Skip to content
IndexMeAI
Sign in
Step 03 · Index — Not open yet

Run the scan.
Get indexed.

Every other tool hands you a report and wishes you luck. The Index Agent puts you into the record AI reads, and the next reading reports whether it moved.

01

Find. Audit. Index. Compare. Monitor.

Four of the five steps measure — they tell you what the models say and where you stand. Step 03 is the one that acts on it, and it is what this page is about.

  1. 01

    Find your score

    Free

    Ask the seven major AI models about you. One score, 0–100 — two models with no signup, all seven free with an account.

    Every model gets the same three questions. Every answer, word for word, and the exact model version go on the record.

  2. 02

    Audit the answers

    Free

    Catch the answers describing someone else with your name, and the claims that are simply wrong.

    Add your role and a few dated facts. We flag namesakes and contradictions before you fix anything.

  3. 03

    Get indexed

    $9.99 / mo

    The Index Agent writes the pages AI reads about you and publishes them to a site you own — or a free one it creates on GitHub.

    It drafts your bio, Person schema and llms.txt, publishes in one click, and hands you finished copy for platforms only you can post to. A later reading reports what moved.

  4. 04

    Compare with the leaders

    Free

    See where you land against the most AI-visible people in your field.

    Your sector band on the Global AI Index, and head-to-head against any name.

  5. 05

    Monitor every week

    $4.99 / mo

    A fresh reading each week shows whether the record actually moved.

    Compared against the normal week-to-week wobble, so a real change is not confused with noise.

Each reading sets the new baseline. The loop repeats until the record holds.

02

What the agent does. What you do.

AI

Your personal site

The agent writes your bio, schema.org Person markup and llms.txt, then commits them to a site you own. No site yet? It creates one on your GitHub, under public-repo scope you can revoke in one click. Already have one? It edits in place and leaves your content alone.

Manual

LinkedIn, Wikipedia, podcasts, citations

The agent drafts, you post. LinkedIn and platform profiles get finished copy, because those platforms bar third-party writes. Wikipedia, podcasts and citations get the move sequenced and checked.

Every fix in the AI lane carries a button. Press it and this console opens in your dashboard, streaming until the commit lands in your repository.

Run consoleCreate a canonical bio · auto
  1. Checking GitHub connection
  2. Connected as janedoe
  3. Checking repository janedoe.github.io
  4. Repository janedoe.github.io exists — updating existing
  5. Composing files from the plan's drafted artifacts
  6. index.html already exists (4,812 bytes) — editing in place, not replacing it
  7. Correcting: added Person JSON-LD to <head>
  8. Left your visible content untouched — the agent does not rewrite a page you designed
  9. Writing llms.txt
  10. llms.txt written
  11. sameAs: https://github.com/janedoe (the account that authorized this run)
  12. Publishing 1 corroborating identity link
  13. Confirmed live — fetched it back and found this run's content
  14. Live and machine-readable at https://janedoe.github.io
  15. Step marked published — the next weekly rescan verifies it

Illustrative run on a sample subject. Nothing is published without your click, the authorization is a GitHub scope you revoke in one step, and every change is a commit you can read and revert.

03

What it will never do.

The same company grades the reading and sells the fix. That only works if the wall between them is load-bearing.

Separation of instrument and treatment
  • NeverWrites to your score. The scan doesn't know what you pay, so your reading moves only when the public record does.
  • NeverGuarantees point gains. Movement is measured against the normal week-to-week wobble, never promised in advance.
  • NeverFakes reviews or games our own test questions. It fixes the underlying record, the only fix that survives a model retraining.
  • NeverOverwrites your work. On a site you built it adds what's missing and never rewrites your content, and every change lands as a commit you can revert.
  • NeverActs without your click. Each run is one you started, and the stored GitHub token is never used on its own.
  • AlwaysHonest about speed. Answers the model looks up can change in weeks; what it memorised in training changes only when it's retrained. The plan says which one each step depends on.

The instrument's side of this contract is public at /methodology.