Back to vendors
A

Almanac Health

Also known as: Almanac Copilot, Almanac

Visit site
Healthcare agentprivateVerified 2026-07-08

Research validated clinical AI platform whose Almanac Copilot acts as a Level 1 autonomous EHR agent that retrieves patient data, drafts notes and places orders under clinician approval, grounded in peer reviewed medical evidence.

Almanac Health is a unified clinical AI platform for evidence based clinical decision support, founded by physician researcher Cyril Zakka MD, whose Stanford work introduced retrieval augmented generation to clinical medicine in one of NEJM AI's most cited papers. Its foundation, Almanac Copilot, is a Level 1 autonomous electronic health record agent that assists clinicians at the point of care: it retrieves and summarizes patient data, drafts notes, places orders and surfaces prioritized alerts, dynamically selecting and chaining tools from a predetermined set that includes FHIR functions, a browser, clinical calculators, a Python interpreter and the EHR database. The platform is grounded in peer reviewed evidence rather than a black box model, is free from pharmaceutical advertising, and is governed by strict institutional controls that keep health systems in authority over how AI generated intelligence is used. As a Level 1 agent it prepares tasks on explicit clinician commands and requires the clinician to review and approve every action, and it is validated through research rather than shipped as a finished product, currently undergoing rigorous clinical validation in academic medical center settings. On the EHR QA benchmark of three hundred common EHR tasks it reached a 74 percent task completion rate, matching much larger models. Backed by a $10 million seed round led by F-Prime with General Catalyst and Lightspeed, which brings total funding to nearly $12 million, Almanac Health is building clinical AI whose incentives are aligned with clinicians and patients.

Vendor details

Canonical URL

https://almanac.chat/

Category

Healthcare agent

Funding status

$10 million seed (April 2026) led by F-Prime with General Catalyst and Lightspeed Venture Partners; General Catalyst led the pre seed with Soma Capital. Total funding nearly $12 million. Based in Boston; founded by Cyril Zakka MD.

Company status

private

Use cases & customers

Primary use cases

Point of care clinical decision supportAutonomous EHR navigation and information retrievalDrafting notes and placing orders under clinician approvalEvidence grounded answers across medical specialties

Target customers

Clinicians across medical specialtiesHealth systems and academic medical centersPhysicians at the point of care

Deployment options

Integrated within existing EHR systemsDeployed under institutional controls in academic medical centers

Integrations

Integrates directly into existing electronic health record systems using the FHIR standard for data interchange. The Almanac Copilot agent dynamically selects and invokes tools from a predetermined set that includes FHIR functions, a browser, clinical calculators, a Python interpreter and the EHR database to complete a request, and grounds answers in a vector database of trusted medical knowledge.

Capability coverage

6.0 / 14 capabilities · 43%

Integrations & Tool CallingUpon receiving a query, Almanac Copilot dynamically selects, populates and chains functions from a predetermined set including FHIR functions, a browser, calculators, a Python interpreter and the EHR database, and the platform integrates specialist knowledge, medical tools and EHR data (Almanac Copilot paper arXiv 2405.07896; almanac.chat). Full
Workflow OrchestrationThe system dynamically selects a subset of APIs and chains them in a valid order (select, populate, structure, chain) to complete multi step EHR tasks, though broader cross workflow orchestration beyond a single request is not documented (Almanac Copilot paper arXiv 2405.07896). Partial
Knowledge Grounding & RAGGrounded in peer reviewed evidence via retrieval augmented generation; the founder developed the first RAG system for clinical medicine (published in NEJM AI), and the agent grounds answers in a vector database of trusted medical knowledge and clinical calculators (Business Wire launch; Almanac Copilot paper). Full
Human Oversight & GuardrailsA Level 1 autonomous agent in which the clinician reviews and approves all actions before they are placed, and the platform is governed by strict institutional controls that keep health systems in authority over how AI is used (Almanac Copilot paper; Business Wire). Full
Security, Identity & GovernanceGoverned by strict institutional controls with privacy focused R&D and institutional data kept secure, but no specific security certifications or identity and access control details are published in retrieved sources (Business Wire; HIT Consultant). Partial
Observability & AuditabilityNo observability, audit logging or run tracing features are described in retrieved sources; the published EHR QA evaluation is research benchmarking rather than product observability (Almanac Copilot paper; almanac.chat). Unable to verify
Memory & State PersistenceThe paper explicitly lists retaining context across multiple interactions as a future enhancement toward a Level 2 agent, indicating persistent memory is not yet a first class capability (Almanac Copilot paper arXiv 2405.07896). Unable to verify
Deployment & Data ResidencyDesigned to integrate into existing EHR systems within academic medical center environments under institutional controls with institutional data kept secure, but specific self host or data residency options are not detailed (Business Wire; HIT Consultant). Partial
Prebuilt Agents / Templates / PacksShips specialist grade knowledge across medical specialties and a library of clinical calculators sourced into its vector database, functioning as prebuilt clinical content, though not a marketplace of prebuilt agents or templates (Business Wire; Almanac Copilot paper). Partial
Triggers & Channel CoverageDelivered as a point of care copilot embedded in the EHR; no multi channel triggers, webhooks or scheduled activation are documented in retrieved sources (almanac.chat; Almanac Copilot paper). Unable to verify
Model Flexibility & RoutingRetrieved sources describe Almanac's own approach and benchmark it against GPT 4, Claude 3 Opus and Biomistral, but no model routing, model choice or bring your own model capability is documented for the product (Almanac Copilot paper). Unable to verify
APIs / SDKs / MCP ExtensibilityThe agent uses FHIR functions and internal tool APIs, but no external developer APIs, SDKs or MCP support for third party extensibility is documented in retrieved sources (Almanac Copilot paper; almanac.chat). Unable to verify
Testing, Debugging & OptimizationBuilt on a strong validation discipline, the agent was benchmarked on the EHR QA dataset at a 74 percent task completion rate across three hundred tasks and is undergoing clinical validation in academic medical centers, though a user facing testing or debugging toolkit is not described (Almanac Copilot paper; Business Wire). Partial
Browser / Computer-useA browser is one of the tools the Copilot can select and invoke to fulfill a query, providing limited browser use rather than general computer use (Almanac Copilot paper arXiv 2405.07896). Partial

Recent platform changes

No recent material changes tracked yet.

Pricing

No public pricing; commercial terms are arranged with the company under institutional agreements.

Not disclosed; negotiated institutional agreement

Contact onlyMedium variable cost

Cost watchouts

Pricing, contract structure and any implementation or integration fees are undisclosed and would be set per institutional agreement.

Variable cost rationale

No pricing model is published, so variable cost exposure cannot be determined; enterprise clinical AI is typically sold on negotiated institutional agreements.

Sales call required

Yes — required for paid access

Free / trial

No public free tier or trial; validation via institutional partnerships

Lowest paid plan

None published; contact the company

Key ambiguities

No pricing is published at all; the product is still in clinical validation, so commercial packaging and price are undisclosed.

Verified 2026-07-08

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.