Productside Webinar
Agentic AI – Product Management Friend or Foe?
How Agentic AI is redefining both your product and your role.
Date:
Time EST:
Agentic AI isn’t just another technological innovation—it’s reshaping entire industries. As companies like Microsoft and Salesforce move toward agent-driven ecosystems, SaaS (as we know it) is evolving. The role of the product manager is also transforming, from strategy author to system curator, with the need to balance trust and control in AI-driven systems.
Join Dean Peters and Roger Snyder for a deep dive into the trends shaping 2025 and beyond. Learn how to adapt, lead, and thrive in this new era.
What you’ll gain:
- Insights into how Agentic AI is disrupting PM roles and SaaS business models.
- Practical strategies for aligning autonomous systems with user and business needs.
- How to evolve your approach to guide AI-driven ecosystems effectively.
- Techniques to stay relevant in a rapidly changing landscape.
Welcome and Housekeeping
Roger Snyder | 00:00–02:30
Welcome everyone to Productside’s webinar, “Agentic AI – Product Management Friend or Foe?” I’m Roger Snyder and I’ll be your host. We are recording; you’ll get the replay and resources. Use the Q&A panel for questions—we’ll save time at the end.
Dean Peters | 02:30–03:30
Thanks, Roger. I’m Dean Peters, product advisor at Productside. Today we’ll explore what agentic AI actually is, what it isn’t, and how PMs can use it without losing the plot.
About Productside and Overview
Roger Snyder | 03:30–05:00
For anyone new: Productside helps product teams connect strategy to outcomes. Today’s agenda: definitions, foundations, impacts on PM work, examples, and guardrails—plus polls and Q&A.
Poll #1 – Where Are You on the Agentic AI Spectrum?
Roger Snyder | 05:00–06:00
Quick pulse: New to agentic AI, experimenting, or shipping with it? Vote now.
Dean Peters | 06:00–06:30
I’m seeing a lot of “experimenting.” Perfect—this talk is aimed at you.
Warm-Up Activity – Describe Agentic AI in Five Words
Roger Snyder | 06:30–07:30
In chat, describe agentic AI in five words. Curious to see your instincts.
Dean Peters | 07:30–08:30
Favorites so far: “Autonomous, useful, risky, fast, emergent.” That captures the tension nicely.
What Is Agentic AI?
Dean Peters | 08:30–13:00
Agentic AI refers to systems that take goals and autonomously plan and execute multi-step tasks—invoking tools, calling APIs, and coordinating sub-agents. Think “goal in, outcomes out,” with humans assigning purpose and guardrails.
Foundations of Agentic AI
Dean Peters | 13:00–18:00
Three pillars:
1) **Reasoning** – planning and reflection loops.
2) **Tools** – actions like search, code, or internal systems.
3) **Memory** – short/long-term context to improve over time.
Without all three, you don’t have agency—you have automation.
Poll #2 – Which Products Are Agentic?
Roger Snyder | 18:00–19:00
Which of these do you consider agentic: scheduling assistants, autonomous research tools, code copilots, or analytics bots?
Dean Peters | 19:00–19:30
Trick answer—some are, some aren’t. Agency depends on autonomy and goal-seeking, not brand names.
Autonomy, Accountability, and Agency
Dean Peters | 19:30–24:00
Autonomy without accountability is chaos. The PM’s role is setting **goals, constraints, and evaluation signals**. We own outcomes—even when agents take actions.
How Agentic AI Impacts PM Work
Dean Peters | 24:00–29:00
Discovery, research, competitive analysis, and synthesis get faster. Roadmapping and prioritization get richer inputs. But the bar for **storytelling, ethics, and governance** goes up.
Poll #3 – Which PM Tasks Could Be Automated?
Roger Snyder | 29:00–30:00
Which tasks should an agent help with first: research, reporting, grooming, or experiments?
Dean Peters | 30:00–30:30
Research and reporting usually win—instant leverage.
Agentic AI in Market Research and Story Generation
Dean Peters | 30:30–35:00
Agents can scrape public data, cluster insights, draft narratives, and propose hypotheses. Keep humans in the loop to check bias and business fit.
Evolving the Product Operating Model
Dean Peters | 35:00–39:30
Shift from artifact-creation to **decision orchestration**. PM + Ops + Data define guardrails; agents execute playbooks; leaders review signals, not slides.
Impact on the SaaS and Business Model
Dean Peters | 39:30–43:30
Agentic workflows change pricing (usage-based), packaging (task bundles), and value proof (jobs-done). Make costs observable and outcomes auditable.
The Agentic Ecosystem and Governance
Dean Peters | 43:30–47:30
RACI for agents: who sets goals, approves tools, reviews logs, and handles incidents? Add **kill-switches, audit trails, and policy tests**.
Experimenting with Agentic Tools
Dean Peters | 47:30–50:30
Start with low-risk domains. Measure value with before/after time-to-insight and error rates. Celebrate learning, not just lift.
Live Demo – LangFlow and Blueprint Examples
Dean Peters | 50:30–55:00
Quick walkthrough of a research agent in LangFlow: goals → tools → memory → evaluation. Then a “blueprint bot” scaffolding product charters.
Key Takeaways for Product Managers
Dean Peters | 55:00–58:00
Own the goals and guardrails. Treat agents like interns that never sleep: helpful, fast, and supervised. Document decisions and measure outcomes.
Q&A and Closing Remarks
Roger Snyder | 58:00–01:02:00
Q1: “What skills should PMs learn first?”—Prompting, evaluation design, and governance.
Q2: “How do we avoid hallucinations?”—Constrain tools, add retrieval, and require evidence in outputs.
Dean Peters | 01:02:00–01:03:00
Thanks for joining—keep experimenting, and keep humans in the loop.
Webinar Panelists
Dean Peters