On September 30, we hosted a breakfast panel and roundtable for Boston AI Week at Rocket Software in Waltham. The topic was growing revenue with an enterprise go-to-market AI infrastructure. Barbara Goose, CMO of Rocket Software, moderated a conversation with Pam Boiros, Co-Founder and CMO of Women Applying AI; Emma Harney, Director of Digital Experience and AI Strategy at Rocket Software; and Sumi Singh, our Chief AI Officer.
Barbara opened with a quick poll. Most of the room was running pilots or building automations, and only a hand or two went up for a fully agentic system. That set the tone for an honest conversation: nobody has this fully figured out yet, and that’s okay.
Four Takeaways That Stuck with Us
1. Productivity isn’t ROI
Saving hours is a start, but your CFO wants to know what you did with them. Sumi suggested setting up evals instead of a hard ROI deadline and giving AI time to mature, the same way organic marketing takes a few quarters to pay off. The goal is to measure outcomes, not output.
2. Adoption runs on skills, guardrails, and culture
Pam pointed out that one training session won’t stick when the tools change every week. People also need simple, on-the-job rules about what data they can use. Rocket’s rule of thumb: if you can access it in your daily job, you can use it in company-approved AI tools. And managers can make or break adoption by showing where they use AI, and where they struggle with it.
3. Guardrails help you go faster
Think of a mountain road. Without guardrails, everyone drives slowly. With them, people feel safe enough to pick up speed. Clear rules make teams more willing to experiment, not less.
4. Data needs context, not just volume
Sumi noted that most companies collect data but not the relationships between it, and that’s where AI finds the “why.” Emma added that before Rocket built agents, the team built context first. One practical tip from the panel: have someone unfamiliar with a process walk through your documentation step by step. They’ll catch the shortcuts and judgment calls that never got written down.
The Morning in Photos
From the Roundtables
After the panel, the room split into small groups. People shared use cases like detailed ICP matching for prospecting, an agent that researches event attendees ahead of time, and synthetic personas to speed up research. A few challenges came up in nearly every group: managing token spend, defining “AI readiness” when the tools change so quickly, and keeping teams from going in different directions.
Let’s keep the conversation going. Follow us on LinkedIn to see what we’re up to next.