Sam Witteveen · 2026-10-02 · notable
Sam Witteveen — 'Image Decision Models for RPA: Forms, Scans and Screenshots'
Sam Witteveen's 2 October 2026 video applies decision models to images for RPA, demoing a form inspector built on two open models, ImaJev 4B and Jev Omni, with per-question confidence thresholds.
Sam Witteveen shows how small image decision models can check forms, scans and screenshots inside RPA workflows.
What is it?
'Image Decision Models for RPA: Forms, Scans and Screenshots' went up on Sam Witteveen's channel on 2 October 2026. The video returns to decision models, this time for images, with robotic process automation (RPA) as the use case.
How does it work?
The first part explains that RPA automates steps but not decisions, why custom deep learning models fell short, and looks at the ImageBench leaderboard and the story behind ImaJev. The demo is a form decision inspector that runs ImaJev 4B and Jev Omni side by side, compares choice and true/false questions, drafts follow-up emails with conditional logic, and sets a confidence threshold per question.
Why does it matter?
Checking screenshots, photos and scanned forms is the step that often breaks RPA pipelines. Treating each check as a 'smart if statement' with a confidence score lets builders route uncertain cases to a person instead of training a custom model for every form.
Who is it for?
automation and agent developers