I would Like to know if there is a way to configure my own Azure Foundry or AWS bedrock models in GEN AI connect rules ? If we are using pega cloud or using our own azure cloud ( we have Prembridge connection), Currently Pega provides these default models in GEN AI connect setting.. As a client we have our Azure subscription and deployed a model there.and i have a end point ready to connect to the model.
Connecting to Own Model support is not available in 25.1 version. However, if you’re leveraging the Agent instead of Gen AI connect, so that you can connect your Model through MCP or A2A or REST APIs if required.
Yes i tried that approach and yes it could work.. Lets say i have my own model which i connected via Exteral agent/MCP or REST… and that gives me some output. would that data (output from my agent) will it be still shared with the models listed in the models tab of the Pega GEN AI rule ?
Yes, it’s possible. I have built a similar prototype and I was able to carry forward the data to Pega GEN AI supported models. Please check this article,
Pega MCP Agent Debate Pattern: Governing Multi‑Agent Intelligence at Enterprise Scale
I mean i do not want to share the data with Pega models.
-
I create a Agent rule.
-
In Advannced tab, i called external agent using MCP protocol
As i am calling my own external agent , passing the promot and getting the response back from my external agent..
What is the purpose of these model now in settings tab:
I DO NOT want to share the details what i get from my external agent. i just want to use that data AS IS.
In this scenario, would the data still be shared with models present in settings tab ? is there a way not to share any data to the models present in settings tab ?