Product update · Integrations

Actions: let your chatbot look things up and send data to your API

Connect your chatbot to your own API. It can check an order, look up stock or submit a request mid-conversation, on your website and social channels.

Gopi Krishna Lakkepuram

· 3 min read

Shipped
Area
Integrations
Available on
Pro and Max

Some questions can't be answered from a knowledge base. "Where's my order?", "Is this in stock?" and "What's my account balance?" need live information from your own systems. Until now, your chatbot could only point people elsewhere. With Actions, it connects to your API and gets the answer during the conversation.

What's new

An Action is a request your chatbot can make to an API you run. You describe when it should be used and what it sends, and the AI decides when a conversation calls for it.

  • Look up information. Order status, stock levels, account details: anything your API can return.
  • Submit data. Send a lead, a support ticket or a request into your own system.
  • Works on every channel. Actions run in website chat, WhatsApp, Instagram DM and Facebook Messenger.
  • A guided setup. Each Action is set up in five steps: Purpose, Request, Authentication, Inputs and Review. Start from Information lookup, Data submission or Custom API.
  • Your credentials stay private. Choose No authentication, Bearer token, API key header or Connected account. Credentials are stored encrypted. They're never shown to the AI or to your visitors, and once saved they can be replaced but not read back.
  • A status line for visitors. You can set short progress text, shown to the visitor while the Action runs.

How to use it

  1. Open your chatbot in Studio and go to the Integrations tab, then Actions.
  2. Pick Information lookup, Data submission or Custom API.
  3. Under Purpose, give the Action a name and explain when the chatbot should use it.
  4. Under Request, enter your endpoint URL and HTTP method.
  5. Under Authentication, choose how your API checks the request and enter the credential.
  6. Under Inputs, list what the chatbot should collect from the visitor, such as an order number or email.
  7. Check everything under Review, then click Add to draft.
  8. Click Save draft and test to try it in a fresh preview of your draft, then publish when it works.

Good to know

  • Up to 10 Actions per chatbot, each with up to 20 inputs.
  • Requests run from Hyperleap's servers, not the visitor's browser. Testing a draft makes real requests too, so use test accounts and data where you can.
  • 20-second timeout. If your API doesn't answer in time, the request is stopped.
  • Public endpoints only. Requests to private or internal network addresses are blocked, and redirects aren't followed, so give the final URL.
  • Visitors see the answer, not the data. The raw request and response stay between Hyperleap and your API. Visitors only see the chatbot's reply.
  • It connects to your API. An Action talks to an endpoint you control. It isn't a ready-made integration with any particular software, so the system on the other end needs an API the chatbot can call.
  • Available on Pro and Max. On other plans, saved Actions stay readable but don't run.
  • New to Actions? Read the Actions help guide or compare plans and pricing.
For developers

Inputs can be strings, numbers or booleans, placed in the URL path (as {inputName} placeholders), the query string, or a flat JSON body. GET Actions use path or query inputs only. Non-secret headers can be added alongside the credential (up to ten). Responses are read up to 64 KB.

Explore Hyperleap AI

An AI front desk that answers customer questions, collects contact details, and books appointments on your website and messaging channels.