Create a Connector
Build a custom Connector once and reuse it across as many AI Fellows and Learning Paths as you need. Frontierz handles invoking the Connector during conversations. Your job is to describe what it does and where it points.
Write a description. The AI Fellow reads this when it decides whether to use the Connector mid-conversation, so make it specific and plain-language.
Pick the type:
- HTTP for a direct call to an endpoint you control.
- MCP for a connection to an AI-compatible data source over the Model Context Protocol.
Provide the endpoint and the authentication token for whichever type you picked.
Optionally, define the input fields the Fellow needs to gather from the user before invoking the Connector (an account ID, a date range, that kind of thing).
The description matters more than it looks. The AI Fellow uses it to decide when to invoke a Connector during a free-form conversation. A vague description ("calls the CRM") leads to vague triggering. A specific one ("returns the customer's open tickets given their account ID") fires when the user mentions a customer or asks about open tickets.
Test it before you publish it
Once the Connector is saved, you can send a real call to your endpoint without having to start a voice session.
A failed test is the cheapest way to catch a wrong endpoint URL, a missing auth token, or a field your backend isn't expecting yet.
Where to use it
Once the Connector exists, attach it from either side:
- From an AI Fellow, to make it available to that Fellow's conversations.
- From a Learning Path, to make it available across every session in the program.