End-to-end encrypted

Human-in-the-loop for AI agents

Hiloop lets AI agents reach humans for approvals, chat, reviews, forms, and live sessions. Zero-trust architecture means the server never sees your content.

Start for free See live demo →

Try it — fully interactive

agent.py

from hiloop import HiloopClient

client = HiloopClient(api_key="hlp_your_key")

# Ask a human for approval
interaction = client.create_interaction(
    type="approval",
    title="Deploy v2.0 to production?",
    priority="high",
    deadline_minutes=60,
)

# Wait for the human response
result = client.await_response(interaction.id)
print(f"Decision: {result.response}")

# All content is end-to-end encrypted.
# The server never sees plaintext.

Everything agents need from humans

✅

Approvals

Yes/no decisions with optional comments. Perfect for deployments, access requests, and safety checks.

💬

Chat

Back-and-forth messaging between agents and humans. Great for support, clarification, and collaboration.

🔍

Reviews

Structured code and content review with feedback. Ideal for PR reviews, document reviews, and QA.

📋

Forms

Structured data collection with custom schemas. For configuration, onboarding, and surveys.

📺

Live Sessions

Real-time frame streaming with 8 render types. Share terminals, dashboards, and progress live.

How it works

Agent creates interaction

Your AI agent calls the Hiloop API to request human input. Content is encrypted client-side before sending.

Human responds

The request appears in the human's inbox with priority routing, notifications, and deadline tracking.

Agent continues

The agent receives the encrypted response via long-poll or webhook and proceeds with the task.

Zero-trust by design

All content fields -- titles, messages, responses, form data -- are encrypted client-side with X25519 + AES-256-GCM before reaching our servers. We literally cannot read your data.

Encrypted (content)

Titles, descriptions, messages, responses, form data, payment details, session frames

Plaintext (metadata)

Type, priority, status, timestamps, participant IDs, routing targets -- only what's needed for routing

SDKs for every stack

Python, TypeScript, Deno, CLI, and an MCP server for AI tools.

Python

pip install hiloop

TypeScript

npm install @hiloop/sdk

Deno

import from deno.land/x/hiloop

MCP Server

npx hiloop-mcp

Embed widgets

npm install @hiloop/elements

Works with your agent stack

Native integrations with leading AI agent frameworks and orchestration platforms.

🤖

OpenClaw

Channel Plugin

Add Hiloop as a channel in OpenClaw with hiloop-openclaw. Your agent gets chat, approvals, reviews, forms, and live sessions -- plus 9 dedicated tools through the standard channel interface.

View integration guide

⚡

Warp Oz

Webhook Integration

Warp Oz coding agents can request human review, approval, and guidance via webhooks. Session-aware tracking links Hiloop interactions to Oz sessions for seamless human-in-the-loop coding workflows.

View integration guide