Bring in what you already know.
Drop in files, read a Postgres query, or connect any MCP server. Contextely keeps the original reachable and remembers where each value came from.
- Drop in filesPDF, DOCX, Markdown and CSV. Nothing to connect.handbook.pdf, 14 pages, 31 records proposed
- Read-only PostgresYour query, read-only, refreshed on a schedule.finance.projects, 212 records, synced 4 minutes ago
- Any MCP serverPoint at a server and choose the tool that lists records.Checked: re-read one record and it matched
Drag and drop
PDF, DOCX, Markdown and CSV. Contextely proposes a collection and a property mapping, and you see the additions before they are saved. Importing the same file again updates rather than duplicates.
Read-only Postgres
Give it a query. It only reads, on a schedule, and checks on save that it can re-read the same record it listed.
Any MCP server
Point at a server and choose its tools for listing and fetching records. Contextely confirms the tool returns the record you asked for.
Every value knows where it came from.
A value is written by a person, synced from a source, or derived by AI from evidence. Each carries its author, its source and when it was last checked.
Synced values refresh on their own. If someone corrects one by hand, the correction stays and is marked as different from the source, with a one-click way back.
- OwnerMei TanakaSynced from Postgres finance.projects, 4 minutes agoFresh
- StatusIn progressEdited by Dana Okafor, differs from source, YesterdayDiffers from source
- SummaryMove payroll to the new provider by JanuaryDerived by AI from payroll-plan.pdf, page 3, 6 days agoDerived
Fresh, or labelled as not
Records from a source can go stale and are re-read when they do. Pages you write yourself can carry a review date, and show "needs review" when it passes. Nothing is presented as current when it is not.
Start with what you already have.
Free for up to five editors. Unlimited readers on every plan.