When you arrive at the event, head down to the session voting area and drop your poker chips into the number box that corresponds with the sessions posted.
We have a ton of amazing session this year, so to help you out, we are previewing all of the proposed sessions leading up to the event. Check them out below!
Attendees can still submit sessions the morning of the event. If you are submitting the day of, please come early so your session is up in time for voting. Voting runs from 8:00 AM - 9:00 AM.
We have a ton of amazing session this year, so to help you out, we are previewing all of the proposed sessions leading up to the event. Check them out below!
Attendees can still submit sessions the morning of the event. If you are submitting the day of, please come early so your session is up in time for voting. Voting runs from 8:00 AM - 9:00 AM.
#1 Stop Being a Jira Clerk: Reclaim 10 Hours a Week for Product Strategy
Eric Reighard, Senior Scrum Manager, Sheetz
Are you a Product Manager or Scrum Master who feels more like a "Jira Clerk" than a strategic leader? In the modern agile landscape, we often spend up to 40% of our week on the "Documentation Tax"—chasing ticket updates, summarizing Slack threads, and translating technical jargon into executive status reports. When the administrative overhead takes over, the human element of product leadership—coaching, discovery, and innovation—is the first thing to suffer.
In this practitioner-led session, Eric Reighard (Senior Scrum Master, Sheetz) shares "in the field" case studies on using AI-native frameworks to automate the ordinary. We’ll explore how to move beyond generic AI "slop" to create a "Scoped Memory" for your team that identifies silent blockers, predicts sprint success, and translates dev-speak into ROI.
Eric Reighard, Senior Scrum Manager, Sheetz
Are you a Product Manager or Scrum Master who feels more like a "Jira Clerk" than a strategic leader? In the modern agile landscape, we often spend up to 40% of our week on the "Documentation Tax"—chasing ticket updates, summarizing Slack threads, and translating technical jargon into executive status reports. When the administrative overhead takes over, the human element of product leadership—coaching, discovery, and innovation—is the first thing to suffer.
In this practitioner-led session, Eric Reighard (Senior Scrum Master, Sheetz) shares "in the field" case studies on using AI-native frameworks to automate the ordinary. We’ll explore how to move beyond generic AI "slop" to create a "Scoped Memory" for your team that identifies silent blockers, predicts sprint success, and translates dev-speak into ROI.
#2 Your AI Feature Isn’t the Problem: Your Design Is
Dr. Lisa Palmieri-Apple, Global Director of Transformation Design & Consulting, Kyndryl
A lot of AI products aren’t failing because the technology is bad, they’re failing because the experience is. Users don’t trust what they don’t understand. They won’t adopt what doesn’t fit how they actually work. No amount of model tuning fixes a product that was never designed with real people in mind.
This session is about what happens when you stop treating AI as the centerpiece and start designing around the human experience instead.
We’ll dig into how co-creation changes the game: bringing users into the process early to shape how AI shows up, behaves, and earns trust. Not as a feedback step at the end but as a core part of how you design.
This will be an interactive session grounded in real-world experience
What attendees will walk away with:
1. A clearer way to think about designing AI from a human perspective
2. Practical co-creation techniques they can apply immediately
3. A better understanding of why their current AI experiences may not be landing and how to fix them
Dr. Lisa Palmieri-Apple, Global Director of Transformation Design & Consulting, Kyndryl
A lot of AI products aren’t failing because the technology is bad, they’re failing because the experience is. Users don’t trust what they don’t understand. They won’t adopt what doesn’t fit how they actually work. No amount of model tuning fixes a product that was never designed with real people in mind.
This session is about what happens when you stop treating AI as the centerpiece and start designing around the human experience instead.
We’ll dig into how co-creation changes the game: bringing users into the process early to shape how AI shows up, behaves, and earns trust. Not as a feedback step at the end but as a core part of how you design.
This will be an interactive session grounded in real-world experience
What attendees will walk away with:
1. A clearer way to think about designing AI from a human perspective
2. Practical co-creation techniques they can apply immediately
3. A better understanding of why their current AI experiences may not be landing and how to fix them
#3 Optimization Obsession: A Case for Humans in the Loop
Claire Johns, Research Analyst, AMG Research
AI is everywhere, but is it bringing real value to your organization? In this talk, we will explore how we can respond when “move fast, break things” actually breaks important things.
Claire Johns, Research Analyst, AMG Research
AI is everywhere, but is it bringing real value to your organization? In this talk, we will explore how we can respond when “move fast, break things” actually breaks important things.
#4 This One Secret Will Spark Innovation
Claire Johns, Research Analyst, AMG Research
Are you wondering how to spark creativity in your work? Good news! It is way more fun than you might think. Join me as I share how knitting has improved my critical analysis skills and share your experience with hobbies making a positive impact on your career too!
Claire Johns, Research Analyst, AMG Research
Are you wondering how to spark creativity in your work? Good news! It is way more fun than you might think. Join me as I share how knitting has improved my critical analysis skills and share your experience with hobbies making a positive impact on your career too!
#5 Bringing Agentic AI into Product
Melissa Wong, Noah Hicks, Graduate Students, CMU
We’ve all seen the demo-ware: a multi-agent stack that looks cool but is too complex, costly, and brittle for production. How do we move from AI hype to a disciplined process for building "right-sized" automation?
In this session, I’m sharing a framework I’m developing to translate real-world workflows into the least complex viable architecture. I want to pressure-test the process with you by walking through a workflow to see how the system decomposes tasks, classifies the AI needed: from deterministic workflows to multi-agent systems and creates a prototype for you with built-in governance.
1. The Classification Engine: Does the proposed AI orchestration for your problem makes sense?
2. The Observability Layer: What metric and trace evaluations do PMs actually need to track?
3. Human-in-the-Loop: Where should the approval layers live to balance speed with safety?
Let’s figure out how to build AI that is sufficient, maintainable, and actually ready for the 6 month roadmap.
Melissa Wong, Noah Hicks, Graduate Students, CMU
We’ve all seen the demo-ware: a multi-agent stack that looks cool but is too complex, costly, and brittle for production. How do we move from AI hype to a disciplined process for building "right-sized" automation?
In this session, I’m sharing a framework I’m developing to translate real-world workflows into the least complex viable architecture. I want to pressure-test the process with you by walking through a workflow to see how the system decomposes tasks, classifies the AI needed: from deterministic workflows to multi-agent systems and creates a prototype for you with built-in governance.
1. The Classification Engine: Does the proposed AI orchestration for your problem makes sense?
2. The Observability Layer: What metric and trace evaluations do PMs actually need to track?
3. Human-in-the-Loop: Where should the approval layers live to balance speed with safety?
Let’s figure out how to build AI that is sufficient, maintainable, and actually ready for the 6 month roadmap.
#6 "Not My Problem" How Product Orgs Break and How to Fix Them
Rick Pollick, Product and Delivery Leader
In matrixed organizations, the most common answer to "who owns this?" is silence. Teams write requirements and hand them off. Leadership chases updates across five tools. Dashboards show what execs want to see instead of what teams need to act on. This session is a candid, interactive discussion about what it actually takes to walk into that environment and fix it, drawn from real experience in large healthcare orgs.
We'll break down how to create visibility that drives real decisions instead of decorating reports, why designing around personas changes everything, and how to coach a team through a culture shift without resorting to command and control. Bring some of your own examples of broken reporting, unclear ownership, or org friction. We'll discuss better approaches together. The best answers are probably already in the room.
Rick Pollick, Product and Delivery Leader
In matrixed organizations, the most common answer to "who owns this?" is silence. Teams write requirements and hand them off. Leadership chases updates across five tools. Dashboards show what execs want to see instead of what teams need to act on. This session is a candid, interactive discussion about what it actually takes to walk into that environment and fix it, drawn from real experience in large healthcare orgs.
We'll break down how to create visibility that drives real decisions instead of decorating reports, why designing around personas changes everything, and how to coach a team through a culture shift without resorting to command and control. Bring some of your own examples of broken reporting, unclear ownership, or org friction. We'll discuss better approaches together. The best answers are probably already in the room.
#7 Cognitive Debt: Use AI to Grow Yourself—Don’t Let Your Brain Atrophy!
Sandeep M Asokan, AI Developer, BNY Mellon
In recent months, research from MIT titled "Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant for essay writing task" has sparked a vital conversation about our partnership with technology. The study posits that long-term, uncritical reliance on AI creates a "cognitive debt"—a measurable decline in neural connectivity, linguistic variety, and behavioral ownership of our work. This raises a critical challenge for the modern professional: How do we leverage the immense productivity of AI without compromising our mental sharpness? To address this, we will explore practical, "anti-debt" strategies designed to keep you cognitively resilient while navigating an AI-augmented landscape.
