Product strategy, rather than technology alone, is what connects AI work to business and customer value.
2
Teams should spend time understanding the problem, its context, and the problem underneath it before building a solution.
3
AI products need measurable business outcomes, customer feedback, and a product person who can connect data teams with the wider business.
Summary
Stuart Winter-Tear argues that AI projects often fail to produce value because teams start with what they can build instead of the problem they need to solve. He uses the product quartet to test an idea across desirability, viability, feasibility, and datability. This asks whether customers need it, whether it fits the business, whether the team can build and support it, and whether suitable data is available. He also discusses the uncertainty of generative AI, the need for feedback and transparency, and the limits of large project plans in an area where the next step depends on incomplete information. Product strategy should connect to business strategy, define measurable outcomes, and produce smaller wins before larger investments. Winter-Tear sees product teams as translators between data and business, with technology coming later in the process.
AI work needs a business and customer reason before a technical solution
Winter-Tear says the ability to research and build something does not mean that customers or the business will adopt it. He attributes poor returns on data and AI projects partly to data teams being isolated from the business and the market. His central claim is that product strategy is more important than technology when trying to win with AI. Teams should focus on the problem, the business outcome, and the customer value before deciding what to build.
The product quartet tests whether an AI idea is worth pursuing
The product quartet adds datability to the familiar product questions. Desirability asks whether the idea solves a customer problem and creates market demand. Viability asks whether it fits the wider business strategy, has measurable return, and can create a defensible advantage through areas such as pricing, packaging, or branding. Feasibility covers whether the team can build and support it. Datability asks whether the required data exists and is good enough.
Teams should use frontline work to find safe ways to augment people
Winter-Tear says generative AI creates opportunities in both existing workflows and in the weaknesses of the technology itself, including security, privacy, governance, and hallucinations. He supports surfacing how frontline workers already use these tools instead of simply penalizing them. Their workflows can reveal opportunities for discovery and augmentation. He would rather make employees much more productive with generative AI than use it only to remove a smaller share of the workforce.
Innovation needs small next steps because the information is incomplete
Winter-Tear rejects the idea that innovation follows a straight line from an initial idea to a finished product. Teams often have to choose the next right thing while dealing with ambiguity and unknowns. Large project plans and grand product road maps do not fit that setting. He still says teams need a clear vision of where they are going, while accepting that the route will change as they learn.
Understanding the problem means examining the problem underneath it
Winter-Tear warns against building solutions in search of problems. He describes the problem space as several layers of investigation: what the problem is, its context, and the problem under the stated problem. The real issue may differ from the first request. Teams need to spend time understanding that situation before they start building or turning an idea into a solution.
Generative AI needs product controls around uncertainty and trust
Generative AI brings unexpected behaviour and operates in a world of probabilities rather than deterministic rules. Winter-Tear says this can be useful, but it needs care in critical environments. He also calls for explainability in the architecture and, where possible, for customers. Feedback loops let customers report when a model is wrong, while education and transparency clarify what a model can and cannot do.
Product strategy has to connect measurable value to business strategy
Winter-Tear says teams need a clear vision for delivering measurable business and customer value. Mature KPIs help prove return and gain support. Product strategy should fit the wider business strategy and goals, because prototypes disconnected from those goals do not get buy-in. He recommends monetizing incrementally, showing quick value in the near term, and using that evidence before pursuing larger plans.
Product people can translate between data and the wider business
Winter-Tear describes a product role that connects data with the wider business, drawing on the idea of a human API from MIT and McKinsey. This person needs to understand and speak both business and data language. He presents product as well placed to take on that responsibility, provided it can translate technical work into business outcomes. In his framing, technology is the last mile and product strategy is the first.
