Key Moments
Why Claude can’t be your PM (yet) | Anthropic CPO Panel
Lenny's PodcastKey Moments
AI models like Claude are not yet replacing Product Managers, but their rapid evolution demands PMs be more adaptable and decisive than ever to bridge the gap between human needs and technological capabilities.
Key Insights
Product management's core function remains a bridge between human problems and technology, but the pace of technological change has accelerated from 5-10 years to every 2 months, making adaptability crucial.
A Product Manager's role is becoming more important, not less, in an AI-driven world, as they are essential for managing stakeholders, ensuring alignment, and overseeing complex projects that AI alone cannot handle.
Building AI-native software requires a shift in thinking, where the interface itself can be modified by agents, enabling new behaviors and more personalized user experiences.
The ability to rapidly test multiple product directions and parallel experiments is key, with a focus on identifying and consolidating winning ideas into a cohesive user experience, rather than creating fragmented features.
While AI can perform many tasks, PMs are still vital for synthesizing information, making decisive judgments, and guiding development in uncertain environments, acting as a 'convener' between AI and human collaborators.
The gap between AI model capabilities and actual user application is a key challenge, and product teams must build user-friendly products that democratize access to powerful AI features for a wider range of users.
AI's acceleration reshapes product management
The fundamental role of product management – bridging human problems with technological solutions – remains unchanged. However, the speed at which technology evolves has dramatically increased, compressing the typical 5-10 year cycle of change to as little as two months. This rapid iteration presents a significant challenge for human adaptation. While technology advances at breakneck speed, the core human problems and needs that product managers address have not changed as drastically. The key, therefore, lies in maintaining a focus on user needs while adopting new perspectives on how technology can solve them, continually re-evaluating and resetting assumptions about solutions based on current knowledge.
The enduring importance of the Product Manager
Despite initial fears that AI might render Product Managers (PMs) obsolete, the panel argued that the role is becoming even more critical. A personal anecdote illustrated this: a PM lead insisted on bringing in a dedicated PM for a project even when AI (Claude) was already involved in managing tasks. The PM's contribution was invaluable, ensuring comprehensive communication, stakeholder alignment, and attention to crucial details like customer success enablement and security protocols that AI alone might overlook. This highlights that even with advanced AI tools, human oversight is necessary for coordination, decision-making, and ensuring all aspects of a product are seamlessly integrated and aligned with user needs.
The 'convener' role in an AI-integrated workflow
In the rapidly evolving AI landscape, the Product Manager acts as a crucial 'convener' or synthesizer. This role involves bringing together the capabilities of AI models, like Claude, with human teams and user needs. AI can surface information and suggest connections, but it lacks the organizational influence, the ability to schedule meetings, or the proactive drive to bring disparate elements together. PMs must fill this gap, ensuring that AI's potential is translated into actionable products. This requires strong judgment, clear decision-making, and the ability to navigate uncertainty, guiding teams through the exploration of new possibilities without losing sight of the ultimate user problem to be solved.
Shifting from intricate UI design to strategic decision-making
Skills that were once highly valued in product management, such as deeply understanding user psychology for intricate UI interactions (e.g., where a finger might land on a phone), are becoming less critical. This is because rapid iteration and A/B testing of multiple versions are now feasible and cost-effective. Instead, the focus is shifting towards strategic decision-making, adaptability, and the ability to discard deeply ingrained expertise when new approaches become more relevant. PMs need to be comfortable with uncertainty and continually experiment with new paradigms, even outside their core competencies, to stay effective.
Embracing chaos and fostering adaptability
The current environment, characterized by constant change and uncertainty, can feel chaotic. However, product leaders encourage reframing this chaos into a structured yet flexible process. This involves creating psychological safety for teams to experiment, acknowledging the emotional toll of constant adaptation, and encouraging open communication about the challenges. The goal is to make navigating uncertainty feel manageable and productive, allowing teams to explore new avenues without fear of failure, thereby unlocking true innovation and productivity.
The rise of agent-native software
The next frontier in software development is 'agent-native' design, where software interfaces can be dynamically modified by AI agents. This paradigm shift moves beyond simple AI integrations or AI-powered features to a state where agents can autonomously perform complex tasks, suggest workflows, and even adapt the user interface itself. This enables novel behaviors and a more personalized, efficient user experience. Anthropic is exploring this by using Claude to observe development processes and generate user interfaces that can be further customized by internal teams or third parties, blurring the lines between AI-generated and human-created software.
Balancing innovation with user adoption
A significant challenge for companies like Anthropic is balancing the drive for cutting-edge, agent-native software with the needs of a diverse user base. While some users are eager to adopt the latest flexible interfaces, many others, including large enterprise clients, require stability and predictability. The strategy involves meeting users where they are, offering adaptable solutions that can evolve over time. This means embracing extensive experimentation, learning what works through parallel testing, and then translating those insights into stable, understandable products without overly constraining innovation or alienating existing user bases.
Consolidating parallel experiments into a cohesive product
Running numerous parallel experiments is essential for identifying winning product directions. However, integrating these successful experiments into a cohesive and intuitive product experience is complex. The challenge lies in avoiding a fragmented user interface with numerous tabs or separate apps. The approach involves building foundational infrastructure that supports shared memory and functionality across different features, ensuring that new capabilities feel like a natural extension of the core product rather than bolted-on additions. This requires strategic decisions about which features form the core user experience and how supplementary functionalities can be seamlessly integrated, making the product feel intuitive and unified.
Mentioned in This Episode
●Software & Apps
●Books
Common Questions
While AI tools like Claude can handle some tasks, they currently cannot fully replace human product managers. The role is evolving, requiring more emphasis on human judgment, adaptability, and the ability to orchestrate complex projects and stakeholder needs.
Topics
Mentioned in this video
An AI model from Anthropic, discussed as a tool for observation, UI creation, and potentially managing projects, though not yet capable of fully replacing human PM roles.
A software tool or model that the interviewer humorously questions why it wasn't used instead of a human PM for certain tasks.
A specific version of Fable, speculated by the interviewer to be a more advanced version that might have been used.
A software tool that fewer attendees have used compared to Fable, indicating a potentially newer or less adopted technology.
A tool or process used to find information within an organization, which helped uncover a relevant issue.
An older version of a model that someone might be using, leading to a misunderstanding of current AI capabilities.
More from Lenny's Podcast
View all 56 summaries
30 minStop planning for 2027: how OpenAI builds product 90 days at a time | Tara Sesha and Nan Yu (OpenAI)
22 minRoles aren't converging—they're expanding | Tamar Yehoshua (Atlassian CPO)
95 minMolly Graham: The grief, burnout, and opportunity hiding inside the AI transition
23 minMarty Cagan: Strong Opinions, loosely held
Ask anything from this episode.
Save it, chat with it, and connect it to Claude or ChatGPT. Get cited answers from the actual content — and build your own knowledge base of every podcast and video you care about.
Get Started Free