Skip to main content
Hero Light

​
Human Approvals for AI Agents

Transform your AI agents from limited question-answering tools into powerful business automation engines - with just one line of code. Add secure human oversight to critical operations while maintaining full control over your automation workflows.

TypeScript Quickstart

Add human approval workflows to your TypeScript/JavaScript agents

Python Quickstart

Integrate human approvals into your Python-based agents

​
Why Human Approvals?

Build Trust

Give stakeholders confidence in AI automation with transparent human oversight

Ensure Compliance

Meet regulatory requirements with built-in audit trails and approvals

Reduce Risk

Add safety checks to critical operations without sacrificing automation

Scale Safely

Deploy AI agents for high-stakes tasks with appropriate controls

​
How It Works

Approvals act as intelligent checkpoints in your AI agent’s workflow:
  1. Pause: AI agent reaches a critical decision point
  2. Present: System displays context and data to approvers
  3. Verify: Human reviews and approves/rejects the action
  4. Continue or Halt: Based on human decision
  • Processing financial transactions
  • Sending mass communications
  • Modifying system configurations
  • Accessing sensitive data
  • Executing bulk operations
  • Synchronous and asynchronous approval flows
  • Conditional auto-approval rules
  • Multi-approver workflows
  • Rich context display
  • Complete audit trails

​
Simple Integration

import { needsHumanApproval } from '@your-org/approvals'

// Add approval requirement to any operation
const sendEmail = needsHumanApproval({
  type: 'sync',
  title: 'Email Campaign',
  ask: (args) => `Approve sending email to ${args.recipientCount} customers?`
})

// Use it in your agent
await sendEmail({
  recipientCount: 1000,
  subject: 'Important Update',
  template: 'newsletter'
})

​
Why We Built This

Organizations face several challenges when deploying AI agents for critical operations:

Trust Gap

Challenge: Organizations hesitate to give AI agents unrestricted accessSolution: Add transparent human oversight

Compliance Needs

Challenge: Many industries require human verificationSolution: Built-in approval workflows and audit trails

Development Overhead

Challenge: Building approval systems is time-consumingSolution: One-line integration with complete features

Scaling Concerns

Challenge: Managing oversight at scale is complexSolution: Automated routing and efficient workflows

​
Next Steps

Browse Recipes

Explore pre-built approval workflows for common scenarios