Autonomy you can prove.
The control plane for every agent in your organization. Give each agent an identity, a scope, a budget and an audit trail; enforce policy at runtime; and show regulators, boards and customers exactly how your AI behaves.
Agents
1,284
Actions today
2.4M
Blocked
312
telecom-sia/change-guardian
config.push
cx-sia/frontline
refund.issue ≤ $200
eng-sia/builder
repo.merge(main)
aiops-sia/remediator
k8s.rollout.undo
Illustrative dashboard
Agent registry & identity
Every agent registered with an owner, a purpose, a risk tier and its own credentials.
Runtime policy engine
Allow, deny or require approval for any action — enforced on every step, not just at login.
Audit & observability
Tamper-evident logs of every prompt, tool call, decision and approval.
Cost & risk controls
Budgets per agent, team and project; kill switches; anomaly alerts.
Governing
Policy as code, enforced in real time
Write policies once — who an agent can act for, which tools it may call, which data it may see, how much it can spend — and Cana enforces them on every step.
- Attribute-based access for agents
- PII detection and redaction
- Prompt-injection and data-exfiltration defenses
- Human approval gates by risk tier
Agents
1,284
Actions today
2.4M
Blocked
312
telecom-sia/change-guardian
config.push
cx-sia/frontline
refund.issue ≤ $200
eng-sia/builder
repo.merge(main)
aiops-sia/remediator
k8s.rollout.undo
Illustrative dashboard
Governing
Built for frameworks regulators recognize
Map controls to the frameworks your auditors use and export evidence on demand, so AI governance becomes a report you generate, not a project you staff.
- Mappings for EU AI Act, NIST AI RMF and ISO/IEC 42001
- Model and agent cards generated automatically
- Evaluation history as evidence
- Data residency controls
Mappings for EU AI Act, NIST AI RMF and ISO/IEC 42001
Model and agent cards generated automatically
Evaluation history as evidence
Data residency controls
Governing, step by step.
- 01
Register
Inventory every agent, model and tool.
- 02
Scope
Assign identity, permissions and budgets.
- 03
Enforce
Apply policies at runtime on every action.
- 04
Prove
Export audit evidence and reports.
Under the hood.
Identity
Per-agent credentials, SSO, SCIM
Policy
Attribute-based, versioned, testable
Audit
Tamper-evident, exportable, SIEM streaming
Frameworks
EU AI Act, NIST AI RMF, ISO/IEC 42001 mappings
What teams build with it.
Stop agent sprawl
Discover every agent across teams and bring it under one policy.
Prepare for audit
Generate evidence packs for internal audit and regulators.
Contain spend
Budgets and alerts before a runaway agent burns your quarter.
Good questions.
Can Cana govern agents not built on Cana?
Yes. Through the MCP gateway and SDK, Cana can register, monitor and enforce policy on agents and tools across your stack.
Is Cana certified?
Our security program is designed around SOC 2 and ISO standards. See the Trust page for current status and our roadmap.
How do I get started with Governing?
Start a Cana chat, then open Governing from the Solutions area when it fits the work you need to do.
How do I control what an agent can do?
Choose the permission level that fits the task: Ask, Accept Edits, Research and Plan, or Bypass All.
Can I connect tools to this work?
Cana can use connected apps and remote MCP servers when you have connected and allowed those tools.
Better together.
Platform
Cana MCP
Connect an MCP server to a Cana chat, a Relay or a Cana App, so the agent can use the tools it exposes.
Solution
Deploying
Run agents where your data and regulators need them: Cana Cloud, your VPC on AWS, Google Cloud, Azure or Oracle, on-premises, at the edge, or fully air-gapped.
Enterprise
Cana for Enterprise
Governed, sovereign agents for the whole organization.
Explore Governing. Join the waitlist.
Join the waitlist for early access to Super Intelligent Agents and updates as Cana opens up.