withHuman GTM brain

withHuman provides an open-source approval layer that lets teams review consequential actions taken by AI agents.

category

AI agent approval infrastructure

market profile

  • text: Teams deploying agents that can act on external systems need a way to route high-impact actions to human reviewers.
  • text: Dedicated approval services compete with built-in human review in agent frameworks and workflow tools.

personas

  • title: AI platform engineer
    pains:
    • Each agent framework implements approvals differently.
    • Tool calls need to pause and resume without rewriting the whole agent.
  • title: Security or AI governance lead
    pains:
    • Agents can initiate consequential payments, refunds or production changes.
    • Auditors need evidence of who approved actions and which policy applied.
  • title: Support operations manager
    pains:
    • A support agent needs to resolve small refunds automatically while escalating larger ones.
    • Review requests must reach the right on-call person with enough context.

icp

  • text: Engineering-led companies developing multiple AI agents with access to external tools or production systems.
  • text: Teams with approval policies and designated reviewers for support, finance or engineering operations, from small teams to enterprises requiring SSO.

message pillars

  • text: Keep consequential agent actions subject to review.
  • text: Work across agent frameworks and open standards.
  • text: Send reviewers the action and context needed to decide.

objections

  • objection: Will adding approvals slow down every agent action?
    response: Use policy thresholds to approve low-risk actions automatically and route exceptions to a reviewer.
  • objection: Will the approval layer cover an agent that bypasses the gated tool?
    response: Review credentials and network boundaries alongside the integration; the launch post explicitly discusses direct API bypass risk.
  • objection: Can we deploy it under enterprise requirements?
    response: Evaluate deployment options and security controls with the vendor; pricing lists enterprise cloud, on-prem and BYOC options.

competitors

  • name: HumanLayer
    why: Human-in-the-loop infrastructure for agent actions.
  • name: LangGraph
    why: Agent framework with human review and workflow control.
  • name: n8n
    why: Workflow automation with human approval steps.

keywords

  • AI agent human approval
  • human in the loop agent
  • agent tool call approval
  • AI agent governance
  • MCP approval gateway
  • agent approval protocol

voice

  • technical and concrete
  • pragmatic
  • control-focused

signals

  • text: A team is moving support or finance agents from suggestions to tool calls that can modify external systems.
  • text: An organization is adopting multiple agent frameworks and needs common approval policies and audit logs.

stack

  • MCP
  • Agent Approval Protocol
  • Slack
  • Discord
  • Stripe

pricing

model: freemium
starting price: $0 hosted; Scale $20 per user / month

motion

hybrid

channels

  • GitHub
  • Product Hunt
  • SEO
  • developer community

confidence

high

Founders

  • name: Chris Battarbee
    title: Founder
    role: Founder
    bio: I'm one of the founders of withHuman - We're building a human approval layer for AI Agents Before withHuman, I worked at Palantir as a Software Engineer where I built out Foundry's compute platform.

Back to YC companies