Skip to content

Blog

When engineering speeds up, the bottleneck moves to product

A workshop mapped a full SDLC against agents. Almost every step compressed, except deciding what to build next, and most organizations are not staffed for that shift.

Written by

Organization and roles

Published: 14 July 2026 Updated: 1 min read

From our experience: we ran an agentic engineering workshop with a client. Day two, we mapped their full SDLC on a whiteboard and asked the same question for every step: what would change if an agent did this?

Almost every step compressed

The answers came fast. Spec generation, an agent can do most of it. Test writing, yes. Code generation, definitely. Review, mostly. Deployment, yes. The whiteboard ended up almost entirely teal.

Almost. There was one step the room went quiet on.

“What to build next.”

I’d say that was the moment the workshop tipped. Around twenty engineers in the room, half of them senior, realizing that the bottleneck they were optimizing wasn’t actually theirs. They could compress engineering throughput substantially with the right harness. The product side, the stakeholder alignment, the strategic decisions: those didn’t compress at all.

The phrase that came up next was “we’d just sit there waiting.” Said half-joking. Not really joking.

The bottleneck moves upstream

This is the bottleneck migration no one warned anyone about. AI speeds up implementation. It also moves which step in the pipeline is the slow one. When implementation drops from weeks to hours, the back-pressure doesn’t disappear. It moves upstream. The product team becomes the bottleneck. Or the stakeholder review. Or the “do we even know what we’re building” conversation.

Organizations are staffed for slow engineers

Most engineering organizations are not staffed to absorb that. The PM-to-engineer ratio is built around the assumption that engineers are slow. When engineers stop being slow, the ratio breaks.

It’s the conversation I see senior engineering leaders postponing right now. They’re focused on the AI rollout. The hard part lands on the product side six months later, when engineers have spare capacity and the product team can’t feed them fast enough.

The bottleneck always lives one step upstream of where you just fixed it.

Does that match what you’re seeing on the product side already?

Further reading

  1. Coding agents are an organizational-design problem Cheaper local generation pushes the coding-agent constraint into team practice, the shared platform, governance, and learning. One practitioner's four-level model of that shift, held up against what DORA can and cannot support. 6 min read
  2. Your org chart is the context architecture An agent asking a question routes through the same knowledge layers a new hire would. Federated context follows the org chart, with a neutral mediator for the deadlocks. 2 min readShort post
  3. When the factory dispatches the work, the manager runs the line When the factory handles assignment and status, the engineering manager's job moves from dispatcher to line operator. Five responsibilities now define the role. 2 min readShort post

All 15 posts on Organization and roles

ClientsCase study: A large DIY retailer in Germany

Test the constraint in your system

The Factory Readiness Assessment connects this research to your value stream, your data, and the next decision.

Training on this topic: Agentic Engineering for Product Managers and Product Owners
See the assessment

Want to check one team first? The self-check: twelve yes or no questions, three minutes, instant result, no account. Take the self-check