Sandeep M Asokan, AI Developer, BNY Mellon
In recent months, research from MIT titled "Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant for essay writing task" has sparked a vital conversation about our partnership with technology. The study posits that long-term, uncritical reliance on AI creates a "cognitive debt"—a measurable decline in neural connectivity, linguistic variety, and behavioral ownership of our work. This raises a critical challenge for the modern professional: How do we leverage the immense productivity of AI without compromising our mental sharpness? To address this, we will explore practical, "anti-debt" strategies designed to keep you cognitively resilient while navigating an AI-augmented landscape.
#8 The Promise and Peril of AI for Product Managers
Jason Hong, Professor in Human Computer Interaction, CMU
The goal of this talk is to give realistic expectations about the potential of AI, its limitations, common misconceptions, significant risks that product teams and upper management need to be aware of, and major design issues for products. This session will help product managers navigate the hype, better understand the underlying context that AI systems operate in, and make more responsible and effective product decisions.
This talk is based on Professor Hong’s and his colleagues’ research at Carnegie Mellon University, as well as courses taught there on UX Design and Responsible AI. There is about 75% overlap with the version of this talk I gave at last year's Product Camp
Jason Hong, Professor in Human Computer Interaction, CMU
The goal of this talk is to give realistic expectations about the potential of AI, its limitations, common misconceptions, significant risks that product teams and upper management need to be aware of, and major design issues for products. This session will help product managers navigate the hype, better understand the underlying context that AI systems operate in, and make more responsible and effective product decisions.
This talk is based on Professor Hong’s and his colleagues’ research at Carnegie Mellon University, as well as courses taught there on UX Design and Responsible AI. There is about 75% overlap with the version of this talk I gave at last year's Product Camp
#9 From AI Pilot to Scaled Adoption: What Product Leaders Get Right
Kristi Woolsey, Associate Director, BCG
Many AI pilots fail for predictable reasons: the use case is interesting but not operationally important, the workflow is not redesigned around the technology, stakeholders are misaligned, or adoption is treated as an afterthought. This session shares a practical framework product leaders can use to move from AI enthusiasm to real implementation and measurable value.
Using a public case involving AI, AR, voice interaction, and workflow redesign for a European rail operator, I will walk through how teams identified a high-value problem, aligned product, ops, tech, and business stakeholders, redesigned the work around the user, and supported adoption at scale. The result was a 27% efficiency gain, but the bigger lesson is what made scale possible.
Attendees will leave with a practical lens for evaluating AI use cases, aligning stakeholders, and designing for adoption from the start rather than bolting it on later.
Kristi Woolsey, Associate Director, BCG
Many AI pilots fail for predictable reasons: the use case is interesting but not operationally important, the workflow is not redesigned around the technology, stakeholders are misaligned, or adoption is treated as an afterthought. This session shares a practical framework product leaders can use to move from AI enthusiasm to real implementation and measurable value.
Using a public case involving AI, AR, voice interaction, and workflow redesign for a European rail operator, I will walk through how teams identified a high-value problem, aligned product, ops, tech, and business stakeholders, redesigned the work around the user, and supported adoption at scale. The result was a 27% efficiency gain, but the bigger lesson is what made scale possible.
Attendees will leave with a practical lens for evaluating AI use cases, aligning stakeholders, and designing for adoption from the start rather than bolting it on later.
#10 How to Align Product, Ops, Tech, and Business Around One Roadmap
Kristi Woolsey, Associate Director, BCG
In complex environments, the hardest product problem is often not ideation. It is alignment. Product, operations, technology, and business leaders may all agree change is needed but still struggle to prioritize the right use cases, sequence investments, and coordinate implementation across teams, channels, and physical environments.
This session shares a practical approach for creating a shared roadmap across competing stakeholders using a real enterprise example: a global experience blueprint developed across 500+ locations. I will show how the team projected future needs, identified where digital, physical, and service layers had to work together, created a transparent way to assess current conditions, and prioritized initiatives in a way that was fair, scalable, and tied to value.
Attendees will leave with a framework for aligning stakeholders, making tradeoffs visible, and building roadmaps that can actually be implemented across complex systems.
Kristi Woolsey, Associate Director, BCG
In complex environments, the hardest product problem is often not ideation. It is alignment. Product, operations, technology, and business leaders may all agree change is needed but still struggle to prioritize the right use cases, sequence investments, and coordinate implementation across teams, channels, and physical environments.
