Humans Do

humans.do

You are an expert in humans.do — the .do service for integrating human workers into autonomous workflows using the same syntax as AI agents.

When to Use

Activate when the user needs human approval, review, or input inside a .do workflow.

The Key Insight

Humans use the exact same tagged template syntax as AI agents. No special handling, no callbacks, no webhooks to configure manually.

import { priya, ralph } from 'agents.do'
import { legal, ceo, designer } from 'humans.do'

// AI agent
const spec = await priya`define the feature spec`

// Human worker — identical syntax, workflow pauses until they respond
const approved = await ceo`approve this spec: ${spec}`

// Back to AI
const code = await ralph`build ${approved}`

Routing

Messages automatically route to wherever the human is:

import { slack, email, sms } from 'humans.do/channels'

// Route to specific channels
await legal.via(slack)`review this contract`
await ceo.via(email)`approve the $50k spend`

Default routing is configured per-human in org.ai.

Approval Patterns

// Simple approval gate
const approved = await ceo`approve the partnership with ${partner}`
if (!approved) return

// Multi-approver (all must approve)
await pr.approvedBy(quinn, tom, priya)

// First-responder (any one approves)
await legal`sign off on ${contract}`

Timeout & Escalation

import { withTimeout, escalate } from 'humans.do'

// Escalate if no response in 4 hours
const review = await withTimeout(
  legal`review ${contract}`,
  { hours: 4, escalateTo: ceo }
)

Best Practices

• Use humans.do for decisions with accountability or legal weight
• Keep human tasks atomic — one clear question, one expected response
• Set timeouts on human steps — workflows shouldn't block indefinitely
• org.ai is where you configure who each role maps to and their routing preferences