🧍🧍‍♀️

humans.sh

on-demand human cognition for AI agents

$ humans ask "Jane Doe" "Can you check in on John? I saw he raised a wellbeing flag. Would be great if you could see how he's doing." --sync
 Message sent to Discord
 Awaiting response
 Response:
 Sure thing!

AI-to-human prompting

For centuries, humans have called out to machines to complete tasks. The printing press, the factory mill, the personal computer.

It's time to invert control. AI is not just a modern-day calculator, it is also a capable operator.

This creates an opportunity to rethink automation: machines that initiate and manage workflows, calling on humans when help is required.

Collaborative & Async

We give AI agents a shell prompt to request input from humans within their organisation.

Those requests are routed to available humans who can respond in their own time.

By making it frictionless for humans to participate in automated workflows, better quality outcomes can be achieved versus relying on AI alone.

Features

  • AI-human collaboration
    Free-flowing communication via Discord (more integrations coming)
  • Task management
    Adapting to individuals' work patterns, schedules and personal capacity
  • Roles and responsibilties
    Enabling AI agents to make informed decisions about who to assign tasks

Coming later:

  • Performance monitoring
    Velocity and progress tracking, enabling strategic planning
  • Incident escalation
    An Andon Cord for both AI and human agents
  • Explainability
    Audit trails for incident resolution and compliance

Created by @djgrant_