This session shares a practical approach for creating a shared roadmap across competing stakeholders using a real enterprise example: a global experience blueprint developed across 500+ locations. I will show how the team projected future needs, identified where digital, physical, and service layers had to work together, created a transparent way to assess current conditions, and prioritized initiatives in a way that was fair, scalable, and tied to value.
Attendees will leave with a framework for aligning stakeholders, making tradeoffs visible, and building roadmaps that can actually be implemented across complex systems.
#11 Sprint to Insight: Agile Strategies for High-Impact AI Products
Ganesh Kumar Suresh, Agile Product Delivery Leader and Enterprise Lean-Agile Change Agent
Data science doesn’t fit neatly into traditional Scrum frameworks. Long experimentation cycles, evolving hypotheses, and delayed feedback loops often lead to slow delivery and limited impact.
This session explores how to bring agility into data science by shifting from delivery-focused thinking to learning-focused systems. We introduce experiment-driven planning, discovery-based standups, and Kanban workflows designed specifically for AI and machine learning teams.
Through practical examples—including retail demand forecasting, recommendation engines, and a coffee shop case study—we show how teams can iteratively move from data to insight to decision. The focus is on enabling fast feedback, reducing waste, and aligning teams around measurable outcomes.
Attendees will gain a repeatable approach to integrate Agile and AI—helping teams deliver value faster, adapt continuously, and turn experimentation into real business impact.
Ganesh Kumar Suresh, Agile Product Delivery Leader and Enterprise Lean-Agile Change Agent
Data science doesn’t fit neatly into traditional Scrum frameworks. Long experimentation cycles, evolving hypotheses, and delayed feedback loops often lead to slow delivery and limited impact.
This session explores how to bring agility into data science by shifting from delivery-focused thinking to learning-focused systems. We introduce experiment-driven planning, discovery-based standups, and Kanban workflows designed specifically for AI and machine learning teams.
Through practical examples—including retail demand forecasting, recommendation engines, and a coffee shop case study—we show how teams can iteratively move from data to insight to decision. The focus is on enabling fast feedback, reducing waste, and aligning teams around measurable outcomes.
Attendees will gain a repeatable approach to integrate Agile and AI—helping teams deliver value faster, adapt continuously, and turn experimentation into real business impact.
#12 Start a Quarterly Advisory Council
Sarah Yack, Experienced Product Manager
It’s painful when we can’t find real problems to solve that will create COMPETITIVE ADVANTAGE. Sometimes you need to get the organization on board with discovery! Do it. Do it now. Start a QUARTERLY ADVISORY COUNCIL to influence continuous discovery. Get your sales teams on board by asking them to invite people. Don’t let them come to it because “you can hear for yourself what they have been trying to tell you for years”, but DO provide rich market analysis, sales enablement feedback, strategy recommendations and a new killer roadmap from it, all while MAPPING IT DIRECTLY TO GROWTH INITIATIVES and the bottom line.
- Learn how to pitch it to your Executive Leadership Team
- Learn how to get the Sales Team to champion it
- Learn about the best format RIVD (Review, Inform, Validate and Discover)
Sarah Yack, Experienced Product Manager
It’s painful when we can’t find real problems to solve that will create COMPETITIVE ADVANTAGE. Sometimes you need to get the organization on board with discovery! Do it. Do it now. Start a QUARTERLY ADVISORY COUNCIL to influence continuous discovery. Get your sales teams on board by asking them to invite people. Don’t let them come to it because “you can hear for yourself what they have been trying to tell you for years”, but DO provide rich market analysis, sales enablement feedback, strategy recommendations and a new killer roadmap from it, all while MAPPING IT DIRECTLY TO GROWTH INITIATIVES and the bottom line.
- Learn how to pitch it to your Executive Leadership Team
- Learn how to get the Sales Team to champion it
- Learn about the best format RIVD (Review, Inform, Validate and Discover)
#13 Superior Innovation Portfolio Performance Workshop
Matthew Arnone, Strategy Consultant
No matter the industry, all leaders must answer two questions about their innovation portfolios: 1) Did I pick the right initiatives? and 2) Will those initiatives deliver to expectations?
To answer these questions reliably, we need to change how we think about innovation portfolio management. In this session, we'll reorient from standard project management concepts toward how three laws of the business universe – Momentum, Disruption, and Volatility – give us powerful insights to address both questions.
Through an eclectic toolset, we'll show you how to rapidly understand how Momentum, Disruption, and Volatility operate in your environment and how you can use them to improve ROI, accelerate delivery, and build seriously unfair advantage in your markets.
