DAGL

Build the check on a case you know, then run it on every row.

Follow one supplier from the first step to a checked answer. Then run the same check on every supplier on your list, with the cost shown before it starts.

Try it on your data See use cases →
01

Start with a case you know.

Pick one real row and note the answer you expect. Everything you build is tested on it first.

Row 2 · pinned case
SupplierNorthwind Foods
CertificateBRCGS food safety
You expectApprove
02

Build one step at a time.

Start, fetch, read, ask AI and send. Each change is a new version, so you see which change helped.

  1. StartOne row from your supplier list
  2. FetchThe certificate from the supplier folder
  3. ReadThe standard, the sites and the expiry date
  4. Ask AIDoes it meet our supplier policy?
  5. SendThe answer back to your supplier system
03

Check it against your expert.

Put your expert's answers in the next column. DAGL counts where they agree, and shows the step that went wrong where they do not.

How you know the answers are right →
46 of 50 cases agree with your expert example data
04

Know the cost before you run.

See how many rows it will read and what it will cost. Nothing is sent until you start it.

What AI costs per answer →
Rows to read412
Answers already saved0
Estimated cost$0.84
▶ Run 412 rows
05

Run it on every row.

Open any answer to see what it read, the instructions it followed and the version that made it.

Alder & SonsReject
Read
FSSC 22000 certificate, page 1
Found
Expiry date 15 December 2025
Instructions
Supplier policy check, version 5
AI
[Model name] · $0.002
06

Keep it right.

VersionsCompare two versions on the same cases, and go back whenever you need to.
When the AI changesRecheck your saved cases on the new version before you switch.
On a scheduleRun the check every day, week or month without anyone starting it.
Where results goSend answers to the tools your team uses today.

Who does what.

The expertReads real cases and improves the instructions in plain words, without code.
The builderConnects the data and sets up each check once.
The ownerSets the budget and approves what goes live.

Bring one check your team does by hand.

Start with one real case and see the working behind every answer.

Try it on your data