Pilot programme
Run a pilot on your own documents
A short, measured trial on documents you actually process, ending in a written report you can take to a decision.
How the pilot runs
1. Scoping call
Thirty minutes to agree which document type, which system receives the data, and what success looks like in numbers.
2. Sample set
You send 50 to 200 representative documents, including the awkward ones. Anonymised is fine.
3. Schema build
We build the extraction schema and validation rules with you, then publish it as a reusable template.
4. Live run
The set is processed in your own workspace. You review exceptions in the queue exactly as you would in production.
5. Delivery test
Results are delivered to your endpoint or automation platform over the API and webhooks, with retries observed.
6. Report and decision
We hand over the pilot report and agree whether to roll out, extend or stop. No obligation either way.
What the pilot report contains
- Field-level accuracy per document type, measured against your corrections
- How many documents passed without a human touching them
- Every exception, why it was flagged, and what would prevent it
- Time and cost per document, compared with your current process
- The integration recommendation, including what your side needs to build
- The reusable schema and template, yours to keep