Ideas for what
comes next.
Practical thinking on AI, software and building companies ready for change.
All insights
Latest firstStop Testing AI Like Normal Software
Unit tests tell you whether the rules still hold. They cannot tell you how often an AI system makes a good decision. