Design partners

Help us build the permission layer
for coding agents.

We're looking for a small number of teams who are actively building with Claude Code, Codex, or Cursor and want a human in the loop before their agents touch production infrastructure.

What we're building

BehalfID is runtime action authorization for AI agents. Before a tool call runs, the agent checks whether it's permitted. If the action requires approval, the agent pauses and waits for a human to approve in the dashboard. If it's denied, the action stops. Every decision is logged.

The first workflow we've focused on is deploy authorization for coding agents. The agent can deploy to staging freely; production deploys pause for human approval via the BehalfID MCP server. Setup takes about five minutes.

what the agent sees when blocked
APPROVAL REQUIRED — do not execute this action.

Action:      deploy_production on vercel.com
Approval ID: apr_Def456uvw

Approve at: https://behalfid.com/dashboard/approvals

This is an early product. The core enforcement loop works end-to-end. We're looking for real usage to find the edges.

Who we're looking for

  • Teams using Claude Code, Codex, or Cursor in a real development workflow — not just evaluating.
  • Engineers who want approval gates before agents touch production: deploys, database migrations, secret rotation, external communications.
  • Teams with at least one person who can spend 30 minutes setting up the integration and giving us honest feedback.
  • Small teams or solo builders— you don't need a company or a use case that looks like enterprise.

What you get

  • Direct setup help.We'll walk through the integration with you and unblock anything that doesn't work.
  • Early access to features as we build them — better approval notifications, team support, webhook routing, and more.
  • Direct input on what we build next. Your use case will directly influence the roadmap.
  • Free Pro plan for the duration of the design partner engagement.

What we need from you

  • A 30-minute setup call to walk through your workflow and get BehalfID wired in.
  • Honest feedback on what's confusing, broken, or missing — in writing or on a short call.
  • Occasional check-ins as we ship new features. No long-term commitment; you can stop any time.
Important:BehalfID enforces where you integrate it. If your agent calls an API directly without going through the MCP tool or SDK, that call is not covered. We'll help you understand exactly what is and isn't enforced in your setup before you rely on it for anything critical.

Get in touch

Email us at design-partners@behalfid.com with a sentence or two about what you're building and what kind of approval gates you're trying to enforce. We respond within one business day.

Or create an account and try the deploy approval workflow yourself — it's fully functional and takes five minutes to set up.

Create accountRead demo guide

What we won't promise

We're a small team and this is an early product. We won't promise enterprise SLAs, SSO, SOC 2, or a white-glove onboarding process. We will promise to respond quickly, fix things that are broken, and build what actually helps your workflow.

If you need production-grade guarantees now, BehalfID is probably not ready for your use case yet. We're building toward that, and design-partner feedback is how we get there faster.