39: Brian Donohue - How to handle the untameable beast of AI products at Intercom
Sep 30, 2023
auto_awesome
Brian Donohue, product executive at Intercom, discusses the challenges of building AI products and the unpredictability of outcomes. They explore the importance of customer service, tying business and technology together, and creating artificial micro-markets. They also delve into the uncertainty of AI displacing jobs and the future impact of AI advancements.
Building ML software presents unique challenges due to uncertainty and the potential for dramatic swings in progress.
AI-powered chatbots have transformed customer service, providing high-quality conversational experiences at scale.
Balancing goals in growth and core product teams is essential, recognizing their distinct objectives and challenges in product development.
Deep dives
The Challenge of Building ML Software
Building ML software presents unique challenges compared to traditional software development. There is always uncertainty and the question of whether the ML model will actually work or not. Time frames and feasibility are difficult to predict due to the potential for dramatic swings in progress and new breakthroughs. Despite the challenges, there is immense potential in the field, with years of runway ahead to extract value from advanced models like GPT-4.
The Impact of AI on Customer Service
Intercom, a customer service software company, has witnessed the transformative impact of AI on customer service. In the past, support was often an afterthought, but AI-powered chatbots have changed the game, providing high-quality conversational experiences at scale. With advancements in language models like GPT-4, the capabilities of chatbots continue to evolve, and the landscape of customer service software is being reshaped by AI technologies.
Balancing Goals in Growth and Core Product Teams
Balancing goals in growth and core product teams presents its own set of challenges. While growth teams focus on business metrics and driving adoption, core product teams are concerned with solving customer problems and building valuable capabilities. It is essential to define the center of gravity for each team and align their goals accordingly. While growth teams prioritize metrics like adoption, core product teams may focus on outcomes such as extended value or depth of product usage. By recognizing the distinct objectives and challenges of each team, companies can effectively navigate the changing landscape of product development.
The Challenges of Working with AI at Scale
Working with AI at scale presents unique challenges. As teams grow and more people get involved, adaptation and collaboration become more difficult. The transition from small teams to larger teams can result in a loss of agility and efficiency. Collaboration between teams, particularly with machine learning (ML) teams, becomes complex due to time pressures and the need for centralized engineering. Uncertainty and the need for continuous learning are major factors when working with ML teams. Additionally, the psychology and adaptability of customers in accepting AI technology are key considerations for product development.
The Importance of Building Better Products
Building better products involves more than just incremental improvement. The best products must have a significant advantage over existing solutions to attract customers. Achieving a 10x improvement is crucial for gaining market traction. While the ease of product creation and go-to-market strategies have improved, competition in existing markets remains intense. Therefore, having a superior product alone is not enough to succeed. The ability to access and understand customer psychology, adapt to market dynamics, and provide exceptional value are essential components of successful product development.
This episode gave me a lot of peace of mind. Intercom is dealing with the exact same problems that we all have when building around and with AI.
Estimations are impossible and things that are feeling impossible are possible the day after. Brian sat together with me to talk about it all including Intercom’s coming-of-age story and how they look at customer service through this new distorting lense of AI and Machine learning that is rocking our boat.
It’s a great learning piece on anything product and leadership.
Season 3 of the ProducTea: We spill the tea on how to Go to Market through Product-led Sales and Product-led Growth in B2B and the realities of senior leadership.
Timestamps: 06:30 What makes Intercom so special, and why is good customer service so difficult? 11:50 Tieing business and technology together, while AI is making quantum leaps from unuseable to “this is really good” 18:19 We are all working in a playground of opportunities that is very hard to get under predictable control 25:30 I have no idea how long this entire project might take. Days or months. No idea 31:00 Unpredictability of your output in product, you only know when you have it 36:00 Outcome-driven goals when outcomes are impossible to define? 44:00 How to build an innovative product inside a large organization by creating an artificial micro-market 52:30 We still don’t know exactly what AI displaces and it won’t. Support vs. Replacement and the resulting uncertainty.
