The problem
- Extracted data is corrected by hand in spreadsheets passed around by email. A script works on one laptop and nobody else can install it. An idea needs a working version before anyone commits to it, and the data has to stay in-house.
FROM WORKFLOW TO WORKING TOOL
Internal tools & prototypes
Useful software for your own team: a review screen for extracted data, a data-entry app, a working prototype of an idea.
Written scope. Tests on your material. Source code delivered.
Select a step to explore the workflow.
What you get
- Internal tools: simple web apps to review, correct and export data
- Review and data-entry apps built with Streamlit, a Python framework for internal apps
- Prototypes: a working version of an idea, to test it early
- Docker packaging, so the tool runs the same on any computer with Docker
Deliverables
- Source code you own, with automated tests
- Installation on a computer or server you control
- Instructions to run it, and a short handover call
How we measure it
- Automated tests run on every build. The tool is checked on your own sample files against the acceptance checks written into the proposal, and the results are reported, including what fails.
Limits
- No production apps with logins, payments or databases have been shipped yet; we will not sell one as if we had.
- No hosting, no service-level agreement (SLA), no 24/7 support.
- Public consumer apps and mobile apps are out of scope.
Related work
Synthetic dataRead the project: Bank reconciliation that explains itself
Bank reconciliation that explains itself
Bank-statement lines matched to invoices, each with a confidence and a reason; unclear cases go to a review list.
Test results and limits
0 wrong automatic matches on all 8 synthetic sets; on the harder stress sets about half the payments went to review instead of being matched (measured 5 October 2026).
Synthetic dataRead the project: KPI dashboard that checks your data first
KPI dashboard that checks your data first
Sales and expense CSVs checked line by line before any chart, then revenue, margin, cash and overdue invoices.
Test results and limits
171 of 171 planted faults found, 0 false alarms, KPIs equal to a separate reference calculation to the cent on 4 of 4 synthetic sets (measured 5 October 2026).
Bring your workflow into focus.
Send up to 3 samples or describe one process. We reply by email with what is feasible and how we would measure it.