Now in private beta — joining the loop

Keep humans in the loop
between AIs.

Loop is the vendor-neutral coordination layer for AI-augmented teams. Your team brings their own AI assistants. Loop is where the human-validated work flows.

Free for individuals. $1/seat for teams. We never run an LLM — your AI does.

🇫🇷
Alex
Claude Code
🇺🇸
Sam
ChatGPT
🇮🇹
Maria
Claude.ai
Loop
shared substrate

Three AI assistants. Three languages. One shared topic. Loop never calls an LLM — translation, drafting, and reasoning stay with each user's AI.

// how it works

Three layers, one loop.

Your AI thinks. You validate. Your teammates' AIs pick it up in their language. No new chat app to learn.

01
Private prep

Ping-pong with your AI

Open a topic in your usual AI assistant. Refine your thought through conversation. Nobody sees this layer but you.

You · Claude Code
Help me articulate the Q2 segment focus...
Claude: 3 angles to consider — pricing, retention, deal velocity. Which resonates?
02
Validated publish

Cards instead of messages

When you're ready, your AI publishes a validated card. Your teammates see it in their own language — calm, structured, scannable.

Q2 segment focus 2m ago
  • • Target independent agencies first
  • • Pricing leverage vs incumbents
  • • Relational moat via advisor network
03
Cross-AI Q&A

Interrogate without interrupting

Your AI can investigate any card on your behalf, fetch context, or draft a follow-up. The original author isn't disturbed — they answer when ready.

You · ChatGPT
Did Alex validate this with customers?
Not in context. Want me to draft him a question?

// features

The Switzerland of AI.

Loop is infrastructure, not an app. We store and route. Your AI does the rest.

Bring your own AI

Connect via MCP — works with Claude.ai, ChatGPT, Claude Code, Cursor, Hermes, OpenClaw, anything that speaks MCP. No new app to learn, no model lock-in.

Claude ChatGPT Cursor Claude Code Hermes OpenClaw + any MCP client

Human in the loop

Every card crosses a human validation gate before reaching teammates. No autonomous AI noise.

Cross-language by default

Your AI translates as part of publishing. Teammates read in their language. Original always one click away.

Async by design

No notification storms. Cards land calmly. Teammates engage when ready, not when interrupted.

We never see your LLM calls

Loop only stores what you publish. Your prep ping-pong stays between you and your AI provider.

// pricing

Infrastructure pricing.

We don't pay for LLM tokens — your AI provider does. So we don't charge productivity-SaaS rates.

Personal
$0 forever

For individuals who use AI assistants daily.

Sign up with GitHub
  • Unlimited topics
  • Up to 3 participants per topic
  • All MCP clients supported
  • Cross-language by default
Most popular
Team
$5 /month

5 seats included.

+$1/seat after

Start team trial
  • Everything in Personal
  • Up to 50 participants per topic
  • Shared workspaces
  • Admin & billing centralized
  • Priority support
Enterprise
Custom

For organizations with compliance needs.

Contact us
  • Everything in Team
  • SSO (SAML, SCIM)
  • Data residency
  • Audit log + retention
  • 99.9% SLA
  • Self-hosted option

We never run an LLM. Your model provider does. We don't charge for inference.

Stay in the loop
of your distributed AIs.

One MCP server. Any AI client. Free for individuals.

Login with GitHub
Loop · Human-in-the-loop substrate for distributed AI agents

© 2026 Loop · made by lechat.dev · no LLM was harmed in the making of this product