You’ll be glad you signed up for our interactive workshop and go home with a practical toolset that enables you to have the conversations you’ve always wanted about your innovation portfolios but couldn’t – until now.
Matthew Arnone, Strategy Consultant
No matter the industry, all leaders must answer two questions about their innovation portfolios: 1) Did I pick the right initiatives? and 2) Will those initiatives deliver to expectations?
To answer these questions reliably, we need to change how we think about innovation portfolio management. In this session, we'll reorient from standard project management concepts toward how three laws of the business universe – Momentum, Disruption, and Volatility – give us powerful insights to address both questions.
Through an eclectic toolset, we'll show you how to rapidly understand how Momentum, Disruption, and Volatility operate in your environment and how you can use them to improve ROI, accelerate delivery, and build seriously unfair advantage in your markets.
You’ll be glad you signed up for our interactive workshop and go home with a practical toolset that enables you to have the conversations you’ve always wanted about your innovation portfolios but couldn’t – until now.
#14 The AI Margin Trap: Pricing Products When Every Feature Costs You Money
Sudhir Gupta, AI Product Strategist, 74 Software
Every AI feature you ship now has a variable cost attached. A delightful experience can quietly become a margin-destroyer at scale or a pricing page so confusing that users churn before reaching the value.
This is an open discussion on the tension between AI as a cost center and AI as a value multiplier. We will walk through the dominant pricing models - flat subscription, usage-based, credits, hybrid, outcome-based and stress-test them against both unit economics and user psychology. The model that protects your margin is often the one that confuses your users most.
I will share lessons from two builds on opposite ends: a consumer EdTech startup pricing for first-time digital learners, and a multi-agent AI system where every model call had a real dollar cost. Then the room takes over - what's working in your products, what's breaking, and where you've had to choose between intuitive and sustainable.
Come with a pricing problem. Leave with sharper questions and a few war stories.
Sudhir Gupta, AI Product Strategist, 74 Software
Every AI feature you ship now has a variable cost attached. A delightful experience can quietly become a margin-destroyer at scale or a pricing page so confusing that users churn before reaching the value.
This is an open discussion on the tension between AI as a cost center and AI as a value multiplier. We will walk through the dominant pricing models - flat subscription, usage-based, credits, hybrid, outcome-based and stress-test them against both unit economics and user psychology. The model that protects your margin is often the one that confuses your users most.
I will share lessons from two builds on opposite ends: a consumer EdTech startup pricing for first-time digital learners, and a multi-agent AI system where every model call had a real dollar cost. Then the room takes over - what's working in your products, what's breaking, and where you've had to choose between intuitive and sustainable.
Come with a pricing problem. Leave with sharper questions and a few war stories.
#15 Customer Conversations: Your Superpower
Steve Johnson, Product Success Coach
Most product managers aren’t struggling because they lack frameworks, data, or AI tools. They’re struggling because they’re talking to the wrong people. Instead of learning directly from customers, they rely on second-hand input—sales requests, executive opinions, support tickets, dashboards, and now AI-generated summaries of all of it. The result? Noise. Misalignment. And products that look good on paper but miss in the market. AI can process more data than you ever will—but it can’t replace direct human understanding. It can’t see hesitation. It can’t hear what’s not being said. And it certainly can’t build trust with a customer who doesn’t believe you yet. That’s where product managers have an advantage—if they choose to use it.
In this session, you’ll learn how Customer Conversations cut through the noise and become your most powerful differentiator. Not as a research technique, but as a leadership behavior that builds market insight, credibility, and better decisions.
Steve Johnson, Product Success Coach
Most product managers aren’t struggling because they lack frameworks, data, or AI tools. They’re struggling because they’re talking to the wrong people. Instead of learning directly from customers, they rely on second-hand input—sales requests, executive opinions, support tickets, dashboards, and now AI-generated summaries of all of it. The result? Noise. Misalignment. And products that look good on paper but miss in the market. AI can process more data than you ever will—but it can’t replace direct human understanding. It can’t see hesitation. It can’t hear what’s not being said. And it certainly can’t build trust with a customer who doesn’t believe you yet. That’s where product managers have an advantage—if they choose to use it.
In this session, you’ll learn how Customer Conversations cut through the noise and become your most powerful differentiator. Not as a research technique, but as a leadership behavior that builds market insight, credibility, and better decisions.
#16 Have a great idea? Patent it for $130. I have 140+ and I’ll show you how.
