
- Executive Summary (verbatim, for description): [PASTE IT]
- One-line disambiguating description (how to tell me apart from other people with my name): [e.g. the [CITY]-based founder of [COMPANY]]
- Job title: [TITLE]
- Location (city, country): [CITY, COUNTRY]
- Education (institutions and qualifications, for alumniOf): [FILL IN]
- Companies: [COMPANY 1] (url, one line on what it does, my relationship: founder or employee), [COMPANY 2] (same details)
- Topics I want associated with me (knowsAbout): [3 TO 8 TOPICS]
- Headshot image URL hosted on my own domain: [URL]
- My real profile URLs for sameAs: [LINKEDIN URL], [X URL], [GITHUB URL], [YOUTUBE URL], [ANY OTHERS]
- Today's date (for dateModified): [DATE]
- Date the /about page was first published (for datePublished): [DATE, or same as today]

Output the complete <script type="application/ld+json"> block for my /about page, then the separate homepage AboutPage block from rule 10 ready for the <head> of my homepage, followed by a short checklist confirming which of the rules above the output satisfies.

Check it worked. Paste https://yourdomain.com/about into validator.schema.org. It should parse with no errors and show one connected graph, and the Person description should match your canonical string exactly, character for character.

Wire your profiles outbound

What you get: Visible profile links that match sameAs exactly

Why

Understanding is not enough; engines want corroboration. The first half of Barnard's self-confirming loop is outbound: your Entity Home links visibly to every profile you control, telling machines 'these pages are also me, go and check'. The sameAs array in your schema is the machine-readable version of the same claim, so the visible links and the array must match exactly.

Do this

1 List every profile you genuinely control: LinkedIn, X, GitHub, YouTube, Instagram, plus any industry directories that list you.
2 Run the prompt below to generate the 'Elsewhere on the web' section and reconcile it against your sameAs array.
3 Add the section to your /about page as visible, clickable links.
4 Delete or ignore dead profiles rather than linking to them; a link to an abandoned page corroborates nothing.

Hand it to Claude

You are building the outbound half of my corroboration loop: a visible "Elsewhere on the web" section on my /about page that links out to every profile I control, matching the sameAs array in my schema exactly.

Rules:
1. Only include the URLs I paste below. Do not add platforms I did not list.
2. Links to profiles I personally own and write get rel="me". Links to third-party pages about me (directories, press) do not get rel="me".
3. Anchor text is the platform name plus my name where natural (for example "[YOUR NAME] on LinkedIn"), never "click here".
4. Output plain semantic HTML: an h2 reading "Elsewhere on the web" and a ul of links. No styling, no scripts.
5. Then compare this list against the sameAs array I paste from my schema, and report any URL that appears in one list but not the other, because the two must match one for one.

Profiles I own (these get rel="me"): [LINKEDIN URL], [X URL], [GITHUB URL], [YOUTUBE URL], [ANY OTHERS]
Third-party pages about me (no rel="me"): [DIRECTORY OR PRESS URLS, or "none"]
My current sameAs array from Phase 3: [PASTE THE sameAs ARRAY FROM YOUR SCHEMA]

Check it worked. Count the links in your 'Elsewhere on the web' section, then count the URLs in your sameAs array. Same number, same URLs, and every link opens the correct live profile.

Close the corroboration loop

What you get: Every bio repeating your canonical words

Why

This closes the loop, and it is the step most people never do. Each profile must link back to /about specifically, not your bare homepage, and repeat your Executive Summary as its bio. Engines then cycle home, out to a profile, back home, and meet identical facts on every pass; each consistent cycle raises confidence. Slightly different bios on each platform are the single most common failure in the wild, because to a machine, variety reads as contradiction.

Do this

1 Run the prompt below to get per-platform bio text and the exact URL for every website field.
2 On each platform, set the website link to https://yourdomain.com/about, not your homepage.
3 Replace each bio with the generated text: verbatim where it fits, trimmed from the end where it does not.
4 Check that your name reads identically on every platform.
5 Expect this to be the tedious phase. It is also the one most people skip, which is why doing it works.

Hand it to Claude

You are closing the inbound half of my corroboration loop. Every profile I control must (a) link back to my Entity Home at https://[YOURDOMAIN.COM]/about, not my bare homepage, and (b) carry my Executive Summary as its bio, as close to word for word as each platform allows.

