Writing
Notes on building AI and SaaS products: how I work, working with founders, and lessons from my own studio products.
AI can generate an interface. Understanding your product is the harder problem.
Figma's work on a specialized AI editing model points to a better question than how fast an interface can be generated: how well does the tool understand the product it is changing?
Read
An AI agent's login is not permission to do everything
Okta's Agent Gateway points to the product question behind agent governance: not just whether an agent can reach a system, but which actions it should be allowed to take once inside.
Read
Your best AI prompt should become a workflow
Figma Weave points to a useful shift in AI tools, from one-off generations to processes a team can inspect, refine and reuse.
Read
Build the workflow, not the feature list
Why a strong first release is a complete solution to one specific problem, not a smaller version of everything you want to build.
Read
Does your product actually need a chat interface?
AI can be valuable without becoming a chatbot, and the right interface depends on the task, the information and the control users need.
Read
The demo works. What still needs to happen before launch?
A working demo proves a path is possible, but a launch needs decisions about everything that happens around that path.
Read
Do you need a prototype, an MVP or a production application?
The right deliverable depends on what you need to learn, not on which label sounds most impressive.
Read
The questions I want answered before we discuss features
Drawing on product design, software development and interviewing, these are the questions I want a product conversation to answer before it becomes a feature list.
ReadPage 1 of 4 · 32 articles