Heed intercepts AI agent actions before they execute and routes them to a human for approval. One API call. Full audit trail.
Works with LangChain, OpenAI Agents SDK, AutoGen, CrewAI, or plain HTTP.
If your agent can POST, it can use Heed.
Your agent followed the rules exactly. The rules were wrong for this case. Nobody reviewed it before the money left the account.
The agent was configured to send emails. Nobody specified that "send" meant "send to everyone, right now, with the draft copy."
Your agent fixed the bug and pushed. Nobody told it that main is production, Friday is sacred, and this particular fix had a side effect.
These aren't hypotheticals. They're the support tickets developers are filing right now.
The problem isn't that your agents are bad. It's that consequential actions need a human in the loop — and nothing in your current stack provides that.
A single POST before any consequential action. Include what the agent wants to do, why, and where to send the decision.
Approval cards land in Slack or email with full context. One click to approve or reject. No login required. Works on mobile.
Decision POSTs to your callback URL with HMAC signature. The action executes. Everything is logged. You have a complete audit trail.
LangSmith, Langfuse and Humanloop are post-mortems. Heed is a gate.
Trace and log agent actions after they happen. Great for debugging. Too late to stop the $47K refund.
Intercept before execution. Human confirms. Then it happens — with a logged, audited paper trail.
Configure rules per tenant. Actions that match auto-approve instantly. Everything else holds for human review. No code changes needed to update policies.
Email and Slack cards with approve/reject buttons. No login required. Works on mobile. The decision takes 10 seconds, not 10 minutes.
Every request, decision, and webhook delivery is logged append-only. SOC 2 and EU AI Act Article 14 ready. Know exactly what happened and when.
Decisions POST back to your callback URL, signed with HMAC-SHA256. Three retries with exponential backoff. Your agent always gets the answer.
Separate tenants, API keys, policies and audit logs per product or client. Stripe-style hd_live_ / hd_test_ key format.
Any language. Any agent framework. LangChain, OpenAI Agents SDK, AutoGen, CrewAI — or plain HTTP. If it can POST, it can use Heed.
"High-risk AI systems shall be designed and developed in such a way, including with appropriate human-machine interface tools, that they can be effectively overseen by natural persons during the period in which the AI system is in use."
— EU AI Act, Article 14 (Human Oversight)If your agents touch healthcare, finance, legal, government, HR, or critical infrastructure, Article 14 applies. Heed gives you the human control layer and the audit trail to prove it.
No credit card required. Upgrade or downgrade anytime.
Free to start. One API call to integrate. Works with every agent framework.