Niches · Human approval & handoff

Pause a long workflow on a human answer and resume it once, with its state

Two or three companies' docs show a full fit.

2companies, full fit
7companies, fit with a limit
32companies carry the tag
19share of the category's work
7first-hand accounts

What counts as done

The run survives a worker restart during the wait, resumes on the answer, receives only the answer for its own key, and an answer sent before the wait registered is not lost.

Measured, for our own tests: it resumes within 10 seconds of the answer. No docs we read state these figures, so they don't count against a listing here.

How agents ask for it: wait for the manager's approval then continue, pause until the reviewer signs off, even overnight, resume the agent when they reply

Does the job 2

The docs show it meets the test above, and nothing we read says otherwise.

ListingWhy
Temporal BB
Temporal Technologies Inc.
Workflow state and timers live in the event history, so a worker restart replays to the same wait; the workflow ID scopes the Signal and a Signal handled before the wait still sets the state. No resume latency is given, and the page rules out sub-second needs. source
withHuman BB
Metoro Inc
Requests persist server-side by ID, so an adapter can re-poll after a restart, and webhook mode sends a signed request.resolved retried for seven days. Each poll waits at most 30 seconds and no resume latency is stated. source

Does the job, with a limit 7

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

ListingWhy
gotoHuman D
gotoHuman UG (haftungsbeschränkt)
The review is held server-side and the result is pushed by webhook, retried up to 7 times over hours with an Idempotency-Key, and keyed by reviewId. Resuming the run is the builder's code, no resume latency is stated, and fetching by ID is a Team-plan Reviews API. source
Hatchet B
Hatchet Technologies, Inc.
The wait survives requeue and reads the event from the durable log; CEL filters plus a scope with lookback catch early answers. No resume latency is stated, Ruby lacks lookback, and issue 4714 (cross-scope waiter match) was fixed in August 2026. source
Inngest BB
Inngest Inc
Waits hold no process and resume on a matching event or a unique signal, matched by field or if expression. The docs say an event sent before waitForEvent listens is not matched (issue 1433), and no resume latency is stated. source
Orkes Conductor Human tasks C
Orkes, Inc.
The task is held server-side by the Conductor cluster and completed by task ID, so worker restarts and early answers do not apply. No figure is given for resume delay after completion. source
Pushary C
RalphNex OÜ
Questions persist server-side under a correlationId until answered or cancelled (connector questions close after 10 minutes), but the agent must keep polling wait_for_answer at 48 to 55 seconds a call, and a late reply cannot restart an ended turn. source
Restate BB
Restate GmbH
Signals, awakeables and promises survive retries and restarts, a signal resolved before the wait is stored, and awakeable IDs are unique. Resume latency is not stated, and issue 4566 (resolves lost in leadership changes) was fixed in April 2026. source
Trigger.dev B
API Hero Ltd
Waitpoint tokens are held server-side, scoped by token ID with a per-token public access token, and complete over HTTP or the SDK. No resume latency is stated, and issue 2295 reported a run resuming 11 minutes after completion. source

Doesn't do this job 2

Listings in the category that do something else, with the reason.

ListingWhy
Big Arrow BB
Franz Enzenhofer (individual, Vienna)
bigarrow wait blocks in the agent's own shell; a card survives the session but nothing persists or resumes a run after a restart, and there is no correlation key beyond the process id. source
Permit MCP Gateway C
Permit Inc.
The wait lives in the open MCP call: an agent disconnect cancels the request and the default window is 5 minutes, so nothing survives a restart. source

What stops agents finishing

  • event sent before the wait registers is never matched
  • resume delayed minutes by queueing or concurrency limits
  • signals cannot be resolved over HTTP in the released server
  • cross-scope events satisfy the wrong waiter

Around it

Before: workflow-automation.start-run. After: human-in-the-loop.approve-action.

Where this comes from

Weight 19 of 100 in Human approval & handoff: 7 of 26 accounts, all GitHub issues on Inngest, Trigger.dev, Hatchet, Restate and LangGraph

The same as data: /api/v1/use-cases/human-in-the-loop.durable-wait.json, or the MCP tool get_use_case with id human-in-the-loop.durable-wait. Other jobs in this category: /api/v1/niches?category=human-in-the-loop.

Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person. A mark says what the docs show, not what a run measured.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.