Tablefire
A template-driven document extraction platform for accountants, bookkeepers, and auditors in the US market.
What it does
Users define a reusable template once: text anchors for automatic document-type detection, extraction regions, normalization rules, and export mapping. The platform then applies it to every matching document.
Per-field OCR confidence scores route low-confidence values to human review, and a stage-by-stage audit trail makes every exported figure traceable back to its source location in the original PDF.
Verification, not extraction
The product's differentiator is not extraction but verification. Accounting professionals can't sign off on a number they can't trace, so every stage of the pipeline is auditable by design.
Amazon Textract does the reading. The template engine, the confidence routing, and the audit trail are the layer that turns its output into something a professional can rely on.
Role
I am the sole engineer. I also defined the product, its positioning, and its pricing model.