The Infinite Machine Problem: When AI Can Ship Everything, How Do You Decide What's Worth Building?
For most of product development's history, the binding constraint was production. Building software was expensive, slow, and required specialized skill. The cost of production forced prioritization: you could not build everything, so you had to decide what was worth building. That constraint was uncomfortable, but it was also useful. The scarcity of engineering capacity created a natural selection pressure on ideas. Features that could not justify their production cost did not get built. Features that survived the prioritization gauntlet had at least cleared a bar, however imperfect.
Generative AI is dismantling this constraint. Code generation tools allow small teams to produce in weeks what previously required months. Content generation tools produce in hours what previously required days. Prototype generation tools create in minutes what previously required design sprints. The production bottleneck is collapsing. And as it collapses, it is exposing the problem that the production bottleneck had been, inadvertently, managing: we do not actually have a reliable way to decide what is worth building. We have opinions, stakeholder requests, competitive anxieties, and feature lists. What we have consistently lacked — and what the removal of the production constraint makes urgently necessary — is a principled, evidence-based approach to the question: will this create value for the people we are building for?
When production is cheap, evaluation becomes the scarce resource — and the behavioral outcome question becomes the new constraint.
Why Abundance Makes Judgment Harder, Not Easier
The intuitive response to AI-enabled production abundance is optimism: we can now build more things, faster, so the product will improve faster. This optimism misunderstands where product value is created. Value is not created by production — it is created by changing user behavior in ways users find beneficial. Production is the mechanism. User behavior change is the result. And the relationship between production velocity and behavioral value creation is not linear. You can build faster without producing more value if you are building the wrong things faster.
The economics of this problem are now acute. When producing a feature cost three months of engineering time, the team had three months of sunk cost forcing careful thought about whether the feature was worth building. When producing the same feature costs three days of AI-assisted development, the team has three days of sunk cost — not enough friction to force the evaluation rigor the decision deserves. Teams that adopt AI development tools without simultaneously adopting more rigorous product discovery frameworks will find that they are shipping more features, faster, with lower quality product-market fit — because the filter that production cost was providing has been removed without being replaced.
The Behavioral Outcome Framework as the New Production Constraint
The framework that replaces the production constraint is the behavioral outcome question: Who will do what differently, by how much, within what timeframe, as a result of this feature? This question, which Jeff Gothelf and Josh Seiden articulate as the core of outcome-based product thinking, has always been the right evaluation framework for product decisions. But teams that were constrained by production capacity had a simpler heuristic available — can we afford to build this? — that could short-circuit the harder question. When production is cheap, the simpler heuristic fails. The behavioral outcome question becomes mandatory.
Teams that apply this framework rigorously discover that the abundance problem AI creates is actually a prioritization clarity problem. Most feature ideas, when subjected to the behavioral outcome question, reveal vague or unmeasurable intended effects. 'Add an AI-powered search' is not a behavioral outcome. 'Increase the percentage of users who find the document they're looking for on the first try from 34% to 70%, reducing average search sessions from 3.2 to 1.4' is. The second formulation is specific enough to test, which means it is specific enough to prioritize against. In an AI-abundant product environment, the teams that win are those that have disciplined their idea generation with outcome specificity — not those that have maximized their production velocity.
In an AI-abundant world, the real scarcity is customer attention, not production capacity
Rethinking the Product Review in an AI-Enabled Team
If AI tools allow your team to generate ten times as many feature ideas, and produce those ideas ten times faster, then product reviews designed for pre-AI teams will be overwhelmed. A review designed to evaluate five features per quarter cannot meaningfully evaluate fifty. Product leaders who try to apply existing review processes to AI-generated idea volume will find that reviews become either cursory (evaluating more ideas less carefully) or bottlenecks (slowing down the AI-enabled production capacity with an evaluation process that cannot scale).
The solution is to move the evaluation earlier in the process — before production, not after it. Teams using AI development tools should apply the behavioral outcome question at idea generation time, before any production work begins. Idea generation in an AI-enabled team is not the constraint; evaluation is. Building an evaluation front-end — a consistent, lightweight process that every idea passes through before it receives any production investment — is the organizational design response to production abundance. The evaluation asks the behavioral outcome question, identifies the highest-risk assumptions embedded in the idea, and specifies the smallest experiment that would provide decision-quality evidence. Only ideas that clear this front-end receive production investment.
The New Scarcity: Customer Attention
When AI makes production infinite, the binding constraint moves from production to consumption. Humans have a finite capacity to learn, adopt, and benefit from new features. An AI-enabled team that ships fifty features per quarter may be creating more for users to ignore than for users to use. The real scarcity in an AI-abundant product world is not features — it is customer attention, cognitive capacity, and the behavioral change energy that users must expend to incorporate new product capabilities into their workflows.
Product teams that understand this shift in scarcity design differently than teams that do not. They treat each feature not just as a production decision but as an attention allocation decision: is this feature worth a portion of my users' limited cognitive budget? This question forces a different kind of prioritization than the production-cost question. The most expensive feature to ship, in an AI-abundant world, is not the one that requires the most code. It is the one that requires the most behavior change from users who are already managing the cognitive load of their existing workflows. Lean UX's emphasis on the smallest possible change that drives the desired behavioral outcome is not just an efficiency principle in this environment — it is a design principle for respecting the scarcity that actually constrains value creation.
The Bottom Line
The infinite machine problem is not a technology problem. It is a product discipline problem. AI has removed the friction that was doing some of the work of product prioritization. Teams that replace that friction with principled behavioral outcome thinking — that apply the 'who does what by how much' question rigorously at idea evaluation time — will find that AI abundance becomes a genuine advantage. Teams that do not will find that they are shipping more, learning less, and creating less value per feature than they were before AI made everything easier to build.
Related Posts from Sense & Respond Learning
Stop Building 'Zombie' Features: How to Prune Your Backlog with Outcomes
The Case Against Annual Roadmaps: Why Quarterly OKRs Serve Leaders Better
Measuring What AI Actually Changes: Behavioral Outcomes in AI-Augmented Products
Further Reading & External Resources
Who Does What By How Much? — Jeff Gothelf & Josh Seiden — The behavioral outcome framework that becomes mandatory when production cost disappears
Lean UX — Gothelf & Seiden (O'Reilly) — The foundational text on building less and learning more in product development
The Paradox of Choice — Barry Schwartz — Research on how abundance undermines value — directly applicable to feature-abundant products
Want to go deeper? This post is part of the Sense & Respond Learning resource library — practical frameworks for product managers, transformation leads and executives who want to lead with outcomes, not outputs.
Explore the full library at https://www.senseandrespond.co/blog
