How private in-browser document AI works
You add a document, pick a workflow, and get structured results. Standard mode runs in your browser: the file is not uploaded. Deep Analysis is optional when you want a stronger server-side model. No account is required for the free version. See private AI document analysis pricing if you need Deep Analysis volume.
Add your document
Paste text or upload a PDF, plain text, or Markdown file. Once the document is in the app, Standard mode processes it locally so we can find the right passages. With Deep Analysis we encrypt the request to our API and run a stronger model via privacy-constrained providers listed in the Trust Center.
Summarize
Get a structured overview and key points instead of a generic paragraph. Use this when you need to understand a long contract, policy, or report quickly without reading every page first.
Review
Surfaces items that may need attention, with evidence from the document. Use this when you want potential issues, obligations, or unusual clauses called out rather than a high-level summary.
Extract
Pulls dates, amounts, parties, obligations, and other fields into tables. Use this when you need structured data from a document rather than prose.
Compare
Place two documents side by side and highlight what changed. Useful for contract versions, policy revisions, or redlines you do not want to send to a public chatbot.
Timeline
Build a chronology across up to three files in a session. Use this for matters, projects, or reports where dates and sequence matter.
Ask
Question the document and get grounded answers that point back to the source. For longer texts, PIA builds a lightweight search index so the model sees the most relevant passages instead of the whole file. In Standard mode that indexing happens in your browser too.
Standard vs Deep Analysis
In Standard mode the AI runs entirely in your browser. Your document never leaves your device. We use open models that load and run locally, so even the content sent to the model stays on your machine. There is no server logging your text or training on it.
If you use Deep Analysis, we encrypt your request to our API, decrypt it in memory, and send the needed text over TLS to privacy-constrained AI providers. We do not store your document or prompts. Providers process content for the request under their published policies; we do not use it for training. Refreshing the page clears everything from the app.
In short: add a doc, pick a workflow, get a result. Your data stays private, and when you close or refresh, it’s gone. See who this is for if you work with contracts, reports, or unpublished drafts.