The table that keeps itself current.
Certificates, schedules and compliance packs arrive on somebody else’s timetable, and the spreadsheet tracking them is out of date the moment one does. Put the table in auto mode and each new document is answered as it lands, with every answer citing the page it came from.
Free credits on sign-up. No card, no subscription.
| Row | Documents | Cover | Limit | Expires | Named insured matches |
|---|---|---|---|---|---|
| Meridian Steel — EL certificate.pdf | Employers’ liability p.1 | £10,000,000 p.1 | 31 Mar 2027 p.1 | Yes p.1 | |
| Falcon Fabrication — PL certificate.jpg | Public liability p.1 | £5,000,000 p.1 | 14 Nov 2026 p.1 | No — trading name only p.1 | |
| Kestrel Structures — PI certificate.pdf | Professional indemnity p.1 | £2,000,000 p.1 | 1 Aug 2026 p.1 | Yes p.1 | |
| Ashfield Erectors — schedule.pdf | Public liability p.2 | Not stated | 30 Sep 2026 p.2 | Yes p.1 |
Retyping is where the errors come from.
Nobody misreads a certificate. What happens is that a limit gets keyed into a spreadsheet with a digit missing, six months before anyone needs it — and by then the certificate is in a folder nobody can name.
Every cell here carries the page it was read from, so the check is a click rather than a search through an inbox. Corrections are recorded rather than overwritten: type over a cell and the original stays underneath, with who changed it and when.
| Document | Expires | Limit |
|---|---|---|
| Harlow Plant — EL certificate | 12 Feb 2027 p.1 | £10,000,000 p.1 |
What auto mode does, and what it does not.
A table in auto mode runs itself against every new document added to its workspace. That is the whole behaviour, and for a stream of arriving paperwork it is the difference between a table that is current and a task somebody has to remember.
It is not a reminder system. Ragextract will extract an expiry date and cite the page it read it from; it will not watch that date, raise an alert, chase a supplier or escalate anything. What you get is a column you can sort and export — the acting on it stays wherever acting on it happens today.
Scanned and photographed documents are read too — JPEG, PNG and WebP, each as a single page — which matters here more than anywhere else on this site, because a certificate is as likely to arrive as a phone photo as a PDF.
Where it fits.
A row is one arriving document. If what arrives is a batch that belongs together, that is a bundle rather than four rows.
| The work | What a row is | Start from |
|---|---|---|
| Supplier compliance packs | A certificate or policy schedule | |
| Accreditation and licence tracking | An accreditation certificate | |
| Inbound report processing | A report, as it arrives | |
| Vendor onboarding files | A submitted document |
Every column here is one you write — no preset ships for arriving paperwork, because the fields are whatever your process already checks. The nearest worked setup is subcontractor trade bids, which shows the shape over short documents, and columns covers picking the output type so a limit sorts as a number and an expiry sorts as a date.
Priced by what arrives.
A stream has no project cost, only a monthly one — and it is the pages that arrive, not the table watching for them. A table sitting in auto mode with nothing to do costs nothing at all.
| A month of arrivals | Volume | Cost |
|---|---|---|
| 200 certificates a month, 4 pages each | 800 pages | 800 credits |
| 4 questions asked of each as it lands | 800 cells | 800 credits |
| The table itself, once built | — | Free |
Start with last month’s arrivals.
Upload twenty of them, write the four columns you retype today, and compare the table against the spreadsheet you already keep. Where they disagree is the interesting part, and it is usually not the table that is wrong.
Ragextract extracts and cites what a document says. It does not monitor dates, send reminders or chase anyone.