Product conductors
The AI-driven unification of product management, design, and engineering
For at least the past two decades, the product manager has functioned both as translator of customer problems into technical requirements, and as a nexus between different functional areas of a business. Typically the job covers a list of responsibilities:
Working with customers to understand their workflows, pains, and problems
Writing detailed product requirements docs (PRDs)
Coordinating activities between departments (engineering, design, marketing, sales)
Running meetings, keeping everyone in sync on status updates
Maintaining timelines and roadmaps
Documenting specs and user stories
Shuffling backlogs and managing Jira tickets
Gathering feedback on the product and incorporating back into the cycle
The PM is a company's coordinator, scheduler, planner, and go-between, with much of their day devoted to glorified secretarial work. As Claire Vo says: "dates and docs".
And that's not to diminish the PM's contribution — someone has to manage the coordination costs and keep the train on the tracks. The product manager spends much of their time as a translation layer: between customers and engineers, between sales and engineering, between marketing and design.
But despite being bogged down by cross-team coordination, the PM's critical responsibilities often get overshadowed: rich customer engagement, hands-on product exploration, prototyping, and deep knowledge through constant use of the product. These high-impact activities — where PMs truly add strategic value — are tragically underutilized, yet they represent the core of what makes a great product leader.
With AI catalyzing unprecedented change, I'm interested in how PMs might combine their unique customer demand perspectives with an unlocked ability to go deeper on solutions singlehandedly, trying out new ideas as fast as they think them up.
The explosion of productized AI brings a raft of new tools that'll evolve how we build software. From Cursor to v0 to Bolt to ChatPRD, LLMs are making their way to the front lines of building things — they're no longer just for writing, searching, and summarizing.
The AIs are getting extremely good at writing code and moving pixels. The interfaces to work with them — to channel the underlying models like Claude or o1 toward useful ends — are advancing as fast as we can figure out how to use them.
I think what we'll see over the coming months and years is a gradual collapse of several roles into one. Or certainly a trend from more to fewer. Product managers, designers, engineers, QA testers, market research analysts — the nuts-and-bolts groundwork of these jobs will become the territory of the AIs. The place for humans is the connectivity between them. The AIs have the skills to execute; what they lack are the goals, the context, and the "soft" knowledge of customer relationships.
We won't be fragmented product managers working in our silo of larger process. We'll become product conductors of an AI orchestra.
Some claim that with this onrush of AI, product management is a disappearing role. Others will say it's the engineers becoming replaceable. I think both perspectives are getting at the same directional theme: that teams will flatten and shrink as AIs take over much of the legwork.
From the PM side, I have two broad reactions:
If a PM continues business-as-usual with the way the role is done today — yes, that job is toast
If a PM embraces what's possible now with AI for idea development, research, validation, building, iterating, prototyping, testing, and more, the PM has the potential to be a whole team
And I think you could swap "engineer" for "PM" and say the same thing. What's required with the new wave is a willingness to experiment into neighboring areas. To spread yourself into a wider set of skills.
I see the PM in a unique position compared with other roles. Great PMs are the keymasters of product businesses. They have the customer relationships, the problem knowledge, the solution landscape intel, and enough technical aptitude to be most essential in the AI era.
I say that PMs are the "glue" of the organization (admittedly, with bias). They’re a binding agent that holds together tech, marketing, sales, and customers. They hold the most context about the business as an entity — who it sells to, what it builds, why it builds it, how it works — no one else except a founder/CEO has the same level of context. And context across a breadth of an organization is the scarcest resource of all.
With AIs at their disposal, the PM — keeper of the context — brings exactly what the AI doesn't have. We even use the phrase "context window" to describe the blobs of information we send to AIs to inference against. Without context, they don't do much at all. To do their best work for us, AIs need a wide yet targeted frame to work with.
If you want an AI to help you build software, you need to bring to the table what it should do, who for, and why. Bring interesting context and direction, get good results.
This situation will hold true for so many domains, not just software companies. The keeper of the context will be the one with the most power, at least if they choose to wield it.
As with prior technological paradigm shifts, there's a tendency for human expertise to be abstracted, where we’re manipulating proxies for some more complex system, process, or programming language. From humans working in factories to humans designing factories for robots. To a large extent, that's what's happening here. We go from writing Assembly code to writing C++ that compiles to Assembly to scripting languages that compile to C++, and up and up through layers of abstraction. AI is pulling us up (in some cases way up) the abstraction ladder.
It's inevitable that the next Big Ideas will be enormously levered on AIs. 3-person teams will build billion-dollar businesses. But this achievement relies on the leverage of a network of virtual minds deftly orchestrated by a conductor above. A strategic division of labor that recognizes the unique value of human context and conviction, combined with the bottomless execution resource of the AIs.
AIs writing PRDs. AIs helping you write prompts for writing PRDs. AIs building your ideas. AIs testing others' code. AIs coordinating with one another.
Will the skilled PM learn to combine the messiness of customers, demand signals, and trade-off management with the leverage they get with AI for building? Will it be the engineers or the PMs replaced first? How it works itself out doesn’t matter to me. What matters is that we all realize this collapse is on the horizon, and that we learn to adapt to the new environment.
Nice: "...context across a breadth of an organization is the scarcest resource of all."