A new product starts with no mentions in AI answers. Record a baseline in launch week, make your site easy for AI engines to read, and publish the pages they need to describe you correctly.
100 free credits when you sign up. No credit card.
Launch baseline
A startup gets into AI answers by being easy to read and easy to describe. Record a launch-week baseline of prompts on ChatGPT, Claude, Gemini, Perplexity and Google's AI answers, make sure AI crawlers can fetch the site, and publish clear pages on what the product is, who it is for and how it compares. Re-run the same prompts every few weeks to catch the first mentions. Compare plans and credits
AI engines answer from what they find on the web and what their models learned. A product that launched last month has little of either.
Without a first measurement you can't tell whether launch content, PR or a directory listing changed anything in AI answers.
A launch site can easily ship with AI crawlers blocked in robots.txt, no structured data and thin pages. Fix that before you publish more.
“Best X for Y” prompts already name established tools. Start with prompts specific to your angle, where the list is shorter.
Start narrow. Your angle and your name first, then the category prompts you want to grow into.
Prompts · fieldnote.app
Track your brand prompt too: when an engine does name you, check that it describes the product correctly.
Five steps you can start before launch day and repeat every few weeks after it.
Run the free SEO, AEO and GEO audit. Check that AI crawlers are allowed, that key pages have schema and that each page answers one clear question.
Free auditRun 10 to 20 prompts on ChatGPT, Claude, Gemini and Perplexity, plus Google AI Overviews and AI Mode, in launch week. Every run is stored, so this becomes your starting line.
AI visibility tracking| Prompt | Engines naming you | Cited |
|---|---|---|
| what is fieldnote | 2/6 | 1 |
| crm that bills per client | 1/6 | 0 |
| fieldnote vs pipewise | 0/6 | 0 |
Use the keyword planner to see search volume, difficulty and whether Google already shows an AI Overview for each question.
Keyword planner| Keyword | Volume | KD | AI Overview |
|---|---|---|---|
| crm for freelancers | 880 | 21 | Yes |
| crm that bills per client | 90 | 8 | No |
| best crm for agencies | 2,400 | 64 | Yes |
Write what engines need to describe you: what it does, who it's for, pricing, comparisons and FAQs. Publish to WordPress or any site through a signed webhook.
AI content writerRe-run the same prompts and compare with the baseline. Opt the ones you are working on into the citation tracker for a weekly re-check.
Citation tracker| Prompt | Week 1 | Week 6 |
|---|---|---|
| what is fieldnote | 2/6 | 5/6 |
| crm that bills per client | 1/6 | 3/6 |
| fieldnote vs pipewise | 0/6 | 2/6 |
Everything a small team needs to go from zero mentions to a measured presence, in one account.
Crawls your site and scores it for search engines, answer engines and generative engines, with the fixes listed. Run it before launch and after every big release.
Site audit
Google search volume, difficulty, the live SERP and whether an AI Overview appears, so you pick questions a new site can win.
Keyword planner
Start with low-difficulty questions that match your angle.
Analyses your site, suggests prompts and topics, researches competitors for the outline, writes the draft and scores it for SEO.
AI content writer
Runs your prompts on six engines and stores every run, so launch week becomes a baseline you can compare against.
Mention rate since launch
Week 1 to week 6. Sample data.
Early wins are small and specific. Measure them against your launch baseline.
Share of runs that name you in launch week, per prompt and engine.
The first prompts where one of your URLs appears in the sources, and on which engine.
Audit scores for SEO, AEO and GEO readiness, release by release.
Whether the answer to “what is [your product]” matches what you actually sell.
The first visits from ChatGPT, Perplexity and other assistants.
44%
Early wins are narrow. Plan for that and measure against the baseline.
Engines that answer from training data may not know a new product for months. Engines that search the web can cite a new page sooner, for specific questions.
Clear, crawlable pages make you easier to describe; they don't force an engine to name you. Track the change across runs.
Runs on ChatGPT, Claude, Gemini and Perplexity don't pretend to be in a country or town, so write the market into the prompt. Google AI Overview and AI Mode checks take a country and language per query.
There is no automatic daily monitoring. Run the prompt set before each review; prompts opted in to the Action Plan citation tracker are re-checked weekly.
Something else? Ask the team
It gives you a dated baseline from launch week, shows which prompts already name competitors, and tells you which pages the engines cite. That turns “AI doesn't know us” into a short list of pages to publish and prompts to re-check.
In launch week, or before. A baseline taken before your launch content goes live is the only way to show later what that content changed.
SEO aims for a position in Google's results. AI search optimisation aims for a mention and a citation inside an answer. The groundwork overlaps (crawlable pages, clear answers, structured data), but you measure it by prompts and engines, not keyword positions.
Yes, for specific queries. AI Overviews cite pages from across the results, so a new page that answers a narrow question clearly can be cited before it ranks for broad terms. Check the queries one by one with the AI Overview tracker.
Look for coverage of the engines your buyers use, stored run history and a way to act on the gaps. AI Rank Lab is our product: it covers ChatGPT, Claude, Gemini, Perplexity, AI Overviews and AI Mode, and adds the audit, writer and publishing in the same account.
Signup includes 100 free credits with no card, enough for about one audit or one prompt run. Paid plans start at $69 a month with 10,000 credits.
Paste your site for the free SEO, AEO and GEO audit, then run your first prompts and keep the result.
