Skip to main content
A custom action lets your agent reach your own software during a call: look up an order, check an account, or create a job in a system the app store does not cover. It is the one part of Brilo that needs someone technical, and it is usually about an hour of their time.

​
What it is

Most actions connect to a tool Brilo already knows about. A custom action connects to yours. Your developer describes where your system lives and what to send it, and after that your agent can use it like any other action, including reading the answer back to the caller mid call.

​
When you need one

  • Your tool is not in the app store. Check the app store first, since the list grows. See Connect your tools.
  • You want your agent to read your own data to a caller, for example an order status or a balance.
  • You want a record created in your own system rather than in a third party tool.
You do not need one for booking appointments in Google Calendar, updating HubSpot, or sending texts and emails. Those already exist.

​
What to send your developer

Send these four things and they will not need to come back to you:
  1. What you want to happen, in your own words. For example: “when a caller asks about their order, the agent should look it up in our system and read back the status”.
  2. Which system it lives in, and who administers it.
  3. The developer reference: Custom actions.
  4. Your API key, from developer settings in your dashboard.
What they will need from your side is the address of the thing your agent should call, how it authenticates, and what it expects to receive. If your system has documentation, send that too.

​
What it is fair to expect

For a developer who has built an integration before, a single custom action is about an hour. If you are told it is a multi week project, either the requirement is larger than you described or something has been misunderstood: ask them to look at the reference page first.

​
After it is built

The custom action appears alongside the others in your agent’s Actions tab, and from there it is yours to manage:
  • Tell your agent when to use it, in the Prompt tab. This is still the most common reason an action never runs.
  • Decide whether it runs during the call, which the caller waits for, or after it.
  • Decide what your agent says when your system does not answer. See What are actions.

​
FAQ

Yes, with a custom action. Once built, your agent can look something up and read the answer back to the caller, or create a record in your system, exactly like any built in action.
Yes. This is the main part of Brilo that needs someone technical, and it is normally about an hour of work. Everything a caller experiences on a call can otherwise be set up without one. See Do I need a developer.
About an hour for a developer who has built an integration before, plus your time deciding when your agent should use it and testing it on a real call.
Yes, and that is the main reason to build one. The action has to run during the call for this, so it needs to be quick: the caller is waiting while it happens.
The action fails and the call continues. Your agent tells the caller it could not complete the task, and what it offers next comes from your instructions, so write that fallback down.
No. A custom action runs during a call and can bring information back. A webhook fires after the call and only sends data out. Use a custom action for mid call lookups and a webhook for records and reporting. See Send data to your systems.
No, they are part of your plan. What costs you is the developer time to build one.