Notes on RPA, agentic AI, and ServiceNow — mostly the operational lessons that don't show up in the docs.
RPA and agentic AI solve different problems. Here's how to tell which one you actually need, and when you need both.
RPA gets sold as a universal fix for manual work. In practice it solves a narrow class of problems extremely well — and fails predictably outside it.
A breakdown of the three pieces that turn an LLM from a chatbot into an agent that gets things done — and where each one tends to fail.
A few Flow Designer patterns that hold up in production: fan-out approvals, timeout escalation, and keeping flows testable.
ServiceNow can trigger and monitor RPA bots as first-class flow steps. Here's the architecture that keeps that integration from turning into a black box.
The problems that show up once an RPA program grows beyond a handful of bots aren't technical — they're operational. Here's what actually breaks.