I manage several HubSpot instances, and getting data from PDF contracts into the CRM is a massive bottleneck.
Standard OCR + Regex was too brittle. So I built Scanny AI.
It listens for Drive webhooks, uses a vision model to extract keys (like "Total Value") regardless of layout, enforces a strict JSON schema, and patches the HubSpot API.
It handles about 5k pages/hour.
Docs and API keys: scanny-ai.com
0 comments
—
A vision-based PDF parser to validate contracts in Google Drive · HackerTrans
Standard OCR + Regex was too brittle. So I built Scanny AI.
It listens for Drive webhooks, uses a vision model to extract keys (like "Total Value") regardless of layout, enforces a strict JSON schema, and patches the HubSpot API.
It handles about 5k pages/hour.
Docs and API keys: scanny-ai.com