Rules:
1. Where a platform's character limit fits the full Executive Summary, use it verbatim. No improvements, no emoji, no hashtags.
2. Where the limit is too tight, produce a trimmed version that keeps the first sentence (the semantic triple "[YOUR NAME] is a ...") completely intact and cuts from the end, never from the start.
3. Keep my name form identical everywhere: "[YOUR NAME]", no nicknames or initials.
4. Output a table with these columns: platform, character limit, exact bio text to paste, character count of that text, and the URL to enter in the website field (always https://[YOURDOMAIN.COM]/about).

My Executive Summary: [PASTE IT]

My platforms and their bio character limits (edit this list to match the accounts you actually have):
- LinkedIn headline (220 characters) and LinkedIn About section (2,600 characters)
- X bio (160 characters)
- GitHub bio (160 characters)
- YouTube channel description (1,000 characters)
- Instagram bio (150 characters)
- [ANY OTHERS]

Check it worked. Open each profile in a private browser window. The website field should point to /about, not the homepage, and the bio should read as your canonical summary.

Open the doors to AI crawlers

What you get: AI crawlers explicitly welcomed in

Why

None of this counts if the machines cannot read it. Barnard calls this Deliverability: your facts must sit in plain crawlable HTML, your robots.txt must explicitly welcome the AI crawlers (many default configurations block them 'to protect content', which defeats the entire project), and a sitemap must tell the index what exists. llms.txt is cheap extra insurance, not a strategy; the crawlable page is the strategy.

Do this

1 Run the prompt below with your platform name to generate robots.txt and llms.txt, and to find out where each file goes.
2 Upload both files to your site root.
3 Confirm your platform generates an XML sitemap and that robots.txt points to it.
4 Set the meta description and og:description on your homepage and /about to your Executive Summary (trimmed from the end only where a limit forces it), lead each page title with your name, and repeat the Summary as the first factual paragraph of your homepage.
5 Settle your canonical URLs: pick www or non-www, pick one trailing-slash style, and stick to both choices forever.

Hand it to Claude

You are making my site readable by AI engines. Generate the deliverability files for my site at https://[YOURDOMAIN.COM], which is hosted on [PLATFORM, e.g. WordPress, Squarespace, Webflow, Framer, Netlify, GitHub Pages].

1. A robots.txt that explicitly ALLOWS the AI and search crawler fleet: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, anthropic-ai, PerplexityBot, Perplexity-User, Google-Extended, Applebot-Extended, Googlebot, Bingbot and DuckDuckBot. No Disallow rules that block any of them from my public pages. End with a Sitemap: line pointing at my sitemap URL.
2. An llms.txt file: a short markdown document starting with my name as an H1, then my Executive Summary verbatim, then a linked list of my key pages. Include a comment noting that llms.txt is cheap insurance, not a substitute for crawlable HTML.
3. For [PLATFORM] specifically, tell me: where to put each file, how to check my XML sitemap exists and is referenced from robots.txt, and how to confirm every page has a canonical URL tag.
4. The exact <title>, meta description and og:description tags for my homepage and my /about page. Each title must lead with my name; each description must be my Executive Summary verbatim, trimmed from the end only where a length limit forces it, keeping the first sentence intact. Remind me that the Summary must also appear as the first factual paragraph of my homepage, not just on /about.
5. Finally, give me one check a non-technical person can run to confirm my /about content is present in the raw HTML rather than loaded by JavaScript (for example viewing page source and searching for a phrase from my summary).

My Executive Summary: [PASTE IT]
My key pages: https://[YOURDOMAIN.COM]/about, https://[YOURDOMAIN.COM]/now, [ANY OTHERS]

Check it worked. Visit yourdomain.com/robots.txt and yourdomain.com/sitemap.xml in a browser; both should load. Then view the page source of /about: if your summary is there in the raw HTML, AI crawlers can read it.

Write name-first Q&A

What you get: Liftable Q&A merged into your graph

Why

When an AI answers a question about you there is no page two; it lifts one answer, whole. So your answers must be liftable: each one starting with your name, self-contained, one to three sentences, and consistent with your canonical facts. Answers that open with 'He' or 'We' depend on context the engine will not carry with it, and answers that drift from your canonical facts cannot be safely cited at all.

Do this

1 Run the prompt below to generate your Q&A pairs and the matching FAQPage schema.
2 Add a featured set of 4 to 6 questions to /about; a separate /q-and-a page can hold the rest.
3 Read every answer alone, out of context, and rewrite any that lean on a previous sentence.
4 Replace your /about schema block with the merged version, one connected @graph that now includes the FAQPage, and re-validate at validator.schema.org.

