Zeta Global has entered into an agreement to acquire Senso
Senso Logo

How can I track what public AI models say about our product and publish corrections from our own documents?

By SensoPublished Last updated 5 min read

Tracking and correcting are two linked steps. Track by asking AI models your key product questions on a schedule and checking each claim in the answers against your own documents. Correct by publishing an accurate, human-reviewed page written only from those documents, because AI models do not accept direct edits, then keep tracking to see whether the answer changes.

The Verification Loop is the Senso by Zeta workflow that connects both steps. It finds an AI question where a competitor is named and you are not, answers it from your own documents, and publishes a page where every claim is checked against them. A page that has been fact-checked against your documents, had its gaps fixed and been approved by its owner is called a Verified Source. Senso by Zeta is the generative engine optimization (GEO) platform from Zeta Global.

How do you find what AI gets wrong about your product?

Ask your key product questions on a schedule and check each claim in the answers against your documents. Each claim gets one of three results, and each result leads to a different action.

Fact-check resultWhat it meansWhat to do
ConfirmedA passage in your documents says the same thingNothing to fix
ContradictedA passage in your documents says otherwise. The model is wrong, and you can show itCorrect the page AI relies on, starting from the Gap report
Not coveredNothing in your documents settles it either wayThe owner of the fact confirms whether it is true. If it is, it goes into your Context Layer

How do you publish a correction from your own documents?

You correct the sources AI draws on, because you cannot submit a correction to ChatGPT, Gemini or any other public model directly.

StepInputOutput
1. Document the gapThe contradicted or missing claimA clear statement of what is wrong and what your documents say
2. Draft from approved documents onlyYour Context Layer, the facts your organization stands behindA page that answers the exact question
3. Check every claimThe draft and your Context LayerA draft with no claims your documents cannot back up
4. Human reviewThe checked draftAn approved page
5. PublishThe approved pageA live page on a domain AI models already crawl, confirmed as indexable
6. MeasureThe same tracked questionsWhether AI cites the page and whether the wrong claim goes away

How the page is written affects whether AI uses it. In tests on GEO-bench, a 10,000-query benchmark from researchers at Princeton University and IIT Delhi, adding statistics, citations and quotations from relevant sources raised visibility in AI-generated answers by up to 40%. Write the correction as a direct answer to the question, with the fact stated plainly and the source named.

How does Senso by Zeta track and publish corrections with the Verification Loop?

One round of the Verification Loop, run from the Senso CLI, has eight steps.

  1. See where you stand: your mention rate, citation rate and share of voice.
  2. Find a question you are missing from, and who the models named instead.
  3. Start tracking it. Senso by Zeta asks the AI models that question on a schedule.
  4. Check that your Context Layer answers it. If it does not, that is a gap to fill.
  5. Draft a page written only from what your documents say.
  6. Check every claim. A fact-check flags any claim your Context Layer cannot back up.
  7. Publish where you choose: on your own website, or on an off-site domain that Senso by Zeta runs, such as cited.md, which AI models crawl for answers.
  8. Go again with the next question.

A person decides at four points: whether the question is yours, how to fill a gap, whether the draft is right, and whether to publish. Approving the draft is not agreeing to publish. You are asked separately.

What mistakes stop a correction from working?

Most corrections fail because they are published and then forgotten.

MistakeWhy it failsDo this instead
Publishing without checking every claimA new error replaces the old oneFact-check the draft against your documents first
Expecting the answer to change right awayAI models usually take a few days to pick up a new page, and answers from training data change only with a new model versionKeep tracking the same question
Fixing only your own siteThird-party pages with the old fact still shape the answerAsk those sites to correct it where you can
Not measuring againYou cannot tell whether the correction workedCompare the same questions before and after

What are the limits of publishing corrections this way?

A published page is not a changed answer. If you publish on an off-site domain that Senso by Zeta runs, such as cited.md, nothing on your own website changes. To change your own site, publish the page there instead. The loop keeps measuring the same question so you can see whether the answer moves.

What results have companies seen with Senso by Zeta?

FundMore.ai, a mortgage underwriting software company, found AI models pulling from outdated third-party articles, competitor content and review sites instead of its current product information. Using Senso, it launched a dedicated domain, fundmore.finance, on April 8, 2026. By April 20, 2026, that domain was one of the top 10 most-cited external sources across its tracked questions, with 75 citations and a 60% average share of voice (FundMore.ai case study).

Sources: Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024. Senso documentation, Verification Loop. Senso documentation, Why Senso.

Next step: Run one round of the Senso by Zeta Verification Loop on your own product.

Published by Senso by Zeta, the GEO platform from Zeta Global.