LangGraph | gotoHuman

An AI agent built with LangGraph integrating gotoHuman to keep a human in the loop.

gotoHuman provides you with a central dashboard to review AI‑generated content, approve critical actions or provide input. And it seamlessly integrates with the AI stack of your choice.

This example workflow uses our Typescript SDK.

It takes the email address of a new sales lead and drafts a personalized initial email outreach. The draft is reviewed and revised by a human before it gets sent out.

Set it up

Create a review form

In gotoHuman, you can simply import the form template used here with the ID OmmAnhbnWmird3oz60q2.
For the webhook, enter the URL where you deploy this app. It is called for each review response to resume your graph.

Reviewers will find new pending reviews in their gotoHuman inbox. You can also opt-in to receive a short-lived public link that you can freely send to reviewers.

Deploy this agent

Clone this Next.js repo, deploy it (e.g. to Vercel) and set up your environment variables

OPENAI_API_KEY = sk-proj-XXX
GOTOHUMAN_API_KEY=XYZ
GOTOHUMAN_FORM_ID=abcdef123
POSTGRES_CONN_STRING="postgres://..."

Run it

Trigger your agent whenever you get your hands on a new email address:

Manually in gotoHuman

Create a new trigger form with a text input field with ID email and include the same webhook as above. You'll see a trigger button appear in gotoHuman.

Via API

HTTP POST [DEPLOY_URL]/api/agent

{
  "email": "[email protected]"
}

Find a new request for review in your gotoHuman inbox as soon as the agent is done with its' research, drafted an outreach message and needs approval.