withHuman
withHuman provides an approval layer for AI agents, intercepting tool calls and holding them until designated human reviewers approve or deny them. Teams define approval rules and pipelines, such as requiring human sign-off for refunds over $20, and agents are paused at those decision points. The platform is framework-agnostic and open source, integrating via the Agent Approval Protocol API, an MCP server, or a gateway that wraps existing tools.
- Enterprise Software
- Developer Tools
- Artificial Intelligence
- AI Agents
- Software Only
Funding
Raised to date
$500KRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Founders
Founder details are not available yet.
Product
Problem
As AI agents increasingly execute real-world tasks—refunding customers, paying invoices, modifying infrastructure—they take consequential, potentially irreversible actions automatically. Without human oversight, these actions risk costly errors, policy violations, or operations that cannot be undone, creating a barrier to trusting agents with meaningful work.
Solution
withHuman provides an approval layer that intercepts agent tool calls and holds them until designated human reviewers approve or deny them. Teams define approval rules and pipelines—such as requiring human sign-off for refunds over $20—and agents are paused at those decision points. The platform is framework-agnostic and built on open standards, integrating via the Agent Approval Protocol API, an MCP server, or a gateway that wraps existing tools. Reviewers respond through email, Slack, Discord, WhatsApp, SMS, or mobile apps, and decisions are recorded in an immutable audit log. Developers can add custom logic through webhook blocks or leverage built-in AI Judge blocks for automated pre-screening. The solution supports escalation paths, team-based routing, and granular permission policies for enterprise control.
Target Audience
Primary users are engineering and operations teams deploying AI agents in production, including coding agents, support bots, and finance automation workflows. The platform targets organizations that need to delegate consequential actions to agents while maintaining accountable human oversight.
Features
- Open-source, framework-agnostic design with hooks for coding agents and an Agent Approval Protocol API
- withHuman MCP server and MCP gateway for intercepting tool calls across agent frameworks
- Configurable approval pipelines with conditions, branch blocks, webhook blocks, AI Judge blocks, and delegate blocks
- Escalation paths and team escalation policies for routing requests to the right reviewers
- Review via email, Slack, Discord, text message, WhatsApp, iPhone, and Android apps
- Immutable audit log tracking who approved, denied, or claimed requests, from which channel, and when
- Organization policies, roles, scoped API keys, single sign-on, verified domains, and directory sync (SCIM)
- Optional SIEM export, on-prem or BYOC hosting, and white-glove onboarding on enterprise tier