I’ve just shared a new video titled Pega Predictable AI: 5 AI Placement Patterns (Agents + GenAI Connect), and I wanted to invite you to take a look and join the conversation.
In this session, I walk through five practical ways to embed AI directly into workflows—showing where agents and GenAI Connect deliver the most value. Rather than focusing on theory, the video uses an example of end‑to‑end agentic workflow to demonstrate how AI can be applied for conversational intake, document analysis, agentic assessment, guided deep dives, and turning unstructured inputs into actionable outputs .
The goal is simple: help practitioners better understand how to place AI intentionally and predictably so it supports the work, stays governed, and brings value. If you’re designing workflows today and wondering where AI can be leveraged, this video should give you concrete patterns you can reus e.
Jay, this is great. It really helps in understanding where the Pega agents augment that workflow connective tissue. A natural next step is to then outline the use case details that help drive value
Thanks @cantc , placement is one aspect of the overall use case. Early on we have been using this use case design template to meet agentic goals. We continue to evolve , but this has been useful early on as we describe agentic goals and design. Would love feedback on its usefulness.
Many real-world automation projects have things like the “Automation Assessment” portion of your demo, only using rules: analyze a bunch of input, produce a list of actions as an output. I’m curious to hear your take on when rules make more sense here vs. an agent.
My own sense is: it depends. On the one hand, if a relatively simple set of rules can completely capture all possibilities with 100% accuracy, then rules are the way to go. On the other hand, if even a very complex set of rules still require a human to review the output, I’m increasingly of the mind that an agent is better: simpler to implement, and more flexible.
Of course that leaves a lot of gray area in the middle, at least as far as I can see.
