What is the Future of the PM Role?
Claire Vo of LaunchDarkly recently predicted that the role of the Product Manager is effectively over as the increasing sophistication of AI LLMs and tools becomes capable of solving PM problems. (You can watch her talk here.) Suffice it to say, it didn't go over too well with the Product Management Community. No one wants to hear their job may soon be replaced by an AI agent. But the truth is, when we look out 3-5 years at how Product Teams (and digital development teams in general) are structured, we know that those teams will not resemble what we have today, and that AI will play a very prominent role. So what will it look like?
AI & Product Management
Before we jump into what AI & Product will look like in 3-5 years, let's start with what it should look like today, in 2025.
Today, many AI products have been introduced to automate and/or streamline Product Management processes. Because a new tool pops up seemingly every day, it can be confusing and time consuming for a PM to determine which are best, how to configure, etc. They need help and support.
​
A product organization must therefore focus on creating processes that make AI adoption easy and convenient for the team. The best way to do this is to croudsource solutioning. A slack channel dedicated to identifying great tools is a great method, along with a general cultural embrace of using and sharing those tools. Product Leaders must evangelize the power of AI tools and provide their team with the bandwidth to scale solutions to the entire team.
Fundamentally, an Empowered PM should not have to write a requirements document or even jira tickets any longer; with the PMs supervision, those can be managed by AI.
​
Beyond task management, there is much that can be accomplished with LLMs today. Any database of knowledge or information can be easily managed by an AI agent today. So for example, if I have a spreadsheet of customer responses to a product and stakeholders want to assess those responses, an AI agent configured through Slack could manage that work, synthesizing data for anyone interested.
​​
The flip side of this is that as tools replace Product Management tasks, they PM must still be grounded to where the rubber hits the road. If a tool exists that helps distill consumer feedback for stakeholders, the PM must still be acutely aware of what customers think and how they feel.
​
Fundamentally, a strong Product Organization should be investing time and resources into available Product AI tools today, scaling those tools to the team and remaining up to date on the latest and greatest time and resource savers.
​
AI & Product Tomorrow​
AI & Product of tomorrow (3-5 years) will be vastly more sophisticated. A Product/Design/Engineering team will consist of both real life roles and AI agents. The team will regularly interact with AI agents as part of their workflow, providing information, feedback, refinement, and so forth, just as they would with a real person. This setup will require empowered, generalist Product Managers with experience and savvy across the entire business.
​
"I’m especially excited by the combination of someone with very strong judgement (product sense) and generative AI tools. But I’m also worried about the prospect of providing those same tools to people that do not have the necessary product foundation." - Marty Cagan
​
Cagan's Product Model of an empowered PM owner whose primary responsibility is Product Discovery - understanding the consumer, their problems, and working with technology and design to create solutions to those problems - aligns extremely well to tomorrow's model.
​
Executional Product Managers will indeed become obsolete without the additional skills of creativity, collaboration, and a deep understanding of the consumer and the nuances of their problems. Product Managers tasked with KPIs and help accountable to them, rather than tasked with a list of prescribed features, will be very well configured to manage the AI-aided product development structure of the near future. What, though, will that look like?
​
Understanding LLM AI Agents
Large Language Models of today like ChatGPT or Perplexity are generalists - they know just about everything. As such they differ from a human being in that they lack a point of view. They tend to be sycophants, extremely agreeable, with no definitive perspective on a given topic, unless you explicitly instruct it to have one. In building AI agents with proficiency in a specific area - design, product, UX, coding, etc. - it is necessary to train them with material and prompting that create their niche skillset, just as would a real employee.
​​
By replicating instances of an LLM training each with a unique criteria of knowledge and information, a perspective is created. One could make a Designer AI agent, an Engineering AI agent and a Product Manager AI LLM fairly easily (this is happening now) that is capable of completing all tasks related to development in a fraction of the time it would take a human being to do the same at a fraction of the cost.
"Wait a minute! What about managing stakeholders? What about understanding the customer?" I think we will be surprised by how comprehensively AI will be able to handle all Product Management and Product Leader tasks. With the normalization of remote work tools like Slack, one could interface with an AI PM Agent and as long as that agents responses and performance were as good or better than that of a human being, much like automated drivers it will eventually be adopted.
​
Until then, there will be a very strong need for empowered Product Teams to manage the development flow - likely fewer but highly competent, generalist Product folks that understand the totality of the business as well as the wholistic view of the Product line and needs of the customer.
​
​
​
​