Velocity and Craft at Scale.
Product, Design, Engineering. The three roles haven’t changed. The collaboration and cadence have. AI work runs in tighter loops, so the team has to flow with it.
The traditional collaboration style isn’t wrong, it’s just dated. It worked at the scale and pace for which it was built. Developed in the early 2000’s for hundreds of people, quarterly roadmaps, and when strategy and execution were cleanly separated. The goals of developing a product that is desirable, feasible, and viable are still the same today. But today’s world is very different and offers advantages for those ready to graduate to a modern method of collaboration.
Fluent Agentic AI designers unleash capabilities while the industry compresses the timeline. The question isn’t who owns what. It’s who’s leading right now, and how we stay in sync while moving fast.
The Traditional Team Structure.
Defined release windows. Sequential phases. Clear lanes between disciplines.
Three lanes, three owners. Swipe down to integrate.
Compressed timelines require disciplines to overlap and communicate organically. There’s no need for middle platforms when people can just talk to each other.
Tighter Loops Require Actual Communication, Not Hand-Off Artifacts.
Strategy emerges from the overlap. Engineering capacity shapes which problems are worth defining. Design surfaces what’s worth building from user research and customer feedback. Product manages the sprint lifecycle and documents team choices to present to the organization.
The Modern Sprint.
Tight loops. Distributed strategy. The team moves through phases together.
Every function is present in every phase. Curves peak where that function leads. Everyone stays engaged. Discovery and Learn are where all roles converge: in a fast loop, framing is a team activity. Design leads Create. Engineering carries Ship. Product loops the learnings back into the next round so the cycle never closes. The team learns together.
What Actually Changed.
The three functions share a language now, so the gap has closed. With no more translation layer, ideas and concepts flow organically, increasing experimentation.
The gap was the cost. Swipe to close it.
For most of the history of product work, Design and Engineering were separated by a translation layer. Design spoke in Figma and intent. Engineering spoke in code. Neither could fully read the other. So Product created specs, tickets, and meetings about the specs and tickets. That layer was never the goal. It was the cost of the gap.
Agentic AI collapses the gap. A designer can now produce something code-adjacent and testable, a real working prototype, and the AI translates in both directions while they work. Design moves toward code. Code moves toward design. All disciplines meet on one shared medium, with Product documenting the why and setting the release strategy.
Once you share a language you stop passing notes through a translator. You just talk. The loop between Design, Product, and Engineering stops being a handoff and becomes a partnership.
This is the unlock. Everything centers around this one fact.
The Prototype Is the Spec.
The spec is a picture. The prototype answers back.
The spec only ever existed to carry intent across that layer. Remove the layer and the document loses its value.
A testable prototype, built with craft and built fast, is the spec. Except you can run it. You can feel it. It’s the thing itself instead of a description of the thing.
The spec was just the menu. Nobody at the table wants to read about the dish when the plate is right there.
Two things have to hold, and neither is optional. Craft: the prototype is a designed artifact, not slop with a pulse. Velocity: the speed is the whole point. A prototype runs orders of magnitude faster and longer than a spec ever could.
Design Validates the Prototype, Product Manages The Sprint, Engineering Builds for Production.
When Design produces the artifact that frames the build, its center of gravity rises. Design leads the Create function, because it’s now making the thing the loop runs on. Once validated, Engineering builds.
The Product Manager’s traditional job, owning the “what” by authoring PRDs and specs, gets absorbed. Part of it moves into the design-led artifact. Part of it moves into the AI. Writing the definition down stops being where the value sits, because the definition now shows up as a working prototype instead of a document.
Read that slow. The role doesn’t vanish. Its center of value moves. From author to orchestrator. Product stops drafting the solo roadmap and starts conducting the trains, keeping everyone aligned and clearing the track so the train never stalls.
Product Manager to Project Manager. The title barely moves. The work moves a long way. The team manages the product now.
Leadership still shifts phase by phase, to whoever’s closest to the work. The why is still co-authored. Engineering still carries Ship. What changed is that Design now makes the artifact the build runs on, so its gravity is heavier across the whole cycle, even in the phases where it isn’t holding the baton. Presence stays constant. Gravity is the thing that moved.
The Takeaways.
The Roles Didn’t Change. The Clock Did.
Build for Today’s Clock.