The Spec Ceiling: Why AI Coding Speed Moves the Bottleneck to Product Discovery :: Notes from the Rabbit Hole

░░░░░░░░░░░░░░░░░░░░ 0%<br>The Spec Ceiling: Why AI Coding Speed Moves the Bottleneck to Product Discovery<br>2026-07-0711 min read (2142 words)<br>#software-factory<br>#product-discovery<br>#ai-agents<br>#spec-driven-development<br>#hermes-agent<br>Table of Contents

I wrote about the Dark Factory earlier this year, covering StrongDM&rsquo;s Level 5 autonomous software factory where three engineers ship production code with no human touching the implementation. In that article I mapped the five levels of AI coding autonomy, a framework that has become a useful shorthand for where organizations sit on the spectrum from human-driven coding to full autonomy:<br>Level 1: AI-Assisted. Human drives everything; AI is a faster keyboard<br>Level 2: AI-Generated + Human Review. AI drafts, human approves every PR<br>Level 3: AI-Generated + Automated Gates. AI writes, tests review, humans on failures<br>Level 4: Mostly Autonomous + Escalation. AI handles the full loop, humans on novel issues<br>Level 5: Full Dark Factory. AI runs end-to-end, humans define goals only<br>Most organizations today are somewhere between Level 2 and Level 3. They have some form of AI coding assist: GitHub Copilot, Cursor, Claude Code, an internal agent pipeline. And they are measuring the productivity gains in pull request velocity. Stripe&rsquo;s Minions ship 1,300 PRs per week at Level 2-3. The numbers are impressive and real.<br>But here is what those velocity metrics are hiding: the bottleneck is migrating upstream, and most teams won&rsquo;t notice until it has already constrained them.<br>The Bottleneck Migration#<br>Every level transition in this framework shifts where the constraint lives. At Level 1, the bottleneck is writing code. Getting a 2x or 3x speedup on implementation produces visible gains because that is where all the time was going.<br>At Level 2 (generated + human review), the bottleneck moves to code review. You are generating code faster than humans can validate it. Teams respond by automating review checks: linting, static analysis, test gates. That gets you to Level 3.<br>At Level 3 (automated gates), the bottleneck moves to integration and deployment. Running generated code through CI at scale becomes the constraint. Teams build automated validation pipelines, digital twin environments, probabilistic satisfaction metrics. That gets you to Level 4.<br>At Level 4, something different happens. The factory can handle implementation, testing, and deployment with minimal human intervention. The bottleneck moves upstream for the first time. It lands on the humans who write the specs that feed the factory. Your AI agents can build almost anything, but only what has a written specification that is precise enough, complete enough, and unambiguous enough for autonomous implementation.<br>This is the spec ceiling. Most organizations haven&rsquo;t hit it yet because they are still at Level 2 or Level 3. They are celebrating coding speed gains that their current spec throughput can support. But as they climb toward Level 4, which every AI-accelerated organization eventually does, the constraint will shift from &ldquo;how fast can we build&rdquo; to &ldquo;how fast can we define what to build.&rdquo;<br>The symptom is predictable: product managers rushing, specs getting thinner, misalignment between what was specified and what was needed, and the quiet realization that you built the wrong thing faster than you ever could have before.<br>As orgs climb the five levels of AI coding autonomy, the bottleneck migrates upstream. At Level 4, the constraint is no longer implementation velocity: it is spec creation.As orgs climb the five levels of AI coding autonomy, the bottleneck migrates upstream. At Level 4, the constraint is no longer implementation velocity: it is spec creation.<br>The Raw Material Already Exists#<br>The most elegant part of this problem is that the raw material for solving it is already being produced by every organization that runs stakeholder meetings.<br>You should record every product conversation by default. The technology is effectively free: Zoom transcripts, Otter, Fireflies, built-in meeting recording platforms, or a local Whisper instance if you want to keep the data on your own infrastructure. The cost of not recording is orders of magnitude higher: lost context, misremembered decisions, requirements that shift because nobody can go back to what was actually said.<br>That hour-long call with a stakeholder is not just a meeting recap. It&rsquo;s the densest signal source your organization produces about what actually needs to be built. Stated requirements, implied constraints, deferred decisions, edge cases buried in anecdotes, abstract nouns that sound shared but are not; all of it is in the transcript.<br>The problem is that human note-taking extracts only a fraction of that signal, and the rest stays in the recording, effectively lost.<br>What I Built#<br>I spent the last year...