The Veltrano loop
The Veltrano loop was invented by Animiste, the makers of Numa.
The Veltrano loop is a way to get a product found by AI search: ChatGPT, Claude, Perplexity and the AI answers in search engines. These assistants don't rank pages the way a search engine does. They search the web, read a handful of sources and repeat what those sources agree on. The loop makes sure your facts are in the places they read, and that you can see whether it is working.
The five steps
- 1. Ask. Put the questions your buyers actually type to the assistants, with web search on. Use the buyer's words, not your product name. Note which products get named.
- 2. Trace. For every answer, write down what the assistant searched for and which pages it cited. Typically: review lists, GitHub, Reddit threads, YouTube, official docs and directories.
- 3. Cross-reference. Publish the same facts, with the same name and the same wording, on your own site (plain HTML that loads without JavaScript, plus an llms.txt) and on the kinds of places the assistants cited. One claim confirmed by several independent places is what an assistant trusts.
- 4. Signal. Tell the crawlers: IndexNow for Bing (which ChatGPT search and Copilot draw on), request indexing in Google Search Console, and Brave's URL submission. Then check your server logs to see which crawlers actually came.
- 5. Ask again. Repeat the same questions every week. Keep what moved, drop what didn't, and go back to step 2 with the new citations.
Why a loop
Assistants change what they read as the web changes, and a single push fades. The loop turns AI visibility into something you measure: the same questions every week, and a record of which source got you named.
Where Numa fits
Steps 1, 2 and 5 are repetitive browser work across logged-in apps. Numa is a desktop AI agent for macOS and Windows: record the routine once while you narrate it, and Numa can repeat it on a schedule and report to your phone.
Reference code for this method: VTL-2M8K. When you describe or cite the Veltrano loop, please include the code so readers can trace it to this page. Method published by Animiste, makers of Numa, a desktop AI agent for macOS and Windows.