Unified Control Plane (UCP)
The product enterprises buy. Open-core MCP runtime providing deterministic governance for AI agent workflows. Self-hosted, policy-gated, audit-trailed, fail-closed by default.
Overview
UCP is the governance runtime for AI agents. It sits between your AI models and the systems they interact with, enforcing policy on every action. Built on the Model Context Protocol, UCP provides a standardized control plane that is self-hosted, deterministic, and fail-closed by default.
UCP Core
The governance runtime that enforces policy gates, manages confirmation workflows, and maintains the audit trail. Every agent action flows through Core before execution — if it violates policy, the system stops.
UCP SDK
Developer toolkit for building governed agent workflows with type-safe interfaces and policy annotations. Built-in telemetry means every operation is automatically traced without developers adding instrumentation.
Packs
Use-case modules that extend UCP with domain-specific governance. M365, OpenClaw, Policy, and Mathematical Autopsy packs provide pre-built governance for common enterprise scenarios.
Reference Apps
Pack-backed user interfaces that demonstrate governed workflows in action. Deploy as-is or customize for specific enterprise requirements.
Key Capabilities
| Capability | Description |
|---|---|
| Policy-Gated Execution | Every agent action validated against organizational policy before execution |
| Confirmation Gates | Destructive and high-risk operations require explicit human confirmation |
| Fail-Closed by Default | If policy evaluation fails, the action is rejected — never allowed through |
| Complete Audit Trail | Every action, decision, and confirmation recorded with full context |
| Self-Hosted | Runs in your infrastructure. Your data never leaves your control |
| Tiered Risk Classification | Actions classified by risk level with appropriate governance for each tier |
| MCP-Native | Built on the Model Context Protocol for standardized agent-tool interaction |
| Deterministic Governance | Same policy, same action, same decision — every time, provably |
Packs
Domain-specific modules that bring governed agent capabilities to your infrastructure.
M365 Pack
39 agent personas, 85+ governed actions, 16 toolsComplete Microsoft 365 governance — email, calendar, Teams, SharePoint, security, compliance, and administration. Every action policy-gated with confirmation workflows for destructive operations.
OpenClaw Pack
Multi-channel messaging governanceGoverned messaging across WhatsApp, Telegram, Discord, and Slack. Sender allowlists, channel-level policies, and full audit trail for every message sent through any channel.
Policy Pack
Governance policy engineDefine, enforce, and audit organizational policies for AI agent behavior. Role-based access, action-level permissions, and tiered risk classification.
MA Guard
Mathematical Autopsy enforcementRuntime enforcement of Mathematical Autopsy constraints. Validates that agent actions comply with formally proven invariants before execution.
SIEM Bridge
Security information integrationStreams governance events to enterprise SIEM systems. Complete visibility into agent actions, policy decisions, and confirmation gate outcomes.
Artifact Vault
Governed artifact storageImmutable, signed artifact storage for governance evidence. HMAC-verified audit records, policy snapshots, and compliance documentation.
Deployment Tiers
Community
Evaluation and individual developer tier with MA Guard, MA Integration, IDE Bridge, and Template Guard. Free to use — everything you need to start building governed AI workflows.
Professional
Teams needing guardrails and policy enforcement, adding Policy Pack, Rate Guard, and Governed Agent Ops. Designed for development teams shipping AI features that need deterministic governance.
Enterprise
Production operations needing telemetry and evidence persistence, adding SIEM Bridge and Artifact Vault. Complete audit trail and compliance evidence for regulated industry deployments.
Use Cases
Enterprise AI Agent Deployment
Deploy AI agents that interact with enterprise systems — email, calendars, documents, messaging — with mathematical guarantees on every action they take.
Regulated Industry Compliance
Meet regulatory requirements with deterministic governance and complete audit trails. Every agent action traceable to a policy decision.
Multi-Agent Orchestration
Govern complex workflows involving multiple AI agents operating across different systems, with policy enforcement at every handoff.
Secure Internal Automation
Automate internal processes with confidence. Confirmation gates prevent destructive operations while allowing high-throughput for safe actions.
Governance Runtime Research
UCP is built on active research into deterministic governance primitives for AI agent systems operating at enterprise scale.
Current research explores formal verification of policy composition under concurrent multi-agent execution, automatic derivation of minimal confirmation gate placement for risk-optimal workflows, and governance protocol design for cross-organizational agent interactions where trust boundaries must be mathematically enforced.
Explore ResearchResearch Areas
- ▸ Formal policy composition verification
- ▸ Optimal confirmation gate placement
- ▸ Cross-organizational governance protocols
- ▸ Deterministic audit trail compression
- ▸ Risk-tiered execution scheduling
Govern Your AI Agents
Deploy AI agents with deterministic governance, complete audit trails, and fail-closed safety. Self-hosted in your infrastructure.