“Which customer contracts have a 60-day termination notice?” is a question about exact words, and the answer has to be complete. StackGrep Engine finds every document that matches, quotes the line, and counts them, so your agent (or your lawyer) can trust the list.
search_collection({ collection: "contracts", query: "([6-9][0-9]|sixty) days.{0,40}(notice|terminat)", regex: true, filter: "type:msa" })
count_collection({ collection: "contracts", query: "limitation of liability", filter: "region:eu" })search_collection(query: "60 days' notice", filter: "type:msa")count_collection(query: "limitation of liability", filter: "region:eu")get_document(id: "initech-msa-2025")We index text. Convert PDFs to text first (most document pipelines already do) and push the text with the PDF's id; your agent can link back to the original.
Yes. Every document can carry metadata, and searches filter on it: type:msa region:eu customer:globex.
Only your organization's API keys and connected agents. Collections are separate per organization, and deleting one removes its documents.
StackGrep Engine is onboarding teams from the waitlist. Bring your documents or your bucket.