The open platform to build enterprise-grade AI agents
Build governed agents, connect them to enterprise tools, and orchestrate business processes across systems, humans, and other agents. All with complete architectural freedom.
Trusted by enterprises worldwide
One open platform to build, orchestrate, and govern agents
Built on the same integration foundation Choreo Labs already runs in production environments globally.
Build
Build AI agents, and connect them to enterprise tools via MCP.
Orchestrate
Orchestrate agentic business processes across humans, systems, and other agents.
Govern
Govern what agents know and do, and observe what they did.
Choreo Agent Builder architecture
Interface
Chat · Voice · Messaging triggers · Webhooks · HTTP
Sandboxes & execution
Isolated runtime for safe tool and code execution
Permission & ID
Human-in-the-loop approvals + identity through Choreo Agent Manager
Observe & eval
Offline regression checks, online runtime evaluation
Governance
Policy and traffic control via Choreo Agent Manager
Tools & skills
MCP servers, prebuilt tools, and 600+ connectors
Agent-to-agent
Agent store discovery, sequential and hierarchical patterns
Memory
Short-term and long-term memory, plus knowledge bases
Orchestration loop
The reason-and-act cycle that turns a model into an agent
The Model
Bring any model provider or use ours
- Choreo Labs
- OpenAI
- Anthropic
- Azure OpenAI
- Google Vertex
- Mistral
- Ollama
- DeepSeek
Build
Everything you need to build enterprise-grade agents
- Build
No-code to pro-code
Start with pre-built agents, or build your own in no-code, and transition to pro-code as agent logic becomes complex.
- Connect
Tools and MCP
600+ pre-built connectors across APIs, events, files, and data available as tools. Build your own MCP servers, or convert existing APIs to MCP.
- Remember
Agent memory
Short-term and long-term memory, plus knowledge bases with a no-code pipeline to ingest unstructured data. Supports Milvus, Pgvector, Pinecone, and more.
- Persist
Agents + workflows
An agent that is a durable, long-running workflow, which can save progress, maintain its state, and resume work. Describe the goal of the workflow, and the agent decides the best path at runtime.
Orchestrate
Automate processes across agents, humans, and systems.
- Systems
Context, data, functionality
Orchestrate autonomous processes across enterprise systems with access to real-time context and data for better decision making.
- Agents
Multi-agent
Orchestrate multi-agent processes across sequential and hierarchical patterns.
Discovery agents through an agent store.
- Humans
Human actions
Include human decision making into processes to minimize an agent’s blast radius.
- Trigger
Agent interactions
Trigger agents via chat channels, event sources, APIs, and file source with support for both conversational and ambient agent styles.
Governance
Governance at multiple levels for any agent you build
- Know
Context engine
Provide agents access to real-time, contextual data to make better informed decisions.
- Do
Evals + human-in-the-loop
Use offline evaluations to identify an agent’s expected decision making during development.
Implement human-in-the-loop workflows to provide human oversight over agent actions.
- Did
Observability
Observe and audit every step and decision of the agentic process, including human actions, with agent tracing.
- Secure
Agent identity
Give agents an identity, and authenticate, authorize, audit, and administer agents via Choreo Agent Manager.
- Govern
Monitor and manage
Monitor and manage agent behavior, including LLM consumption, via Choreo Agent Manager.
Architectural freedom
Build, orchestrate, and govern agents your way
Deployment flexibility
Deploy in any environment, and any infrastructure, based on your compliance and sovereignty requirements.
Open source
Protect yourself from vendor lock-in, and maintain business continuity with 100% open source software.





