FREE TOOL · MODEL-ASSISTED · ONE READ
TrustGrowth extracts the visible homepage text and runs one bounded LLM pass to show what AI currently understands.
No account. One bounded check — never the full crawl or a site-wide score.
WHAT YOU GET
ILLUSTRATIVEWHAT IT THINKS YOU ARE
"A monitoring product for engineering teams, likely self-serve."
AMBIGUITIES
"Unclear whether this is a product or an agency."
TRUST SIGNALS NOT FOUND
No named team · no dated content
When an AI assistant lands on your homepage, it doesn't see your design — it reads your visible text and forms a working theory: what this business is, who it's for, what it offers. This check does exactly that, once: it extracts up to 15,000 characters of visible homepage text and runs a single bounded model pass, returning the description an AI formed, the audience it inferred, the offerings it found, and — most usefully — what stayed ambiguous and which trust signals it couldn't find.
Answer engines summarize you to people who never visit your site; the framing in that summary comes from how legibly your own pages state it. Search engines run the same inference at scale for topical relevance. One bounded read is a snapshot, not a measurement — but a snapshot that misreads you is usually telling the truth about the page.
The product's AI Identity check runs this read on a schedule inside the visibility funnel's Readable stage, alongside GEO readiness scoring across your crawled pages — so "AI misreads us" becomes a tracked, dated finding instead of a one-off surprise.
Convinced? It takes one URL.
TRY ANOTHER CHECK
It fetches your homepage, keeps only the visible text — up to 15,000 characters — and runs one bounded AI read over it. You get back what the model understood about your business and audience from your own words, plus the ambiguities and missing trust signals it noticed.
No, and that distinction matters. There is no search, no browsing history and no prior knowledge of your brand in this read — only the page. It answers "what do my own words communicate", not "what does the internet think of me".
No. It prints the machine's reading of your page; it renders no verdict on whether that reading matches your intent. A divergence is your call. That is not a gap in the tool — it is the tool declining to have an opinion about your positioning.
One page — the homepage — and one model call. There is no crawl, no follow-up question and no second pass. The budget is strict, which is why it can be free.
The result is cached for about 72 hours so the shareable link keeps working, and the run itself — the URL and the outcome, not your page's content as a document — is retained for up to 90 days for abuse and cost control, then deleted. Result pages are not indexed.
The in-product version reads across your audited pages rather than one, sits beside your own positioning for comparison, and keeps a history you can watch change. This is the same reader, pointed at one page, once.