Authorize exact actions before execution, route approval only when policy requires it, and carry every decision and observed effect into current, reusable compliance evidence.
Exact-action authorization
Maetra binds a decision to the approved action envelope, policy versions, workspace, and intended executor. Change the request and the old authorization no longer matches.
Approved action
issue_refund $500 USDPolicy version 7 approved this exact refund request.
Changed request
issue_refund $5,000 USDThe action name stayed the same. The payload did not.
Execution result
ACTION_ENVELOPE_MISMATCHThe earlier decision cannot authorize the changed request.
Each capability can work independently. Together, they connect one AI-system record to applicable requirements, current reusable evidence, runtime policy, security checks, and a traceable action history.
Maetra scans GitHub and connected sources to identify official, shadow, and unreviewed agents, connect them to a governed inventory, and surface material changes for review.
Map AI systems to supported requirements across frameworks such as the EU AI Act, GDPR, NIST AI RMF, ISO 42001, the Colorado AI Act, and SOC 2. Reuse current evidence and documentation from the same underlying system record.
Use Govern in two ways: define exact AI agent approval policies for risks your team already understands, or turn on decision intelligence when each action needs runtime judgment. Maetra returns allow, approval, or block before the agent continues, then binds an approval to the exact action the executor is allowed to perform.
Protect how employees use ChatGPT, Claude, Gemini, and Microsoft Copilot in the browser. Interaction Guard checks covered prompts and supported files before submission, then warns, requests a business reason, or blocks the interaction according to your team's Guards.
Task Guard turns each user-authorized request into an active task contract, checks whether proposed actions remain in scope, guides drifting agents back to the objective, and requests confirmation before material scope changes. It then compares expected and reported effects in a traceable task record.
Scan prompts and tool calls in real time for prompt injection, data exfiltration, privilege escalation, and policy violations before they reach production tools.
Maetra records every discovery, classification, approval, execution attempt, resulting effect, policy change, and security incident. Signed receipts connect what was approved to what actually ran. When configured, independent verification and third-party anchors add external proof of the observed result and audit history.
Configure AI agent approval policies manually when you know the exact condition, approver, quorum, timeout, and notification channel. Or enable AI agent decision intelligence instead: Maetra evaluates each action against policy context, Secure rules, and selected compliance frameworks, then decides whether to allow, block, or send it for human approval.
Use exact rules for known actions, or let decision intelligence evaluate runtime risk and route approval dynamically.
Agents call Maetra at decision points. Maetra returns allowed, pending approval, or blocked with a payload-bound signed decision. Approved actions consume a one-use capability at execution, then attach provider attempts and effect evidence to the same action record, clearly separating reported results from independently verified results.
const decision = await maetra.checkpoint({ agent_name: "finance-agent-prod", action: "transfer_funds", payload: { amount: 84000, currency: "USD" }, timeout_seconds: 14400 }) if (decision.status === "approved") execute() if (decision.status === "pending") awaitWebhook() if (decision.status === "blocked") stop()
Connect GitHub for agent discovery, Slack, Email, WhatsApp, and Telegram for approval alerts, Webhooks for custom workflows, and SIEM for enterprise audit streams.
Pricing
Every self-serve option is visible here. Compare request allowances, policy limits, seats, and the complete feature matrix on the pricing page.
Explore the control path with no card required.
For an individual putting core controls into production.
For a small team beginning shared governance work.
For teams operationalizing governed agent workflows.
For broader control volume and enterprise operations.
Test how Maetra binds a policy decision to one consequential action, then explore how the same control graph keeps compliance evidence current.