Back to vendors
C

Corti

Also known as: FactsR, Corti Assistant, Corti Agentic Framework

Visit site
Healthcare agentindependentVerified 2026-09-29
Entry pricePay as you go, with speech to text at $0.0065 per minute and agents, coding and text generation at $4 and $16 per million input and output tokens. $50 of free credits to start.Full pricing detail

AI infrastructure for healthcare developers: clinical speech to text, medical coding, text generation, EU-hosted Corti Models and an Agentic Framework with connectors, traces and a 20+ agent library, through one API with published usage pricing.

Corti sells a developer platform for healthcare AI, marketed as Symphony, delivered through one API with usage-based credits. It covers medical speech to text, a medical coding API that returns structured predictions with supporting evidence and an audit trail, clinical text generation, and Corti Models, OpenAI-compatible model tiers built on open-weight foundation models and served from a Corti-operated European cluster (Azure-hosted OpenAI models for US projects).

The Agentic Framework runs agents on a runtime that coordinates connectors (a pre-built registry, customer MCP servers, JSON Schema tools, other Corti agents and remote A2A agents), keeps per-encounter contexts with delete and trace-export endpoints, enforces guardrails, and exposes OpenInference traces. An Agent Library offers 20+ pre-built agents such as ICD-10 navigation, code validation, prior authorization and medication reconciliation, and an Agent SDK composes multi-agent workflows. Projects choose EU or US regions, with France and Denmark on request and sovereign in-country or on-premises deployment for enterprise. The same API powers Corti Assistant, an ambient scribe.

Vendor details

Canonical URL

https://www.corti.ai

Category

Healthcare agent

Funding status

$60M Series B; Copenhagen based

Company status

independent

Use cases & customers

Primary use cases

healthcare agent developmentmedical speech to textclinical documentation generationmedical coding APIreal time clinical fact extraction

Target customers

healthtech developersEHR vendorsvirtual care platformshealth systems

Deployment options

SaaSsovereign cloud options

Integrations

REST API with JavaScript and C# .NET SDKs, an Agent SDK, web components and A2A agent cards; agents connect to registry connectors, customer MCP servers, schema tools and other agents; EHR-agnostic.

In practice

You are building a prior authorization tool for a health system. You start from the Agent Library's Prior Authorization and Code Validation agents, connect your own MCP servers, and compose the steps into a deterministic workflow with the Agent SDK.

Your patient data must stay in Europe. You set the project to the EU region, Corti Models processing stays in the EU, and for enterprise a sovereign cloud deployment can run in country or on premises.

Your compliance team asks what an agent did with a chart. Each encounter's context exports OpenInference traces of every LLM call, connector invocation and tool use, and the coding API returns supporting evidence and an audit trail with every call.

Sources & related URLs

Agentic Index coverage score

9.5 / 14 capabilities · 68%

Integrations & Tool Calling Full

Agents take tools through a unified connector model that covers pre-built registry connectors, remote MCP servers the customer brings, other Corti agents, remote A2A agents and custom JSON Schema tools, each with configurable connector authentication.

Sourcedocs.corti.ai/agentic/connectorsread 2026-09-29
Workflow Orchestration Full

An agent runtime coordinates connectors, contexts and sub-agents, and the Agent SDK composes multiple agents into deterministic workflows, parallel fan-out and state-graph routing.

Sourcedocs.corti.ai/agentic/architectureread 2026-09-29
Knowledge Grounding & RAG Partial

Agents reach the customer's knowledge through connectors at run time, and the coding API returns supporting evidence per call, with an embedding model offered as a building block. The platform has no stated maintained retrieval structure over the customer's knowledge.

Sourcedocs.corti.ai/agentic/connectorsread 2026-09-29
Human Oversight & Guardrails Partial

The agent runtime enforces guardrails, type validation and policy-driven constraints that the customer configures, and a Compliance Guardrail Agent is in the library. There is no stated approval step before an agent action commits.

Sourcedocs.corti.ai/agentic/architectureread 2026-09-29
Security, Identity & Governance Partial

Only project admins or owners can create, update or delete resources, with an Admin API for users and customers and bespoke private identity tenants for regulated cases. Corti describes itself as HIPAA-ready, and named attestations appear only in the trust center at trust.corti.ai.

Sourcedocs.corti.ai/about/admin-apiread 2026-09-29
Observability & Auditability Full

Each context exports OpenInference-format traces with spans for LLM calls, connector invocations and tool usage through a trace endpoint, and the coding API returns a complete audit trail with every call.

Sourcedocs.corti.ai/agentic/context-memoryread 2026-09-29
Memory & State Persistence Partial

Contexts are first-class resources grouping messages, tasks and artifacts for one patient encounter, call or workflow, with endpoints to get, list tasks, export traces and delete a context. They have a set scope and a delete path, but no stated lifetime or retention.

Sourcedocs.corti.ai/agentic/context-memoryread 2026-09-29
Deployment & Data Residency Full

Each project chooses its region. EU (Azure West Europe) and US (Azure US East) are public, France (Scaleway Paris) and Denmark are available on request, and Corti Models EU processing stays in the EU. For enterprise, sovereign cloud deployments run in-country, on any cloud or on premises.

Sourcedocs.corti.ai/authentication/environments_tenantsread 2026-09-29
Prebuilt Agents, Templates & Packs Full

An Agent Library of 20+ pre-built agents and connectors, among them ICD-10 Index Navigator, Code Validation, Compliance Guardrail, Medication Reconciliation, Denial Appeals, Prior Authorization, Referral Generator and Nursing Shift Handoff agents, each doing its own job.

