What AI means for your product strategy | Paul Adams (CPO of Intercom)
Paul Adams, CPO at Intercom, shares lessons from past failures and deep insights on integrating AI into product strategy. He discusses Intercom's all-in approach to AI, key frameworks, and the importance of simple, customer-centric product development.
Deep Dive Analysis
18 Topic Outline
Learning from Public Speaking Failure at Cannes
Experiences with Failed Social Products at Google
The Importance of Embracing Failure in Product Development
Intercom's 'Ship Fast, Ship Early, Ship Often' Principle
Integrating AI into Product Strategy: All In vs. Skepticism
Making Time for AI Learning and Experimentation
Strategic Questions for AI Product Integration
Intercom's Pivot to AI and the Development of Fin
Mind-Blowing Capabilities and Implications of AI
Structuring Teams and Avoiding Pitfalls with AI Products
Staying Up-to-Date on Emerging AI Technology
Hurdles and Building Conviction for AI Implementation
Paul's 'Before-After' Framework for Strategic Shifts
Lessons Learned from Intercom's Pricing Strategy
Paul's 'Differentiation vs. Table Stakes' Framework
The Concept of 'Swinging the Pendulum' in Business
Paul's 'Product Market Story Fit' Framework
Intercom's Use of Jobs to Be Done and Four Forces Frameworks
5 Key Concepts
Before-After Moments
These are pivotal points where a significant change occurs, making nothing the same as before. After such a moment, it's crucial to re-evaluate assumptions and learn from customer feedback to understand the impact of the change.
Differentiation vs. Table Stakes
This framework categorizes product features into two types: 'differentiation' (what makes a product uniquely better and attractive to customers) and 'table stakes' (basic, often boring features required to compete in a market). Companies must balance investment in both, as neglecting table stakes can hinder adoption, even with strong differentiation.
Swinging the Pendulum
This concept describes the tendency to overcorrect when addressing an undesirable state in a company or product. For example, focusing too much on differentiation might lead to neglecting table stakes, causing a subsequent overcorrection towards table stakes, and so on. The challenge is to find balance without extreme shifts.
Product Market Story Fit
Beyond traditional product-market fit (building the right product for the right market), this framework emphasizes the importance of a compelling and clear 'story.' A great product in a great market can fail if its value proposition is convoluted or poorly communicated, failing to capture customer attention.
The Four Forces Framework
Derived from Jobs to be Done, this framework helps understand why people switch or adopt new products. It considers the attraction of the new solution, the reasons against adopting it (e.g., existing habits), and anxieties about the new solution, providing a comprehensive view of customer decision-making.
9 Questions Answered
Product leaders should start by understanding their product's core purpose and then ask if AI can perform or augment those functions. This involves mapping what the product does against AI's capabilities (like writing, summarizing, reasoning, taking actions) to determine if a foundational strategic change is needed.
It requires dedicated time; one must actively make time to read and experiment with AI tools. Playing with tools like ChatGPT, including its vision capabilities, and exploring platforms like Rewind.ai, is essential to understand their potential and limitations.
Intercom viewed ChatGPT's launch as a 'before-after moment' and almost entirely re-evaluated its strategy from first principles, focusing on how AI could transform customer support. This led to a 'bet the farm' mindset and the development of their AI chatbot, Fin.
While early, some Intercom customers using Fin have seen huge success, with the AI chatbot answering 50-70% of inbound questions. The biggest challenge has been helping customer support teams adapt to the organizational change and new roles, rather than the technology itself.
Beyond answering questions, AI's ability to reason, write complex code, parse imagery (e.g., diagnosing bike problems from a photo, translating ancient texts), and even replicate human voices and faces to create indistinguishable digital personas are particularly astonishing.
Companies need to invest heavily in hiring and building out a strong machine learning engineering function. While specialists are crucial for foundational technology, it's also important to foster generalists across product teams who can learn to build on this technology and design natural AI-integrated user experiences.
Avoid treating AI as a bolted-on feature or having only a separate 'AI team'; instead, aim to integrate AI learning and capabilities across all product teams. Also, be wary of over-optimism and actively seek out skeptical or negative opinions to balance perspectives, and don't be afraid to lean into the change.
The biggest lesson is to keep pricing simple. It's tempting to add complexity with numerous add-ons, tiers, and different pricing models, but this can lead to customers being unable to understand their bills and create compounded mistakes over time.
This framework helps teams evaluate their roadmap by balancing investment in features that attract new users (differentiation) against those that are merely entry requirements for the market (table stakes). Startups often need to prioritize differentiation, while established companies must ensure they maintain sufficient table stakes to remain competitive.