Ned Uber, Inventor in Residence, Pitt Adjunct Professor, University of Pittsburgh
Most people are stopped from patenting their idea because they think that it costs $10,000. But if you disclose the idea before you file, you could lose it totally. If you’re a good engineer, programmer, product manager, or problem solver, you can write a good provisional patent. You don’t need an attorney. The provisional patent fee is $130 or less.
I’ll explain what is needed and show you examples that I filed and became issued patents. I’ll also share one “secret” technique, morphological analysis, that is little know and even less used, but is incredibly powerful for broadening your design space and strengthening your patent. I’ll walk you through the magic. If you are staring a company your IP is one of your most valuable asset. The stronger it is, the higher the value you’ll receive for your work.
Ned Uber, Inventor in Residence, Pitt Adjunct Professor, University of Pittsburgh
Most people are stopped from patenting their idea because they think that it costs $10,000. But if you disclose the idea before you file, you could lose it totally. If you’re a good engineer, programmer, product manager, or problem solver, you can write a good provisional patent. You don’t need an attorney. The provisional patent fee is $130 or less.
I’ll explain what is needed and show you examples that I filed and became issued patents. I’ll also share one “secret” technique, morphological analysis, that is little know and even less used, but is incredibly powerful for broadening your design space and strengthening your patent. I’ll walk you through the magic. If you are staring a company your IP is one of your most valuable asset. The stronger it is, the higher the value you’ll receive for your work.
#17 How to Create Successful Products by Dan Olsen
Dan Olsen, Product management trainer, consultant, and speaker
We all know that most new product initiatives fail. A big reason why is because the team was focused on the solution they were building and weren’t as clear on the customer problem they were solving. In this talk, product management expert and author Dan Olsen will share advice from his bestselling product book The Lean Product Playbook on how to understand and define customer problems before you start building.
Once you have a well-defined problem space, the next question is: How do we prioritize the different opportunities we could pursue? To answer this question, Dan will share his powerful but easy to use Importance vs. Satisfaction framework, which will help your team prioritize the opportunities that will create the most value for your customers.
Dan will illustrate his frameworks with real world examples and case studies. Come learn how to increase your odds of product success.
Dan Olsen, Product management trainer, consultant, and speaker
We all know that most new product initiatives fail. A big reason why is because the team was focused on the solution they were building and weren’t as clear on the customer problem they were solving. In this talk, product management expert and author Dan Olsen will share advice from his bestselling product book The Lean Product Playbook on how to understand and define customer problems before you start building.
Once you have a well-defined problem space, the next question is: How do we prioritize the different opportunities we could pursue? To answer this question, Dan will share his powerful but easy to use Importance vs. Satisfaction framework, which will help your team prioritize the opportunities that will create the most value for your customers.
Dan will illustrate his frameworks with real world examples and case studies. Come learn how to increase your odds of product success.
#18 AI, Agents, and UX: What the Science Says
Sauvik Das, Associate Professor, CMU's Human-Computer Interaction Institute
There's a lot of hype around AI, agents, and what they mean for UI/UX. Drawing on experience as an HCI professor at CMU, I'll cover what the science has to say about questions like: Can I use AI to critique and fix my UIs? Can I use AI agents to do user research? What are agents anyway and are they going to take my job?
Sauvik Das, Associate Professor, CMU's Human-Computer Interaction Institute
There's a lot of hype around AI, agents, and what they mean for UI/UX. Drawing on experience as an HCI professor at CMU, I'll cover what the science has to say about questions like: Can I use AI to critique and fix my UIs? Can I use AI agents to do user research? What are agents anyway and are they going to take my job?
#19 Is Product Development More Like Poker or Chess? A Poker Player's Lessons
A. Robert Wasson, Ph.D., Engineering and Innovation Management (ETIM), CMU
Is new product development more like poker or chess? From our work at CMU that is adapted from Annie Duke’s book, Thinking in Bets, we now have a better understanding of the similarities between poker and new product development. We see that there are steps that professional poker players take that may be applied to new product development.
In our discussion, we highlight 6 key concepts that leaders of new product development need to understand:
1) Great process, bad outcome.
2) Life is poker, not chess
3) Speak in probabilities
4) Imagine failure before it happens
5) Recalibration is not weakness
6) Sunk costs become an anchor in decision-making
We discuss how other authors refine Duke’s thinking and identify certain actions that product development leaders can take to mitigate these issues.
