coherence beats – As AI tools let small teams ship software in weeks, the advantage shifts from speed to coherence: how an organization’s product, data, decisions, and customer experience move in the same direction. In a world of near-identical apps, the deciding factor becomes
Building software has become dramatically cheaper—fast enough that the question is no longer how quickly companies can code. It’s whether what they build is worth using.
With “the right AI tooling. ” a team of three can ship in weeks what used to take a team of 30 a year to deliver. and the tools are described as getting better every quarter. That speed is spreading across industries—from retail to healthcare to financial services—while competitors churn out new versions at a pace that feels almost instant.
But the flood has a consequence. The gap between the best digital products and the worst is widening. because the barriers to building have collapsed and the standard for what people will actually use—and stay loyal to—has risen. When there are “20. 000 versions of the same product. ” with features comparable. interfaces similar. and subscription tiers nearly identical. switching stops being a major effort. A user can try one offer on Monday and move to another by Friday.
What prevents that churn, in the view presented here, is coherence.
Coherence is the degree to which every part of an organization—its product. its people. its positioning. and its decisions—moves in the same direction at the same time. It works “above design and above brand,” even though it touches both. For teams accelerating right now. coherence becomes the separating factor. because building coherent work is easier in smaller groups: “Teams of three supersede teams of 3. 000. because coherence is easier with a smaller group.”.
That’s where the framing around AI in product development often goes off track. Much of the conversation focuses on speed—“ship faster. ” “automate more. ” and “reduce headcount.” At a major AI conference recently. the pitch at nearly every booth followed a similar script: the workforce is the problem. and AI is the fix. The argument here pushes back against that premise.
Building, it says, has always been the easy part. The hard part has been understanding what to build, for whom, and why they should care. And as anyone can produce an app in “20 minutes. ” the central question shifts to whether that thing deserves to exist at all—whether it has genuine domain intelligence. understands the people it serves deeply enough to earn loyalty. and can deliver an experience people will actually want to use.
The same logic applies to intelligence in general. AI is described as extraordinary at processing, pattern-finding, and generating at remarkable speed. But insight—reading unarticulated needs, and instinct built through years of doing the work—remains human. The organizations moving fastest, the piece argues, use AI to amplify both human insight and customer understanding. That amplification compounds when it serves both the people building the product and the customers using it.
This is also where the “agents” come in.
Agents are becoming a new kind of digital intermediary. increasingly used by customers to decide what to trust and which products to recommend. The mechanism described is stark: an agent reads a product at a structural level—whether data is traversable. whether systems are consistent. and whether what the product promises is actually what its backend delivers. In milliseconds, it decides whether to recommend the brand or route a customer to a competitor.
For that process to go well, coherence has to exist beyond what users see. A brand with coherent product data and a coherent relationship between its promises and its delivery is rewarded by both agents and humans. But a brand with a polished surface and a fragmented backend “will lose. ” and coherence cannot be faked for an agent. The agent will audit the brand’s digital presence in the time it takes a human to load a homepage.
To make the concept practical, the piece offers three questions aimed at measuring coherence directly.
First: if you asked 10 people across an organization what the single most important product priority is this year, would all 10 give the same answer?
Second: when customers interact with a product across three different touchpoints, does it feel like one company built all three?
Third: if an AI agent audited the entire digital presence tonight, what would it find?
It then adds another way to test whether strategy and execution are aligned: look at the last five major product decisions. trace each one back to the stated strategy. and check whether more than two of them fail to connect to a clear strategic rationale. The warning is not about whether the features ship or whether the product functions. The danger is that the experience will feel like it was made by “five different companies,” because functionally it was.
The tools may keep getting faster. The advantage, the argument concludes, belongs to organizations that know what to build and why it matters—because in a market where anyone can ship, coherence is what determines who keeps customers, and who gets selected when an agent makes the call.
Peter Smart is CXO and managing partner of Fantasy.
AI software development product development coherence product strategy AI agents customer loyalty digital experience retail healthcare financial services Fantasy Peter Smart