Comparison
Run it by vs gotoHuman, Slack approvals and workflow human tasks
Oct 27, 2026 · 6 min read
The short version
Most human-in-the-loop tools assume the approver is on your team: they sit in your Slack, have a seat in a review inbox, or are a user in your workflow engine. Run it by assumes the opposite. The approver can be anyone with an email address or a phone, they get their own link, and they don’t sign up for anything.
Facts about other products were checked against their own pages in October 2026. Follow the links for current details.
Review inboxes, such as gotoHuman
gotoHuman is built around an inbox where your team reviews what agents propose. You define a review type in its editor, the agent requests a review (there’s an MCP server for this), reviewers act in the inbox or from an email or Slack notification, and the result comes back by webhook. Plans are priced by seat.
That’s a good fit for an internal team that reviews agent output every day. Run it by fits the other case: one-off decisions from people outside the team, a client, a landlord, a manager at another company, who will open one link and never come back.
Slack and Teams approvals
Approval buttons in a chat tool are convenient when every approver lives in the same workspace. They stop working as soon as one approver doesn’t: a customer, a contractor, a partner on a different chat tool. Run it by’s links work in any browser, and you can still post them into Slack yourself.
Workflow engines with human steps
Frameworks and engines give you the pause, not the page. LangGraph interrupts stop a graph and resume it with a value your code supplies; Orkes Conductor human tasks wait for a form submitted by an assigned user. You still decide how the person is reached, what they see and how they prove who they are.
The two fit together: pause your graph, create a Run it by request, and resume when the webhook brings the verdict.
Side by side
| Run it by | Review inbox (gotoHuman) | Chat-tool approvals | Workflow human tasks | |
|---|---|---|---|---|
| Who can approve | Anyone with an email or phone | Reviewers on your plan’s seats | Members of your workspace | Users of your system |
| What they open | A personal link, no sign-up | The inbox, or a notification | A message with buttons | Whatever UI you build or configure |
| Several approvers and a policy | Built in: any, all, majority or n | Configured per review type | Usually one approver per message | You model it in the workflow |
| Proof it was them | Optional email code per person | Their account | Their chat account | Their account in your system |
| No answer in time | Verdict expired, never a silent yes |
Depends on your setup | Depends on your setup | Timeouts you configure |
| Also collects data | Yes: the same links do forms | Review forms | Limited | Yes, with your own forms |
How to choose
- Your own team reviews agent output all day: a review inbox.
- Every approver is in your Slack and the stakes are low: chat-tool buttons.
- You run a workflow engine and only need a pause: its human step, plus a way to reach the person.
- The approver is outside your team, several people must agree, or you need to prove who decided: Run it by approvals.
Try it with your agent
Free for 100 requests a month, no card. Add the MCP server from Claude Code or Codex with one command, or use the REST API.