Almost every hand went up for "Who wants to turn on Now Assist?" but only a few for "Who's running it in production?"

Energized after an action-packed week at Knowledge25 with clients and fellow IBMers!

Watching the platform evolve over the past six years has been inspiring — especially this last year as ServiceNow dove head-first into Agentic AI and cemented its foothold in CSM.

Below are some favorite — and not-so-favorite — moments:

Highlights

1. Time with IBMers & clients — Cheesy, but it always tops the list.

2. Jensen Huang fireside chat with Bill McDermott — Loved the personal backstory — and the drop of Nemotron 15B, trained on ServiceNow domain data. Clear signal that the ServiceNow x NVIDIA combo is serious about Enterprise AI.

3. Deep-dives on WDF (ServiceNow Data Fabric) — Virtualizing data inside the platform is a must for Agentic AI. If you'd told me a few years ago that ServiceNow would enter the data-fabric game, I wouldn't have believed you. I respect the ambition.

4. Marc Hayem's session on Cognitive Customer Care — A buzzword-free AI session — finally! Real client stories where IBM Consulting drove up to 70% containment in contact-center volumes. Bonus: seeing Virtual Agents I helped build on the big screen.

5. Conan O'Brien and Jessica Williams Unscripted Keynote — Inspiring, hilarious, best way to end the week. "Success is a non-linear journey."

Things I could've skipped

1. The "one-click AI" myth — In one session, almost every hand went up for "Who wants to turn on Now Assist?" but only a few for "Who's running it in production?" Getting Gen AI production-ready is tougher than flipping a switch. Love ServiceNow's build-fast/fail-fast ethos, but let's be clear on the hurdles.

For context: IBM runs one of the largest production deployments of Now Assist — real users, real workload, no cutesy POCs. I was lucky enough to lead it.

2. "AI" demos that are just an OpenAI API call — Bill's vision is platform-embedded AI. A Script Include that ships data off to OpenAI isn't quite it — and we could do that in 2023.

3. LLM-written session blurbs — "Optimize agent-driven AI workflows to maximize productivity and brand satisfaction." Huh? One of my clients guessed an LLM wrote that — and I think she nailed it.

I'll spend the weekend like the last one: knee-deep in Agent Builder, testing new use cases for clients.