No. Your text is sent securely to an AI model for analysis and discarded right after. We don't write it to a database or log file. Once you close or refresh the tab, it's gone from our side too — there's no history or account to retrieve it from later.
Your document text passes through our server (briefly, in memory, to relay it) and Google's Gemini API (which performs the analysis). We don't share it with anyone else, and we don't keep a copy. See the Privacy Policy for more on how Google's free tier handles request data.
No. Trothix is an AI tool, not a lawyer. It's built to flag things worth a second look and explain a document in plain English — it can miss things, misread context, or flag clauses that are actually standard for your situation. For anything with real money or real consequences riding on it, have an actual attorney review the document.
Google's Gemini model. We keep it on the current free-tier-eligible version and periodically check Google's docs to confirm eligibility, since Google adjusts this over time.
Trothix is tuned for contracts, leases, and terms-of-service documents — the kind of agreement an individual might be asked to sign or click "I agree" on. Residential leases, freelance/independent contractor agreements, and ToS excerpts all work well. Try the sample-document chips on the homepage to see the kind of output to expect.
A few worth knowing about:
• It's a v1 tool — no accounts, no file uploads (paste text only), no saved history.
• It runs on a free-tier AI budget, so during busy periods you may see a "we're getting a lot of traffic" message. Waiting a minute and retrying usually works.
• Long documents are truncated after roughly 18,000 characters to fit within the model's processing limits.
• Like any AI tool, it can occasionally misread a clause or miss something a human reviewer would catch — treat it as a first pass, not a final word.
We don't train anything ourselves. Because analysis currently runs on Gemini's free tier, Google's own terms may allow request content to be used to improve their models — this is standard for free-tier AI APIs generally, not something specific to Trothix. If you're pasting something especially sensitive, strip identifying details first.
The free AI tier we run on has a shared request budget across everyone using Trothix at that moment, not just you. If a lot of people are analyzing documents at the same time, we show a friendly "try again in a minute" message rather than letting the request fail silently. It's not personal, and it's not about how much you've used the tool.