Sourcecorti.airead 2026-09-29
Triggers & Channel Coverage Partial

Agents run when the builder's application sends an A2A message or opens a stream, with real-time audio over WebSocket for scribe and dictation. Push notifications on the agent card are marked future scope, and the platform has no stated event, schedule or webhook trigger.

Sourcedocs.corti.ai/agentic/agent-cardsread 2026-09-29
Model Flexibility & Routing Partial

Customers pick among Corti's own model tiers (corti-s1, s1-instant, s1-mini, s1-mini-instant). Corti discloses the foundation models behind them (GLM 5.2 and Qwen 3.6 in the EU, Azure-hosted OpenAI models in the US) and swaps them behind the same IDs. The customer chooses a Corti tier, while Corti routes the underlying provider.

Sourcedocs.corti.ai/models/modelsread 2026-09-29
APIs, SDKs & MCP Extensibility Full

Corti offers a REST API with client-credential authentication, JavaScript and C#.NET SDKs, a TypeScript and Python Agent SDK, web components, a Postman collection and an Admin API, and serves A2A agent cards per agent with endpoint URLs and bearer authentication.

Sourcedocs.corti.ai/agentic/agent-cardsread 2026-09-29
Testing, Debugging & Optimization Partial

Traces are offered for debugging and performance analysis, and Corti validates each model release against its own evaluation suites. Those suites test Corti's own models, and there is no stated evaluation harness, test cases or quality gate for the customer's agents.

Sourcedocs.corti.ai/models/how-models-workread 2026-09-29
Browser & Computer Use Not documented

Agents act through API connectors, MCP servers and A2A agents, with no stated browser, desktop or computer control.

Sourcedocs.corti.ai/agentic/connectorsread 2026-09-29

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Product updates

·Agent capabilityPartially Verified

Updated German and French speech to text models, including the Swiss variants, are now available from Corti. Measured on 10 German and 6 French datasets with the method from Corti's May 2026 benchmark paper, word error rate fell 18.75% for German and 10.46% for French relative to the previous models, and medical term recall rose 4.35% and 5.23%.

Bears on: Agent capability

View source
·Human approval / guardrailsVerified

Corti updated Guided Generation to pass user edited, added, and discarded facts with explicit type metadata. User edits now take precedence over conflicting transcript content, and discarded facts are excluded from generated notes. User added facts and system supplied context also receive explicit priority.

Bears on: Human approval / guardrails

View source
·Deployment / data residencyPartially Verified

Corti partnered with the Danish Center for AI Innovation to launch a Sovereign AI Control Layer for European enterprises. Trifork was announced as the first implementation partner for this new infrastructure.

Bears on: Deployment / data residency

View source
View all 3 Corti updates →Tracked since Aug 2026 · Verified from public vendor sources

Pricing

Pay as you go, with speech to text at $0.0065 per minute and agents, coding and text generation at $4 and $16 per million input and output tokens. $50 of free credits to start.

Credits from one balance, with audio billed per minute and text per million input and output tokens.

Trial available

What is public

Corti publishes rates per unit for every API. Speech to text costs $0.0065 a minute, and coding, agents and text generation $4 and $16 per million tokens. Corti S1 costs $2 and $8, and S1 Mini $1 and $4, with cached input at $0.20 and $0.10. The Acceleration Pack costs $1,000 a month, and Corti Advance and Enterprise terms are set in conversation.

Variable cost rationale

Corti meters audio minutes and tokens. Speech to text bills per minute and the language APIs per million input and output tokens, all drawn from one credit balance, so the bill grows directly with the volume of audio transcribed and text processed. No spending cap is published, and usage continues at pay as you go rates once the free credits run out. The published rates make spend estimable from a team's own volume, and the Acceleration Pack and Enterprise offer bundles in place of pure pay as you go.

Overage / add-ons

After free credits run out, usage continues automatically at pay as you go rates from the same balance.

Sales call required

Mixed (some tiers require a call)

Free / trial

$50 of free credits, no card required, valid for one year

Lowest paid plan

Pay as you go from $0.0065 per audio minute. Acceleration Pack at $1,000 per month.

Cancellation / refund

No upfront commitment or minimum spend on pay as you go. Corti Advance is a rolling monthly commitment.

Support SLA / resale

The Acceleration Pack adds 5 development hours a month with an architecture review, and Corti Advance adds embedded engineering support.

Agentic Index verified 2026-09-29
Official pricing page

Related vendors

Alternatives to Corti

The closest documented capability profiles to Corti among healthcare agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

  • Owkin11.0 / 14Fuller documented coverage on Knowledge Grounding & RAG and Security, Identity & Governance
  • Prosper AI10.0 / 14Fuller documented coverage on Human Oversight & Guardrails and Triggers & Channel Coverage
  • Microsoft Dragon Copilot9.0 / 14Fuller documented coverage on Human Oversight & Guardrails and Security, Identity & Governance
  • Sully.ai9.0 / 14Fuller documented coverage on Security, Identity & Governance and Triggers & Channel Coverage
  • SuperDial9.0 / 14Adds documented Browser & Computer Use
  • Cedar7.5 / 14Fuller documented coverage on Triggers & Channel Coverage

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Compare Corti with other vendors →Compare all healthcare agents →See where Corti ranks →
Data confidence: high

Head to head

Corti vs PhonicCorti vs Innovaccer

Contact us

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