Which file types can Lumipact read?+
PDF and Word documents, plus scanned contracts as images — OCR is built in, so a photographed or scanned agreement is readable too. Files arrive by bulk upload, CSV import for existing registers, or email forwarding to your workspace's ingest address. There is no Google Drive, Dropbox, or SharePoint sync today.
How does the AI extraction work?+
You upload a contract; the AI reads the document and fills in the counterparty, dates, value, renewal terms, and notice period. Each extracted field links back to the passage in the document it came from, so checking a value takes one click rather than a re-read. You review and correct rather than type from scratch.
How accurate is the extraction?+
Accurate enough to draft the record, not to replace your review — which is why every extracted field links to its source passage and the workflow asks you to confirm before the record stands. Quality tracks the document: a clean digital PDF extracts better than a skewed scan. We publish no accuracy percentage, because a single number would hide exactly that variation.
Who can see which contracts?+
Access is role-based and enforced in the database itself, not just hidden in the interface. Contracts can be scoped by legal entity or business unit, so a subsidiary's team sees its own agreements while roles with group-wide access see across entities. Every access is logged, and the full permission matrix is published in the docs.
Can Lumipact run on our own AI key?+
Yes. By default extraction runs on Lumipact's own Anthropic account, and your contract text is never used to train models. If your security review prefers it, connect your own Anthropic API key instead — the AI calls then run under your account, on your terms, with usage visible in your own console.