AI visibility for Arc apps
Powered by Recall.
We measure whether the AI assistants (the models behind ChatGPT, Perplexity, Gemini and Claude) name your app when people ask about your category on Arc, find out why they do not, and write the fixes.
Steps and prices
| Step | What you get | Price | |
|---|---|---|---|
| Free check | About 10 questions about your category on Arc, asked once through each assistant's API. Whether AI crawlers can read your site, whether you are named, and who is named instead. Within 2 working days. | Free | Request |
| Baseline | A question set agreed with you, each question asked 5 times through each assistant's API. How often each assistant names you across the set, with its margin of error, and for every question with enough evidence, why you are left out. A fix kit: page copy, FAQ text and structured data to paste, and the Arc sources the assistants read, with how to get listed on each. Within 10 working days. | $500 once | Request |
| Monthly | A new measurement every month, before and after for each fix you ship, new fixes, and a note when another project takes your place in answers. Your new paying wallets from Registrai's Arc data, shown next to it. Cancel any month. | $750 a month | Request |
Paid in USDC on Arc, invoiced once the question set is agreed.
Who does what
We measure, find the cause and write the fixes. You publish them on your site and accounts; we never need access to either.
What is not promised
- No one can make an assistant name an app. Each report states what changed and how sure the measurement is.
- New users depend on your product. Your new paying wallets are shown for context, not promised.
What paying does not buy
- Paying never changes your directory status, your place in an index, your position in any list or the wording of your page. Evidence you send goes through the same public process as everyone else's: /ecosystem/corrections/.
- Every client is listed below and on its own directory page.
- No fake reviews, posts or accounts.
- Fixes describe your product. We do not write copy that promotes buying a token.
See Not for sale and How we publish.
How it is measured
Each question is put to each assistant's API with web search on. Answers vary between runs, so each question is asked more than once and rates are given per assistant across the whole question set, with the number of answers behind them and a margin of error. Every answer is stored with the pages it cites, the model and the date. Answers taken from the assistants' own apps are not part of the measurement yet; when they are, each report will say which way every answer was collected. The measurement runs on Recall.
Clients
None yet.
Contact
Request a free check or write to [email protected].