@DharaniM5625

To use your own LLM in an on-prem Pega setup, expose your model as a secure HTTPS REST endpoint (chat/completions style) through your network or an internal gateway.

In Pega, create a GenAI provider/connector that calls this endpoint, set the base URL, headers, auth (API key/OAuth), and any required request/response mapping.

Test the connector to confirm Pega can send prompts and read the generated text back successfully.

Then go to App Studio → GenAI Connect (your use case) → Settings, and select your newly created provider/model from the GenAI model dropdown.

Save and publish so the use case runs through your connector, meaning Pega now routes all GenAI requests to your LLM instead of the ootb models.

@DharaniM5625

If your custom AI agents use MCP, you can build custom middleware or an adapter to translate MCP messages into RESTful formats that Pega can invoke.