The HumanHours Make app adds three modules and an instant trigger to any Make scenario, so you can wire ROI tracking into your existing automations without writing code.
Install (private invite)
While the app waits for Make's marketplace review, it ships as a private invite link:
https://www.make.com/en/hq/app-invitation/4a021c4c63191710f5c27327688685d4
Click the link, sign in to Make, and accept the invite. The HumanHours app shows up under Custom apps in your Make organisation. The invite link works for everyone you share it with.
Connection
Add a HumanHours API key connection once per Make org:
- API key —
hh_live_...from /api-keys in the dashboard.
The connection-test calls GET /v1/summary?period=30d; the connection label shows your workspace currency once it succeeds.
Modules
| Module | Type | What it does |
|---|---|---|
| Track event | Action | Sends a single agent event to HumanHours. Idempotency-keyed off the Make execution + iteration index, so retries do not double-count. |
| Get summary | Search | Returns hours saved, cost saved, and FTE-equivalent for a period. |
| Find agent | Search | Looks up an agent by slug; returns display name, total events, last activity. |
| Watch events | Instant trigger | Fires on every event HumanHours records. Auto-registers a webhook on the active scenario, with an optional outcome filter. |
Track event fields
| Field | Required | Description |
|---|---|---|
| Agent ID | yes | slug, e.g. support-classifier |
| Task type | yes | one of the built-in keys, or a custom one defined on /task-types |
| Outcome | yes | success / failure / needs_review |
| Human baseline minutes | no | override the task type's default |
| Agent duration (seconds) | no | how long the agent took |
| Agent cost | no | marginal LLM/third-party cost, used for net ROI |
| Occurred at | no | defaults to now |
| Metadata | no | free-form key/value pairs |
| Idempotency key | advanced | custom dedupe key; default uses the execution + iteration index |
Patterns
End-of-flow tracking. Drop the Track event module at the end of an existing Make flow that runs an AI agent. Wire agent_id and task_type to scenario constants, outcome to success (or branch on the agent's structured output).
ROI summary in Slack. Add a scheduled scenario that runs Get summary monthly, then formats the response into a Slack message with hours saved, cost saved, and FTE-equivalent.
Outcome-triggered alerting. Use the Watch events trigger with Outcome filter = Failure only to fire a downstream alert (Slack, PagerDuty, email) only when an agent fails.