Hey brains trust and Convex team - I’m building a medical research training platform on Convex (cothesis.ai). AI-powered mentorship for medical trainees who need to complete research projects but typically have no methodology training or adequate supervision. Fairly complex data model with reactive aggregated views (Gantt, calendar, Kanban, file manager etc), which is the main reason I picked Convex.
Four questions as I plan the build:
1. HIPAA BAA - the $0 BAA on any plan was a significant factor in choosing Convex. Is there any risk this moves to a paid tier as pricing evolves? Bootstrapped solo founder, so this matters for planning. 2. Plate/Slate editor support - I have a Plate Plus lifetime licence (Slate-based, shadcn native, includes track changes). But @convex-dev/prosemirror-sync only supports TipTap/BlockNote. Any plans for a Slate/Yjs equivalent, or is ProseMirror the direction? This is my biggest architecture decision right now. 3. Dynamic document structures in a static schema - I need documents whose effective shape varies per project based on external configuration (different research methodologies, institutional requirements, compliance frameworks). The config is known at setup time, not truly runtime-dynamic. What's the recommended pattern here: wide schema with optionals, separate config documents, JSON blob, something else? 4. Cross-document aggregation - will regularly need to pull specific data types from hundreds of related documents into aggregated views (scheduling data → Gantt, file metadata → file manager, etc). Recommended patterns, or should I plan to denormalise heavily?
Happy to go into more detail privately - several core architecture structures are covered by provisional patents, so keeping the public description general deliberately.
Based in Perth, Australia, so timezone may be interesting. Cheers and thanks for any suggestions/ideas. Dave