The problem
- “Can this be automated?” needs a measured answer before anyone builds. Prior work is scattered and nobody has checked which sources hold up. Code has no fixed seeds, configuration or documentation, so results cannot be reproduced.
FROM WORKFLOW TO WORKING TOOL
R&D & technical writing
For questions that need research before code: can this be automated, what has already been tried, how do we write it up so others can reproduce it?
Written scope. Tests on your material. Source code delivered.
Select a step to explore the workflow.
What you get
- Feasibility prototypes: a small, measured experiment that answers one question
- Prior-art and literature reviews: what exists, what works, what does not, with every source linked
- Research software: reproducible code with fixed seeds, configuration files and documentation
- Technical reports and documentation: methods, results and limits written down clearly
Deliverables
- A written report with dated, linked sources
- Code and configuration that regenerate every table and figure
- A list of assumptions and open questions
How we measure it
- Every claim is sourced and dated; every number comes from code we actually ran; assumptions and gaps are marked as such. A script regenerates the report’s tables and figures so someone else can check them.
Limits
- No ghost-writing of theses, assignments or academic papers submitted under someone else’s name.
- No regulatory sign-off, for example for medical devices. We can support a qualified person; we do not replace one.
Related work
Synthetic dataRead the project: doc2data
doc2data
Invoice and receipt PDFs and scans to checked, structured data.
Test results and limits
0 field errors on 3 clean synthetic sets of 16 documents; 92.9% (131 of 141) and 91.4% (127 of 139) of fields on degraded synthetic scans (measured 2 October 2026).
Insight
A practical guide to a reproducible feasibility study
Six steps to answer “can this be automated?” before anyone builds, in a way someone else can rerun.
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.