Now in beta

Your agent gets stuck.
You get pinged.
Work continues.

The collaboration layer where AI agents ask humans for help — visual judgment, taste calls, account setup, anything they can't do alone. You respond from Telegram in seconds.

How it works

🛑

Agent hits a wall

Your AI agent is building your product, running marketing, or handling ops — and encounters something it can't decide or do.

📱

You get pinged

A Telegram notification appears with the question and inline buttons. Tap your choice or type a response — takes seconds.

🚀

Work continues

Your agent gets your answer via the API and keeps going. No context switching, no dashboard to check, no email threads.

Four ways to ask

Every task type maps to a clean Telegram interaction.

📋

Pick One

Present options and let the human choose.

"Which of these 3 color schemes feels right?"

Yes / No

Quick binary decisions.

"All tests pass. Deploy to production?"

🔍

Review

Approve, reject, or give detailed feedback.

"Here's the landing page copy. Does this feel right?"

🙏

Do For Me

Ask the human to do something and report back.

"What's the email on your Stripe account?"

Built for autonomous builders

When your AI agents run on autopilot, they need a lifeline.

👁

Visual judgment

Logo options, color schemes, layout choices — things AI can generate but can't judge aesthetically.

🔧

Account setup

Stripe credentials, OAuth configs, DNS records — things that need your identity or access.

🎨

Creative preferences

Tone of voice, brand direction, naming choices — subjective calls only you can make.

📦

Physical world

Take a photo, check a delivery, scan a QR code — tasks that require human presence.

Start building with AwaitHuman

Connect your Telegram, get an API key, and give your agents a human lifeline.

Get Started