AI compliance should follow your ML stack, not a folder of documents. Occlave scans what you run, classifies EU AI Act risk, and drafts technical documentation from that inventory.
Pipelines, endpoints, and training data live in platforms like Databricks. Risk registers and technical files usually live somewhere else, updated on a project schedule. Retrain a model and the paperwork is out of date.
Occlave connects to your ML platform, keeps a live inventory of AI systems, maps Annex III high-risk classification, and writes draft technical documentation you can refresh when the stack changes.
occlave — bash
$ occlave scan --platform databricks --regulation eu-ai-act
→ scanning ML pipelines on databricks…
→ 47 AI systems discovered · 12 classified high-risk (Annex III)
→ generating technical documentation (Art. 9–14)…
✓ draft compliance evidence written to ./occlave/For ML platform teams at EU-facing companies running high-risk AI on Databricks.