Looking for advice: How to lab/test the new Agentic AI Governance features without an enterprise AI governance environment?

Hi everyone,

I’m really excited about the new Agentic AI Governance capabilities in SailPoint ISC, and I’m looking to experiment with them inside my personal lab environment. Here is my current blocker: while I have the SailPoint ISC features available to test, I do not have access to an actual enterprise environment running Agentic AI workflows, autonomous agents, or the specific service accounts/permissions that usually interact with them.

I want to build a realistic Proof of Concept (PoC) to understand how ISC discovers, governs, and secures these AI identities, but I’m struggling to find a lightweight or mock way to simulate them.

How can I simulate Agentic AI identities? Has anyone successfully built a “mock” setup (e.g., using open-source frameworks like LangChain, AutoGen, or simple Python-based autonomous bots running on AWS/Azure) that can be easily plugged into SailPoint for governance testing?

What target systems should I focus on? To make the test effective, what is the easiest downstream system/directory to connect where these simulated agents would hold their permissions or service accounts?

I would highly appreciate any insights, architectural ideas, or tips from anyone who has already started playing around with this!

Thanks in advance,

I would look at LM Studio, OLlama to run the AI model and act like your claude or openai. I would then look at n8n to build the workflow and tie it into openclaw or hermies for the agent.

Please be careful with true agents being so new in the world securing them is going to be very tricky and complex. So I would also use my router to isolate the network this is all running on and after configured I would only let the VA be the device to be able to speak to the internet.

Please let us know if this help. This will be good everyone to be able to replicate in their own labs.