Hand it to Claude

You are writing the Q&A section that AI engines will quote when someone asks about me. In AI answers there is rarely a second page: the engine lifts one self-contained answer, so every answer must survive being read with zero surrounding context.

Rules for every answer:
1. It starts with my name: "[YOUR NAME] is..." or "[YOUR NAME] founded...". Never "He", "She", "They", "I" or "We" as the opening word.
2. It is 1 to 3 sentences, factual, third person, and fully self-contained.
3. It must be consistent with my Executive Summary and the facts below. Nothing invented; if you lack a fact for a question, output [TODO] for that answer instead of guessing.
4. The questions are the literal phrasings people type: "Who is [YOUR NAME]?", "What does [YOUR NAME] do?", "What is [YOUR COMPANY]?", "Who does [YOUR NAME] work with?", "Where is [YOUR NAME] based?", plus 3 to 5 more that fit my situation.
5. The answer to "Who is [YOUR NAME]?" is my Executive Summary verbatim.

My Executive Summary: [PASTE IT]
My facts: [PASTE THE FACTS BLOCK FROM PHASE 2, OR THE TEXT OF YOUR /about PAGE]
My current schema from Phase 3: [PASTE THE FULL JSON-LD BLOCK FROM YOUR /about PAGE]

Output two things: (a) the Q&A as clean HTML (an h2 per question, a p per answer) ready for my /about or /q-and-a page, and (b) my Phase 3 schema block with a FAQPage node MERGED into the same single connected @graph, never a second disconnected script block. Give the FAQPage the stable @id "https://[YOURDOMAIN.COM]/about#faq", set its about and mainEntity to {"@id": "https://[YOURDOMAIN.COM]/about#person"} so every Question stays anchored to me, keep every existing @id unchanged, and make each Answer's text identical to the HTML answers. Output the complete merged block ready to replace the schema on the page.

Check it worked. Take any single answer and read it out of context. If a stranger could hear it alone and know exactly who it is about, it is citable. Then re-validate the page for the FAQPage markup.

Prove you are alive

What you get: A /now page with an honest date

Why

Engines constantly reconcile your page against fresher third-party data, and stale facts lose. A /now page with an honest, visible, recently updated date, matched by dateModified in your schema, tells every layer of the system that this source is alive and still the winning reference. Fake freshness, a date that updates while nothing changes, is worse than none; the machine notices.

Do this

1 Jot down 3 to 6 bullets of what you are actually working on right now.
2 Run the prompt below to generate the /now page with today's real date.
3 Add a visible 'Last updated' date to /about as well.
4 Set the quarterly calendar reminder the prompt gives you. An honest older date beats a fake fresh one.

Hand it to Claude

You are writing my /now page: a short page saying what I am focused on right now, which proves to search and AI engines that my Entity Home is alive and current. Stale entity facts lose to fresher third-party data, so this page and its dates must be honest.

Rules:
1. Keep it short: a dated heading, one intro line, and 3 to 6 bullets of current focus, written in the present tense.
2. Put "Last updated: [TODAY'S DATE]" visibly on the page. Use the real date; never post-date or fake freshness.
3. First person is fine on this one page, but keep every name and fact consistent with my /about page.
4. Nothing aspirational dressed up as current: only what is actually happening now.

What I am currently working on (rough bullets are fine): [YOUR CURRENT PROJECTS AND FOCUS]
My name and company, for consistency: [YOUR NAME], [YOUR COMPANY]

Output three things: (a) the /now page as clean HTML with the visible Last updated line, (b) a one-line reminder I can paste into my calendar as a quarterly recurring event, reading "Update /now and the Last updated date on /about", and (c) if I paste my existing schema block below, the same block with dateModified set to today: [OPTIONALLY PASTE YOUR SCHEMA FROM PHASE 3]

Check it worked. Load /now and check the visible date is real and recent, and that it matches dateModified in your schema. Then actually diarise the quarterly update; a /now page last touched a year ago signals the opposite of what you built it for.

Earn independent corroboration

What you get: A /media page and outreach in flight

Why

Everything so far is you agreeing with yourself, and engines discount that: a closed loop of owned profiles caps how much confidence you can earn. What moves an engine from hedging to asserting is independent sources repeating the same facts: podcasts, directories, industry press, professional bodies. Your job is to make it effortless for them to use your exact words, and to surface the coverage on a /media page so the loop keeps widening.

