AI document analysis for the files you have to get through
AI document analysis means extracting structure from unstructured files: fields, obligations, findings, differences and gaps. ChatOTL supports it by attaching documents to a thread and asking targeted questions, with output you verify against the source.
Summarising a document is the easy request. The valuable ones are extraction, comparison and gap-finding — where the answer has a shape you can act on.
Analyses worth asking for
- Structured extraction: "Return a table of party, effective date, term, renewal, notice period."
- Comparison: attach two versions and ask what changed and which changes carry risk.
- Gap analysis: "What's missing relative to a standard document of this type?"
- Data questions: attach a CSV export and ask what stands out and what looks like a data-entry error.
Supported inputs
PDFs, images, plain text, Markdown, CSV, TSV, JSON, XML, HTML and source code — five files per message, 15 MB each. Text formats are extracted inline; PDFs and images are passed to the model as files.
Build a review step in
Always ask for the location of each extracted value — clause, page or row. An answer you can spot-check in ten seconds is worth far more than one you have to trust.
Limitations
- Not a substitute for legal, tax or compliance review.
- Table and figure extraction can be wrong; verify totals.
- Very large corpora need to be worked through document by document.
Frequently asked questions
Can it fill a spreadsheet from many documents?
It can produce a table per batch you attach, which you paste into your sheet. There's no bulk background pipeline.
Can it compare two contract versions?
Yes — attach both and ask for changes grouped by significance. Confirm each change in the documents before acting.
Is document data kept private?
Uploads sit in a private bucket scoped to your account, and the server verifies file ownership before forwarding anything to a model.
Turn a document into a table
Attach it, name the fields, check the references.
Try document analysis