For most of the last two decades, the answer to “what’s slowing us down?” was almost always the same: development. Getting engineers to build the thing was the long pole in the tent. Product managers wrote requirements, designers handed off mocks, and then everyone waited — sometimes for months — while the engineering team worked through the backlog.
That assumption is being turned upside down.
AI-assisted development tools are compressing engineering timelines at a pace most organizations haven’t fully reckoned with. A feature that once took a team of engineers six weeks to build can now be prototyped in days. Full products are being scaffolded in hours. The code is no longer the hard part.
The bottleneck has moved — and most organizations aren’t ready for it.
The new long poles in the tent are the teams on either side of engineering: the people who define what gets built and the people who bring it to market once it is. Requirements definition, UX design, legal review, compliance sign-off, marketing readiness, sales enablement — these functions have not yet seen the scale of change that development has seen, and they will not be able to keep up.
What’s more, this isn’t a temporary adjustment. The gap between what AI-enabled development teams can deliver and what surrounding functions can absorb is going to keep widening. As AI tooling matures, the speed advantage on the engineering side compounds. Meanwhile, discovery, design, and launch processes often remain largely manual, sequential, and slow by comparison. The snowball is already rolling.
What this means for delivery leaders
This shift has real implications for how PMOs and delivery organizations are structured. For most of my career, the job of a program manager was to protect engineering capacity — to clear blockers, manage dependencies, and make sure developers could stay in flow. That work isn’t going away, but it’s no longer the most critical leverage point.
The new priority is accelerating the functions that feed and follow engineering. That means rethinking how requirements are gathered and validated. It means designing for faster iteration rather than perfect specification upfront. It means building go-to-market motions that can launch in weeks, not quarters. And it means building PMO frameworks that treat discovery and launch as first-class delivery workstreams — not afterthoughts.
Organizations that recognize this shift and adapt their delivery models will be able to move at the speed AI makes possible. Those that don’t will find themselves with faster development teams and slower everything else — and a growing backlog of finished code waiting on the humans around it.
The bottleneck has moved. The question is whether your delivery organization has moved with it.