// data-use
Data use
Last updated
How TOntoN handles the source material you bring into the platform. This complements the Privacy notice with a more specific look at the Vision pipeline itself — what gets crawled, what gets sent to the LLM, and what stays inside your workspace.
Public-source only
TOntoN analyses public material: company websites, blog posts, press articles, careers pages, and the founder / company LinkedIn URLs you explicitly provide. We do not scrape behind paywalls, log-in walls, or rate-limit blocks. We do not ingest private documents, leaked material, or anything you don't have the right to process.
How the crawler behaves
When you add a company with a root URL, the crawler fetches up to ~20 pages from that domain. It respects robots.txt directives and standard rate-limiting. For larger or paywalled sites, you add specific pages as manual sources (paste-text, paste-link, or upload PDF / DOCX / MD / TXT).
What gets sent to the LLM
After source selection, the cleaned text of the sources you confirmed is sent to OpenAI in a structured prompt. The model returns scored evidence quotes per Vision-ontology node. We do not send your account email, organisation name, or other workspace metadata to the LLM. Per OpenAI's API terms we do not opt into training; OpenAI retains the request content for up to 30 days for abuse monitoring, then deletes it.
Where outputs live
Extracted evidence, scores, and reports live in your workspace in our PostgreSQL database (hosted by Railway). Access is workspace-scoped — no other tenant can see your data. Workspace isolation is enforced at the API layer (require_workspace_* dependencies, see MR-0 of the public roadmap) and verified by an automated test suite on every deploy.
Generated PDFs + downloads
When you download a Vision report PDF, the file is generated on the server and streamed to your browser. We do not push your reports to any third-party storage. If you share a download URL, the recipient still needs your auth token (the API rejects anonymous requests), so the link itself is not a leak.
Take-down + correction
If you believe TOntoN has analysed material that should be removed — for example, content the original owner has retracted or material the source asks us to delete — email us at hello@teonto.com with the URL or company name. We will action valid take-down requests within 14 days and clear the affected sources from active analyses.
Contact
Questions or requests, reach us at hello@teonto.com.