A. Robert Wasson, Ph.D., Engineering and Innovation Management (ETIM), CMU
Is new product development more like poker or chess? From our work at CMU that is adapted from Annie Duke’s book, Thinking in Bets, we now have a better understanding of the similarities between poker and new product development. We see that there are steps that professional poker players take that may be applied to new product development.
In our discussion, we highlight 6 key concepts that leaders of new product development need to understand:
1) Great process, bad outcome.
2) Life is poker, not chess
3) Speak in probabilities
4) Imagine failure before it happens
5) Recalibration is not weakness
6) Sunk costs become an anchor in decision-making
We discuss how other authors refine Duke’s thinking and identify certain actions that product development leaders can take to mitigate these issues.
#20 When AI Owns the “How,” Product & Design Must Lead the “Why”
Alina Bengert-Lombardi, UX-driven product strategist and design leader
Traditionally, software development centered on the “how”: architecture, frameworks, APIs, and implementation details. Product and design teams defined user needs and experiences within those technical constraints.
AI-assisted development is shifting that model.
Developers are increasingly prompting AI with desired outcomes instead of manually building every solution from scratch. As AI takes on more of the implementation layer, developers are moving closer to defining the “what.”
This changes the role of product and design.
If AI accelerates the “how,” and developers increasingly shape the “what,” then product and design must lead the “why”:
Why does this matter?
Why is this worth solving?
Why should users care?
This session explores how AI is reshaping product development and why intentionality matters more than ever.
Interactive discussion encouraged.
Alina Bengert-Lombardi, UX-driven product strategist and design leader
Traditionally, software development centered on the “how”: architecture, frameworks, APIs, and implementation details. Product and design teams defined user needs and experiences within those technical constraints.
AI-assisted development is shifting that model.
Developers are increasingly prompting AI with desired outcomes instead of manually building every solution from scratch. As AI takes on more of the implementation layer, developers are moving closer to defining the “what.”
This changes the role of product and design.
If AI accelerates the “how,” and developers increasingly shape the “what,” then product and design must lead the “why”:
Why does this matter?
Why is this worth solving?
Why should users care?
This session explores how AI is reshaping product development and why intentionality matters more than ever.
Interactive discussion encouraged.
#21 Why Enterprise AI Copilots Fail After the Demo
Siddhant Sagar, Senior Product Manager, SkySync
Every enterprise vendor is shipping an AI Copilot. Most get strong demo reactions and weak adoption. The gap between ""this is impressive"" and ""I use this every day"" is where most enterprise AI products quietly die.
Drawing on PM experience at Microsoft on Dynamics 365, I'll walk through why the dominant design pattern is wrong. Most copilots are features bolted onto existing software. The ones that get adopted are workflow-native: they understand the user's work and intervene at the right moment, with the right control, in a way the user trusts.
I'll cover three things:
1. The trust gap - why ""high accuracy"" is not ""user trust,"" and what to measure instead of AI engagement.
2. Copilot vs automation - when users want the AI to act for them vs. stay in control, and how to design for both.
3. Workflow-context - why great enterprise AI starts with the workflow, not the LLM.
Closing with examples from CRM and productivity. Q&A in the final 10 min.
Siddhant Sagar, Senior Product Manager, SkySync
Every enterprise vendor is shipping an AI Copilot. Most get strong demo reactions and weak adoption. The gap between ""this is impressive"" and ""I use this every day"" is where most enterprise AI products quietly die.
Drawing on PM experience at Microsoft on Dynamics 365, I'll walk through why the dominant design pattern is wrong. Most copilots are features bolted onto existing software. The ones that get adopted are workflow-native: they understand the user's work and intervene at the right moment, with the right control, in a way the user trusts.
I'll cover three things:
1. The trust gap - why ""high accuracy"" is not ""user trust,"" and what to measure instead of AI engagement.
2. Copilot vs automation - when users want the AI to act for them vs. stay in control, and how to design for both.
3. Workflow-context - why great enterprise AI starts with the workflow, not the LLM.
Closing with examples from CRM and productivity. Q&A in the final 10 min.
#22 Agentic Failures 101: How Agents Fail Differently and How to Catch it
Melissa Wong, MS candidate, Carnegie Mellon's Heinz College
When your AI solution is LLM calls, evaluating it can be straightforward: did it answer right? When it's an agent (multi-step, tool-using), that question misses most of what can go wrong.
