Pricing
We're currently engaging with design partners. Pilot and enterprise pricing finalize after pilot results inform the model.
Design partner
For 1–2 academic medical centers shaping the product.
- ✓Co-development with the founding team
- ✓Full feature access (gap detection, audio, recipient questions, eval)
- ✓Joint publication opportunities
- ✓Required: clinical co-investigator, IRB approval for any real-data work
Pilot
After design partnership matures into deployment.
- ✓BAA-covered infrastructure (Anthropic, Deepgram, R2, Neon)
- ✓Multi-org SSO
- ✓EHR integration (SMART on FHIR — phased)
- ✓Dedicated success manager
- ✓On-prem deployment option (Llama 3.x via Together)
Enterprise
Multi-site academic systems and integrated delivery networks.
- ✓All Pilot features
- ✓Dedicated infrastructure
- ✓Custom prompt-bundle versioning per site
- ✓Quarterly cross-model evaluation reports
- ✓Co-developed publication pipeline
FAQ
Why design partners before commercial pricing?
Pricing should reflect the value of the validated product. Until our pilots show measurable improvement in handoff quality on real institutional data, we'd rather work in design partnership than commit to a number that doesn't match outcomes.
Is on-prem deployment supported?
Not today. The harness is model-agnostic by design (the structured pipeline, calibration layer, and eval harness work with any text-generation model exposed via a chat-completions-style API), but the on-prem path — Together's BAA tier or self-hosted vLLM — is on the roadmap, not currently shipped. We'll prioritize it with a design partner whose deployment requires it.
How do real-data validation tiers work?
See Security. We use a four-tier strategy (synthetic → MIMIC-derived → published corpora → IRB-approved local data) before any prospective deployment.