HumanLayer

API and SDK designed for AI agents to seamlessly interact with humans for assistance, feedback, and approvals, enhancing human-in-the-loop workflows.

Visit Site

Visit

AI Agent AI API AI Developer Tools AI Workflow Large Language Models (LLMs)

About HumanLayer

HumanLayer provides a comprehensive API and SDK that enable AI agents to contact humans for assistance, feedback, and approval processes. It offers human-in-the-loop capabilities, allowing developers to build AI workflows that incorporate human oversight. Compatible with channels like Slack, Email, and Discord, HumanLayer integrates seamlessly with popular orchestration frameworks and large language models to support reliable, scalable AI-human interactions.

How to Use

Integrate the HumanLayer SDK into your AI system. Use the @hl.require_approval() decorator to pause functions needing human oversight. Reach out to humans through multiple channels and route approvals to specific teams or individuals for streamlined workflows.

Features

Flexible approval routing for precise control
Require human approval before executing critical functions
Advanced workflows with escalation and timeout options
Support for Slack, Email, Discord, and other channels
Framework and LLM agnostic design
Human as a tool with hl.human_as_tool() for enhanced interactions

Use Cases

Gathering human feedback to enhance AI performance.
Automating lead qualification with human input.
Ensuring reliability with human oversight in AI processes.
Obtaining critical approvals for high-stakes decisions.
Managing complex approval workflows across teams.

Best For

AI developersWorkflow automation teamsSales and marketing professionalsHuman resources teamsCustomer support teamsAI system architects

Pros

Enables learning and automatic approvals based on past interactions.
Supports multiple communication channels for flexibility.
Integrates smoothly with existing LLMs and frameworks.
Provides human oversight to ensure AI reliability.
Offers detailed control over approval routing.

Cons

Requires active user participation for approvals.
Increases complexity in AI development workflows.
Limited data retention on the free plan.

Pricing Plans

Choose the perfect plan. All plans include 24/7 support.

Starter

$0/month

Free for developers and hobbyists. Supports all major frameworks with 100 operations per month. Additional operations at $20 per 200. Connects via Slack, Email, and Discord. Up to 10 active users with limited data retention.

Get Started

Premium

$500/month

Designed for teams and scaling. Remove branding with your own app and keys. Includes 2000 operations monthly, with extra at $18 per 200. Features advanced approval workflows with escalations, timeouts, and auto-approvals. Supports MSTeams, SMS, and RCS. Priority support and 180 days of data retention.

Get Started

Enterprise

Custom

Ideal for private deployments on VPC or on-premises. Offers RBAC, SSO, volume discounts, and customizable notification channels. Includes expert guidance for building tailored AI agents for sales, marketing, and HR.

Get Started

FAQs

What defines an active user?

An active user is anyone registered to approve or respond to AI agent requests.

How do Learning and Auto Approvals work?

They automatically approve or deny requests based on historical human interactions and set thresholds.

What are escalation procedures and timeouts?

They manage approval processes across teams, with options for escalation and timeouts, supporting Promise.race() and Promise.all() semantics.

Can I export datasets for training and evaluation?

Yes, you can periodically export live feedback data to high-fidelity labeled datasets for fine-tuning or evaluation purposes.

Is HumanLayer suitable for large-scale enterprise use?

Absolutely, with features like custom deployment options, RBAC, SSO, and dedicated support, it is ideal for enterprise-scale integrations.