Comparison
Run it by vs gotoHuman for AI agent approvals
Nov 4, 2026 · 6 min read
The short version
Both let an agent stop, ask a person, and get the result back by webhook. gotoHuman is built around a team inbox: reviewers check what agents produce every day, edit it before approving, and send it back for another try. Run it by is built around a link: one-off decisions and answers from whoever the agent needs to ask, with approval policies and expiry built in.
Facts about gotoHuman were checked against its own pages in November 2026. Follow the links for current details.
Describing the review
In gotoHuman you first design a review type in its editor, with fields such as text, markdown, images, buttons and selects. The agent then sends data into it through the Python or JavaScript SDK, the HTTP API, n8n, Make or a local MCP server.
In Run it by there is nothing to set up first. The agent sends the reason, the people and, for a form, a JSON Schema with each request. The MCP server is remote: add its URL to your client and sign in, with no key to copy.
Who reviews
gotoHuman reviewers are mostly team members with a seat, working in the web inbox, by email or in Slack. You can also add external reviewers in your workspace settings; they get a unique link and don’t need an account.
In Run it by every approver is reached by link, and you name them per request: 1 to
50 people, with no setup beforehand. Set verify: "email" and each
person confirms a code sent to their own address before the buttons work, so a
forwarded link can’t decide for them.
Several people, and no answer
A Run it by approval has a policy: any, all, majority or a number. The verdict
arrives as soon as it’s decided, and if nobody decides before the request expires,
the verdict is
expired. There is no path where silence turns into a yes.
gotoHuman can assign a review to several people. We didn’t find a policy for how many must approve, or an expiry, in its API reference, so you’d handle those in your own code.
Pricing
gotoHuman is free for 1 user and 300 reviews a month. Team is $79 a month for 5 users, plus $15 for each extra user, with unlimited reviews and the Reviews API. Run it by has no seats: Free covers 250 requests a month, Builder is $9 for 2,000 and Pro is $29 for 10,000, however many people answer.
Side by side
| Run it by | gotoHuman | |
|---|---|---|
| Set up before the first request | Nothing; the agent describes each request | A review type, designed in the editor |
| Where people act | A personal link in any browser | The inbox, email or Slack; a link for external reviewers |
| Several approvers and a policy | Built in: any, all, majority or n | Assign several reviewers; the rule is yours |
| Proof it was them | Optional email code per person | Their account, or the unique link |
| No answer in time | Verdict expired |
Handled in your code |
| Edit the agent’s output before approving | No | Yes, with retries and version compare |
| Pricing | Per request, no seats. Free 250 a month | Per seat. Free for 1 user and 300 reviews |
When gotoHuman is the better choice
- Your own team reviews agent output every day, and a shared inbox with team metrics helps them.
- Reviewers should edit the draft, send it back for another try with a changed prompt, or compare versions. Run it by only takes a decision and an optional comment.
- You build in n8n or Make and want a ready-made node.
- You need data hosted in the EU or an ISO 27001 vendor. Run it by stores data in the US and isn’t certified yet.
Run it by fits when the people who decide are outside your team or change from one request to the next, several of them must agree, or you need to prove who decided. The same links also collect form answers. More in approvals by link.
Try it with your agent
Free for 250 requests a month, no card. Add the MCP server from Claude Code or Codex with one command, or use the REST API.