COPENHAGEN – 22 October 2026 – Corti, the frontier AI lab for clinical-grade AI, today announced a major expansion of its Agentic Framework, introducing a new execution and control layer to address the growing complexity of multi-agent systems in healthcare.
As agentic AI becomes increasingly embedded in healthcare software, the demands on the underlying infrastructure are growing. Corti brings together clinical-grade AI orchestration, connectivity, observability and governance in a single platform, giving healthcare software developers the infrastructure to build and manage their own AI agents and applications, without having to develop and maintain it themselves.
Corti's Agentic Framework is already being used by Dedalus, a leading global provider of healthcare and diagnostic IT solutions, to bring agentic AI capabilities to Electronic Health Record (EHR) systems across Europe. ORBIS Buddy, a fully agentic layer for Dedalus’ ORBIS EHR, is built on the Corti Agentic Framework and Speech Technologies, which moves the ORBIS EHR beyond information management toward a deeply embedded, integrated AI workflow platform, where clinicians direct the system in natural language, by voice or text, and AI agents carry out the work across the record.
As Agentic AI Evolves, Managing Complexity Becomes The Challenge
AI agents are evolving from executing single tasks into interconnected systems capable of collaborating, accessing data and executing actions across applications. IDC forecasts 1.2 billion AI agents operating globally by 2029, performing 217 billion actions every day. This shift introduces new complexities and risks around managing and governing agent behaviour. Gartner predicts that by 2027, 40% of enterprises will demote or decommission autonomous AI agents because of governance gaps discovered after production incidents. In interconnected systems, errors can cascade across agents and workflows, making it harder to understand what went wrong and intervene before consequences escalate.
In healthcare, the shift from systems of record to systems of action raises the stakes. As AI agents execute workflows across patient records and clinical systems, developers must monitor agent interactions, trace failures, evaluate performance and maintain human oversight across interconnected workflows. Uncontrolled automation introduces significant liability. Without the right infrastructure, building and maintaining these safeguards adds to the development burden and slows deployment, while meeting established standards for patient safety, compliance and accountability can become out of reach. Corti's new execution and control layer addresses these challenges, giving developers the visibility and tools needed to manage complex clinical workflows.
“The accelerating adoption of agentic AI in healthcare is fundamentally changing how healthcare software is built and operated. As AI agents become more interconnected and clinical workflows more complex, managing and governing these systems reliably and at scale requires a new level of infrastructure. Corti provides the execution and control layer, removing that burden from developers and making it easy for them to build the next generation of healthcare AI. Dedalus demonstrates this in practice, using Corti's Agentic Framework as the foundation for ORBIS Buddy to transform the EHR from a system of record into a system of action,” said Lars Maaløe, CTO and Co-founder of Corti.
The Execution and Control Layer for Healthcare AI
Corti's expanded Agentic Framework brings together capabilities across three key areas, including new features designed to help developers manage increasingly complex AI systems:
- Intelligent execution and connectivity: A central orchestrator coordinates specialised AI agents, maintains context and connects them to clinical systems and trusted medical knowledge. New Schema Connectors support structured data, building on existing MCP, A2A and sub-agent connectors. Token-by-token streaming provides real-time visibility into agent activity and allows users to interrupt tasks when necessary.
- Governance, traceability and evaluation: Built-in guardrails and human approval mechanisms support oversight. New trace APIs help developers investigate failures and debug workflows, while an open-source evaluation library and Feedback API enable systematic testing and continuous improvement.
- Sovereignty and flexibility: Corti's broader AI stack gives organisations choice over models, infrastructure and deployment, including on-premises environments, while supporting integration with existing healthcare systems.
Together, these capabilities give developers the infrastructure to execute, monitor and govern the AI agents they build, while focusing their resources on developing their own healthcare applications.
From Systems Of Record To Systems Of Action With Dedalus’ Orbis Buddy
Electronic health records bring together the full clinical picture, from patient history and medications to results, notes, orders, alerts and tasks. But accessing and acting on that information still means navigating multiple modules, views and forms, adding up to significant time spent operating software instead of caring for patients.
Dedalus built ORBIS Buddy in collaboration with Corti to meet that challenge, expanding AI from documentation and information management into agentic workflow support. It sits on top of ORBIS as an intelligence layer. In early use, Buddy is helping clinicians access patient information more quickly and interact with ORBIS through natural language, reducing the navigation typically required across multiple screens and modules. It also supports the creation of structured clinical documentation, writing back information into the system as and when needed, helping to ease administrative burden while keeping clinicians in control of every entry.
Behind that simple experience is a production-grade multi-agent system, underpinned by the Corti Agentic Framework and built from highly configured agents that understand clinical workflows and meet the level of clinical robustness that hospital care requires. A central orchestrator coordinates specialized AI agents, each handling a specific part of ORBIS or drawing on an external clinical knowledge source like Thieme and Elsevier, and connects them through open protocols. Because it is native to ORBIS, every action lands where care happens, and the clinician confirms each step.
Read more: Dedalus and Corti usher in the agentic EHR era across Europe
“ORBIS has always been built around the needs of clinicians, with and for them. With Buddy, clinicians can put the full power of ORBIS to work within one easy to use interface. Partnering with Corti gives us a production-grade agentic framework, as well as the governance, compliance, and data provenance capabilities we need to deliver a high-quality product across our hospital network,” said Siobhan McMahon, Director of AI and Transformation at Dedalus.
Advanced Agentic AI Without Sacrificing Control
As agentic AI becomes embedded in critical healthcare workflows, organisations face growing questions about long-term technological independence and their reliance on specific models, infrastructure and providers.
Corti's Agentic Framework forms part of its broader sovereign AI stack, including models built on open foundations and available through an OpenAI-compatible API. Its inference infrastructure, developed in collaboration with NVIDIA, uses NVIDIA GPUs and Triton Inference Server to support real-time AI applications.
For Corti, sovereignty goes beyond where data is stored. It means giving organisations the freedom to choose how AI is built, deployed and governed, without sacrificing access to advanced capabilities.
Availability
The Corti Agentic Framework is available at console.corti.app, with documentation at docs.corti.ai