Is the Future PM a Goal Architect?
Andrew Chen's "Goal Architect" framing assumes PMs are primarily output creators — but PRDs and specs follow from judgment, not the other way around. The hard part was never the artifact; it was deciding what matters, aligning people around it, and owning the outcome — and that's not computational.
My thoughts on Andrew Chen’s post on LinkedIn calling the future of the PM role is to be a “Goal Architect”.
What Andrew Chen calls the future PM is not really a future role. It’s the job, now. If you are a PM or manage PMs who don’t have clear product goals, I’m curious.
There are assumptions here about the current PM role/trajectory and AI capability. Let’s break them down. This builds on my earlier post on Moravec’s Paradox and the future of PM.
1. Assumption: PMs are primarily creators of outputs.
PRDs, specs, stories are visible, so it’s easy to think that’s the job. But outputs follow from judgement. The real work is tradeoffs, sequencing now/next/later/never, navigating org dynamics, aligning stakeholders, shaping positioning, supporting sales. Sure, output-only PMs are vulnerable, but they were vulnerable even before AI.
2. Assumption: Strategy and product definition are separate.
The framing seems to be: “Human PM” own strategy, “AI PM” own product definition. In reality, product definition, sequencing features for differentiation, defining tiers or PLG, deciding what not to build, etc., are strategy. These are decisions where judgement and context matter. AI can support exploration and analysis rather than replace.
3. Assumption: AI can “figure the rest out”.
This assumes creating goals and strategy is enough for AI to handle prioritization, sequencing, tradeoffs, roadmaps, and XFN coordination. Roadmaps are negotiation problems, not math problems. XFN coordination compounds with org size and dependencies. There is risk tolerance asymmetry, incentive asymmetry, power asymmetry that affect the negotiated roadmap and what gets shipped when. AI can generate options, highlight tradeoffs and simulate decisions rather than replace.
The real question isn’t whether PMs become “Goal Architects” (they already should) but which parts of the PM job are computational and which are more judgement, negotiation, influence, accountability.
AI may create outputs. But the hard part was never the artifact. But judgement, negotiation, sequencing under uncertainty, accountability, asymmetries, etc., are the inputs that shape the outputs.
Call it a PM or a “Goal Architect”, title is secondary. The job was never writing documents. It has always been: decide what matters, align people around it, own the outcome.
What do you think? Share your thoughts on this post on LinkedIn.