A point of view on how great software gets built in the AI era, and what that changes about product leadership. It comes from building, and we revise it as the work teaches us.
AI is not eliminating Product, Design, Engineering, or Research. It is expanding what each discipline can contribute as the cost of building software falls.
The strongest teams will not erase specialized craft. They will preserve the distinct perspectives of each discipline while making the boundaries between them more permeable.
Everyone builds, but they build different things.
The advantage is no longer shipping more software. It is learning faster through software.
For decades, software organizations optimized for specialization, scale, and coordinated handoffs. That made sense when software was expensive to build.
AI is changing the equation. The cost of prototyping, implementation, iteration, research synthesis, and validation is falling, and a small team can move from an important problem to working software far faster than before.
As execution gets cheaper, the real constraints shift toward:
The role of product leadership is to create the conditions for a multidisciplinary team to solve important customer problems.
The modern product leader is not primarily a backlog owner, a requirements writer, or a coordinator of functional handoffs. The role is becoming broader and more integrated.
Every team starts somewhere different. The goal is one stage forward.
This is not an argument that traditional product organizations are obsolete. Many companies have real constraints: existing systems, customers, compliance needs, organizational scale. The market is in transition, and the opportunity is to help teams move one meaningful stage forward from wherever they are today.
Functions are distinct. Product defines, Design designs, and Engineering implements. AI is mostly an individual productivity tool.
Teams prototype more, roles overlap selectively, and AI accelerates parts of the workflow. The broader organization stays largely unchanged.
Small multidisciplinary teams own important problems end to end. AI is part of the operating model. Leadership focuses on clarity, learning, and outcomes.
AI is leverage. What matters is where we choose to apply it.
The thesis is not only about how software gets built. It is about what is worth building. Problems that once required large organizations, enormous capital, or hundreds of people may now be approachable by small, experienced teams with conviction and judgment.
The hopeful part of this moment is not that companies can build more cheaply. It is that important problems may be more solvable than they have ever been.
Madrona exists to put this thesis into practice. Our products and client work are the evidence, and we share what we learn as we go. This is a working document; the moment it stops changing, it has stopped doing its job.