19 Actionable Insights
1. Dedicate Time to Explore AI
Set aside dedicated time to read about AI and actively play with AI tools like ChatGPT or Rewind. This is crucial to avoid being left behind by the rapidly transforming technological landscape.
2. Map Product to AI Capabilities
Start with your product’s core premise and the problem it solves, then ask if AI can do that, partially do it, or augment it. This helps strategically identify where AI can replace or assist existing functionalities.
3. Embrace ‘Ship to Learn’ Culture
Adopt a ‘ship fast, ship early, ship often’ mentality, understanding that mistakes will happen more often than not. This allows for rapid learning, quick changes, and continuous improvement, especially with new technologies.
4. Focus on Customer Problems
Center product development on understanding what customers are truly trying to do and the energy they have around those problems. This ensures you build solutions for important issues, not just tactical features.
5. Achieve Product Market Story Fit
Ensure your product not only fits the market but also has a clear, simple, and compelling story that explains why it’s better. A great product in a great market can still fail if the story is convoluted or missing.
6. Balance Differentiation & Table Stakes
Strategically trade off between building differentiated features that attract customers and essential ’table stakes’ features required to compete. Startups often need more differentiation, while established products balance both.
7. Integrate AI Across Teams
Avoid creating a separate ‘AI team’ and instead encourage everyone across product teams to learn about and integrate AI. This prevents bolting on AI and fosters a more pervasive understanding and application of the technology.
8. Continuously Learn About AI
Stay up-to-date with the fast-moving AI landscape by reading blogs (e.g., OpenAI’s), newsletters (e.g., Matt Rickard’s), and following relevant people on platforms like X (Twitter). Allocate specific time for this learning.
9. Keep Pricing Models Simple
Resist the temptation to add complex tiers, add-ons, and multiple pricing models. Simple pricing helps customers understand their bill and reduces confusion.
10. Navigate Ambiguity with Conviction
In rapidly changing fields like AI, be prepared for high levels of ambiguity and develop conviction based on informed judgment and experience. This helps lead teams through uncertainty without stifling alternative voices.
11. Seek Diverse AI Perspectives
Actively read and consider alternative opinions from those skeptical or critical of AI, not just optimistic views. This helps balance perspective and avoid blindly jumping on bandwagons.
12. Lean Into AI, Don’t Fear It
Overcome the fear associated with AI’s potential impact on jobs and roles by leaning into the technology. Focus on how roles will change and adapt, rather than succumbing to doomsday scenarios.
13. Use ‘Before & After’ Framework
Identify significant changes or launches as ‘before and after’ moments, then actively engage in learning and research after the ‘after’ state. This helps assess if decisions were right or wrong and what happened post-change.
14. Strategic Overcorrection (Pendulum)
Recognize when an aspect of your organization or product is in an undesirable state and be willing to ‘swing the pendulum’ to overcorrect. Learning often comes from crossing boundaries and realizing mistakes, then course-correcting.
15. Apply ‘Four Forces’ to Decisions
Utilize the ‘Four Forces’ framework (attraction of new solution, reasons not to adopt, habits, anxieties) to understand why people make decisions, especially when considering product adoption or switching. This helps address underlying motivations.
16. Only Work on What Matters
Prioritize your tasks and focus strictly on what is most important. This personal motto helps reduce distractions and ensures effort is directed towards high-impact activities.
17. Don’t Worry About Uncontrollables
Consciously reduce stress and improve focus by not worrying about things outside of your control. This helps maintain a calmer perspective and directs energy more effectively.
18. Practice Kindness in Interactions
Always strive to be nice to people, recognizing that you don’t know what personal challenges others might be facing. Kindness often goes further than people realize and avoids regrettable actions.
19. Effective Reference Call Question
When conducting reference calls for candidates, ask: ‘What feedback will I be giving this person in their first performance review?’ This question provides insightful and candid information that candidates cannot easily dodge.
6 Key Quotes
This is a, like, meteor coming towards you. This is going to radically transform society. And I think if people don't explore AI properly, it will leave them behind.
Paul Adams
You've got to map your product and what AI can do and what it will be able to do. And then ask yourself, okay, what are we going to do?
Paul Adams
The number one thing I would say is keep it simple. Keep it simple. It's so tempting to, like, with us, for example, a lot of SaaS products, you know, have add-ons where you're like, hey, you know, we built X and that's like $10 or $100,000 to spend what kind of product you're selling.
Paul Adams
To know where the boundary is, you got to cross it. And crossing, it's painful. But if you don't cross it, you'll never know.
Paul Adams
Only work on what matters most.
Paul Adams
What feedback will I be giving this person in their first performance review?
Paul Adams