Product and engineering have always had tension. Product wants to ship fast and learn; engineering wants to build right and avoid tech debt. Product pushes scope; engineering pushes back. That tension is productive when both sides are optimizing for the same thing: building something that works and matters.

When agents change what gets built and how fast, that tension shifts. The old tradeoffs—speed vs. quality, scope vs. simplicity—don’t map the same way. And desire for the new way of working isn’t uniform. Product might see upside (faster prototypes, more experiments); engineering might see threat (my code is now generated). Or vice versa. Misaligned desire between functions creates friction that no tool can fix.

Where the Tension Lives Now

Ownership of “what gets built.”
When agents generate specs, user stories, and code, the line between product and engineering blurs. Who owns the output? Who’s accountable when it’s wrong? Desire for agentic workflows can diverge when ownership is unclear. Product may want to “let the agent run”; engineering may want to review everything. Neither is wrong—but unspoken disagreement kills adoption.

Velocity expectations.
If agents make implementation faster, does product get to ask for more? Or does engineering get to invest in quality and architecture? Desire depends on shared expectations. When product assumes “we can do 2x the roadmap” and engineering assumes “we can do the same roadmap with half the stress,” both will be frustrated. Aligning on what “faster” means—and what we’re optimizing for—is essential.

The bridge from prototype to production.
Agents excel at prototypes. Production—reliability, scale, maintainability—is harder. Product may desire rapid iteration; engineering may desire sustainable systems. The tension isn’t new, but agents amplify it. Desire for agentic workflows needs to include a shared view of that bridge: how we move from “agent-built prototype” to “production-ready system.”

Aligning Desire Across Functions

1. Name the tension.
Don’t pretend it doesn’t exist. “Product and engineering have different incentives around agent adoption. Let’s talk about it.” Making it explicit creates space to align.

2. Define shared success.
What does “winning” look like when agents are in the mix? Faster learning? Higher quality? Better developer experience? If both sides agree on the North Star, desire can converge around the path to get there.

3. Clarify ownership.
Who decides what the agent builds? Who validates output? Who owns production readiness? Clear ownership reduces the fear that drives resistance and the confusion that dilutes desire.

4. Invest in the bridge.
If the gap between prototype and production is the friction point, invest in closing it. Tooling, process, roles. Desire grows when both functions see a plausible path from “agent-augmented” to “shipped and sustainable.”

Desire in ADKAR isn’t just individual. It’s organizational. Product and tech need to want the same future—or at least understand and respect each other’s stakes. That alignment is the foundation for Knowledge and Ability to follow.