We are rolling out Heidi II, an autonomous agentic layer that handles administrative tasks around patient visits, moving from passive documentation to actually completing tasks.

Heidi II adds native Agents, Memory, and Peer-Reviewed Research to carry tasks through to completion, not just hand them off to physicians.

Clinical governance centers on safety, with physician-in-the-loop oversight, explicit clinician review and approval for decisions, and ongoing internal safety reviews including risk assessments before and after deployments.

The platform now supports about 2.8 million patient interactions weekly across 110 languages in 190 countries, with a proven track record of over 175 million visits and 400 million follow-on tasks since February 2024.

Initial rollout prioritized English and French, and deliberately withholds agentic capabilities from the UK and EU to align with regional regulatory readiness before wider international deployment.

The system employs continuous adaptive memory to learn clinician preferences and formatting, and can navigate across 4 to 6 disparate systems (like EHRs, calendars, and patient portals) to read data and draft communications, clearly signaling when a task cannot be completed.

Embedded peer-reviewed medical evidence from NEJM, Wiley, Cochrane, and BMJ Group provides integrated clinical decision support with linked full-text articles.

Security and compliance are foundational, with certifications including ISO 27001, SOC 2 Type II, Cyber Essentials Plus, ISO 42001, HIPAA, and Australian Privacy Principles.

Key features include a context-aware Home Dashboard that pre-loads schedules and patient data, plain-language autonomous agents that draft completed tasks for clinician sign-off, and reusable routines that run automatically across patient panels.

Thomas Kelly, M.D., co-founder and CEO of Heidi, notes that the system creates tasks and completes workflows while clinical judgment remains with the clinician.