This is Agentic Failures 101 for PMs. No prior eval experience needed. We'll cover:
- LLM failures vs. agentic failures, in plain language
- Common metrics used to evaluate agents today: what each catches, and where each one fails
- Use Case: How Google's ADK suite does and doesn't catch agentic failures when grounded in a structured failure taxonomy
- Tool: How to use that taxonomy as a bridge so PMs and engineers share one vocabulary for what broke and why
You'll leave with two artifacts: understanding of LLM vs Agentic Failures and the 70-mode Agentic Failure Taxonomy.
Closing activity: give me a failure you see often in your agent, and I'll show you the range of failure modes that can cause it.
Melissa Wong, MS candidate, Carnegie Mellon's Heinz College
When your AI solution is LLM calls, evaluating it can be straightforward: did it answer right? When it's an agent (multi-step, tool-using), that question misses most of what can go wrong.
This is Agentic Failures 101 for PMs. No prior eval experience needed. We'll cover:
- LLM failures vs. agentic failures, in plain language
- Common metrics used to evaluate agents today: what each catches, and where each one fails
- Use Case: How Google's ADK suite does and doesn't catch agentic failures when grounded in a structured failure taxonomy
- Tool: How to use that taxonomy as a bridge so PMs and engineers share one vocabulary for what broke and why
You'll leave with two artifacts: understanding of LLM vs Agentic Failures and the 70-mode Agentic Failure Taxonomy.
Closing activity: give me a failure you see often in your agent, and I'll show you the range of failure modes that can cause it.
#23 AI Copilot, Not Autopilot: When PMs Should Not Use AI
Archana Vijayan Kamath, Product Leader
I can draft your PRD, summarize customer interviews, generate roadmap ideas, and make you look productive fast. But can it tell you what actually matters? Can it sense when a customer does not trust your product? Can it navigate a tense stakeholder tradeoff? Can it own the consequences of a bad product decision?
This session is for PMs who want to use AI without becoming dependent on it. We will explore the hidden traps of AI in product management: shallow discovery, generic strategy, overconfident prioritization, biased recommendations, and polished documents that mask weak thinking.
You will leave with a practical decision framework for knowing when to use AI, when to keep a human in the loop, and when product judgment must stay fully human. Expect real examples, audience discussion, and immediately usable takeaways for discovery, roadmapping, stakeholder communication, and product strategy.
Archana Vijayan Kamath, Product Leader
I can draft your PRD, summarize customer interviews, generate roadmap ideas, and make you look productive fast. But can it tell you what actually matters? Can it sense when a customer does not trust your product? Can it navigate a tense stakeholder tradeoff? Can it own the consequences of a bad product decision?
This session is for PMs who want to use AI without becoming dependent on it. We will explore the hidden traps of AI in product management: shallow discovery, generic strategy, overconfident prioritization, biased recommendations, and polished documents that mask weak thinking.
You will leave with a practical decision framework for knowing when to use AI, when to keep a human in the loop, and when product judgment must stay fully human. Expect real examples, audience discussion, and immediately usable takeaways for discovery, roadmapping, stakeholder communication, and product strategy.
#24 Designing for People Who Know More Than You
Valerie Caña, Graduate Student, CMU
Designing for expert users is different from designing for general consumers. Experts rely on specialized language, tacit knowledge, complex workflows, and judgment built over years of experience. If product teams simplify too early, they risk flattening the very complexity that makes expert work effective.
In this interactive session, we’ll explore how product teams can research and design for sophisticated domains. We’ll cover expert interviews, workflow mapping, decision-point research, mental models, and artifact walkthroughs. Then we’ll look at how AI-assisted prototyping can help teams turn expert workflows into testable product concepts faster without losing product judgment.
Attendees will leave with a practical framework for researching expert users, translating complexity into product direction, and using AI prototypes as learning tools rather than polished demos.
Valerie Caña, Graduate Student, CMU
Designing for expert users is different from designing for general consumers. Experts rely on specialized language, tacit knowledge, complex workflows, and judgment built over years of experience. If product teams simplify too early, they risk flattening the very complexity that makes expert work effective.
In this interactive session, we’ll explore how product teams can research and design for sophisticated domains. We’ll cover expert interviews, workflow mapping, decision-point research, mental models, and artifact walkthroughs. Then we’ll look at how AI-assisted prototyping can help teams turn expert workflows into testable product concepts faster without losing product judgment.
Attendees will leave with a practical framework for researching expert users, translating complexity into product direction, and using AI prototypes as learning tools rather than polished demos.