Do this

1 List every piece of independent coverage you already have: interviews, podcasts, directories, press mentions, guest posts. 'None yet' is a fine starting answer.
2 Run the prompt below to build your /media page, boilerplate bio and target shortlist.
3 Publish /media, even if it is short. Real and thin beats padded and fake.
4 Send the outreach: podcasts and niche publications in your field, offering your boilerplate and asking them to link to /about.
5 Every time you are featured anywhere, hand over the boilerplate. Same words, every time.

Hand it to Claude

You are building my third-party corroboration plan. Profiles I own all repeating the same facts is a closed loop; what moves AI engines from hedging to asserting is independent sources saying the same thing. Help me earn that honestly.

Rules:
1. Never invent coverage. The /media page may list only items I confirm actually exist.
2. The boilerplate bio must contain my Executive Summary word for word; you may add ONE extension sentence of true supporting fact.
3. Suggested targets must be realistic for someone at my current level of visibility: niche podcasts, industry directories, professional bodies, local business press, guest posts. Not "get featured in Forbes".

My Executive Summary: [PASTE IT]
My niche and audience: [DESCRIBE THEM]
Existing independent coverage, interviews or mentions, with links ("none yet" is fine): [LIST OR "NONE YET"]
Communities, associations or publications I already have a relationship with: [LIST]

Output four things:
(a) A /media page in clean HTML listing my confirmed coverage, each item linking to its source, or a starter version with a [TODO] note if I have none yet.
(b) My boilerplate bio paragraph for journalists and podcast hosts to paste.
(c) A shortlist of 8 to 12 realistic third-party targets by type, each with a one-line angle for why they would feature me.
(d) A short, polite outreach email template that includes the boilerplate verbatim and asks the publisher to link to https://[YOURDOMAIN.COM]/about.

Check it worked. Every item on /media opens a real page you do not own. Over the following months, search your name plus your company: independent pages repeating your canonical facts should start appearing.

Monitor the answers monthly

What you get: A monthly audit habit with a log

Why

You optimise what you measure, and the measurement here is the actual answer. Barnard's diagnostic is simple: ask each engine who you are, compare the reply against your canonical description, and treat every wrong or missing fact as a corroboration gap with a specific fix. Knowledge graphs accumulate over months, so this is a monthly habit, not a launch task; the payoff is watching the engines converge on your own words.

Do this

1 Once a month, ask ChatGPT, Claude, Perplexity and Google's AI: 'Who is [YOUR NAME]?' Save each answer.
2 Google your name and note what the results page says about you.
3 Run the audit prompt below with the answers pasted in.
4 Do the one highest-leverage fix it recommends. One per month is enough.
5 Be patient on purpose: knowledge graphs absorb consistent signal over months, and the log is how you see it working.

Hand it to Claude

You are my monthly AI visibility auditor. I have asked the major AI engines who I am, and I am pasting their answers below. Compare them against my canonical facts and tell me exactly what to fix.

My canonical Executive Summary: [PASTE IT]
My key facts (roles, companies, audience, location): [PASTE YOUR FACTS BLOCK]

This month's answers to "Who is [YOUR NAME]?":
- ChatGPT: [PASTE THE ANSWER, or "not asked"]
- Claude: [PASTE]
- Perplexity: [PASTE]
- Gemini or Google AI: [PASTE]
- Google results page for my name (describe the top results in a line or two): [DESCRIBE]

Produce:
1. A drift report per engine: correct facts, wrong facts, missing facts, stale facts, and whether the engine cited my site or my profiles.
2. For each problem, the likely root cause in corroboration terms (for example "an old bio on X still says [OLD ROLE]" or "no third-party source states [FACT]") and which fix it maps to: my /about page, a profile bio, my schema, crawler access, my Q&A, freshness, or a new third-party source.
3. The single highest-leverage action for me this month, in one sentence.
4. A markdown table row I can append to my tracking log: date, engine, verdict (correct / partial / wrong / absent), one-line note.

Be blunt. If an engine has never heard of me, say so plainly, and remind me that knowledge graphs need months of consistent signal, not days.

Check it worked. Keep the log and compare month on month. Success is convergence: the engines gradually answering 'Who is [you]?' in your own canonical words, citing pages you control or earned.

Credit where it belongs

This methodology is Jason Barnard's. As the founder of Kalicube, he pioneered the Entity Home concept, Brand SERP optimisation and the Kalicube Process that teaches search and AI engines who you are, long before AI visibility was fashionable. This guide is simply an implementation path with Claude as the builder; the thinking is his, and you can go much deeper at jasonbarnard.com and kalicube.com.

If you would rather have all of this built and maintained for you, get in touch.