AI & automation
When a tool should ask
How doc2data flags defined inconsistencies for a person to review instead of guessing, what its synthetic tests show, and what its checks cannot prove.
FROM THE STUDIO
Insights
Notes from our own demos and open-source work. Each one says what was measured, on what data, and what we do not know yet.
Internal tools & prototypes
Your documents, processed locally
How doc2data reads PDFs and scans on your own computer by default, and what the optional AI fallback sends and where.
Data, analytics & ML evaluation
A better way to compare models
A single train/test split can make a weak model look good. Repeated cross-validation and a Bayesian comparison give a more careful answer.
R&D & technical writing
From question to reproducible study
Six steps to answer “can this be automated?” before anyone builds, in a way someone else can rerun.
Product
Inside doc2data
An honest status report: what doc2data extracts and checks, its synthetic test results, and what it does not handle yet.
Process
Start with a feasibility check
What happens when you send us 3 samples or a process description, what you get back, and when we say no.
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.