When you start building products with LLMs, you quickly hit the question:

If you can build anything you want, what are the limiting factors?

So far, AI-assisted development cuts cycle time but barely touches lead time. [1]

Cycle time = the time it takes to finish once you start on it

Lead time = the time it takes to finish once you see the need for it

Verification: Did it build what you wanted? Did it handle edge cases? Did it hallucinate? Does it meet your quality bar?

Integration and regression: How does it interact with the rest of the system? How fast can you get the reviews you need? Did it break anything? Does it meet cost, performance, scalability, security, usability, reliability, maintainability constraints?

**Context: **If you don’t tell it, the model guesses. When this keeps happening:

AI: “I solved it!”

You: “Not quite.”

assume missing context first. How do you route the right data, mcps, tools, rag environment without drowning it in a mess of logs?

Feedback loops: Agents only make progress when they can check their work. Create tests, tools, validation scripts, evals so they can run safely for longer.

**Essential complexity: **Some difficulty comes from the problem itself. AI can cut accidental complexity, but the essential kind stays. [2]

**Infrastructure and secret management: **You can vibe thousands of lines of code in minutes, but wiring stable environments, access controls, and secrets without spraying keys everywhere still takes deliberate work.

Even with these bottlenecks, cycle time is way faster than it used to be. Lead time has barely moved.

Choosing the right bet: This has always been the hard part. Of everything you could build, what deserves focus now? What opportunity matters most? What does success look like under your principles and constraints?

Shiny objects: When everyone can ship shallow ideas, depth compounds. Jumping between ideas resets most items on this list.

Buy-in and collaboration: You still need political and cultural oxygen. Teammates, users, and community need to believe the bet might work and be worth supporting. [3]

**Creativity: **Models help with breadth, but still struggle to 1) choose the right solution space, and 2) transfer insight between domains.

Organizational knowledge sharing: Each person runs their own model with its own context. A shared AGENTS file or RAG helps, but the systems rarely learn together or reflect the company’s principles.

This is still all internal. When real customers show up, the bottlenecks tighten.

**Attention & distribution: **“If you build it, they will come” still fails. Does your product amplify your distribution and vice versa? [4]

Qualitative performance: Talk to users, watch behavior, run interviews and surveys. Does the product actually solve their problem?

Quantitative performance: As experiments stack up, each A/B test takes longer to reach significance. At scale, statistical power becomes its own bottleneck.

Not all of these have to be bottlenecks. Better context and stronger feedback loops unlock real gains. But constraints like prioritization, verification, attention, and statistical limits still dominate. None of this is new. AI just pulled the easy work forward exposing the real constraints. Ignore them and you get busywork at scale. Design for them and you get a faster, tighter system.

[1] I’m borrowing these terms from lean; purists define them slightly differently.

[2] Fred Brooks wrote about essential complexity vs accidental complexity a while back.

[3] Watch out for idealistic pessimists

[4]

’s Four Fits Growth Framework comes in handy

No posts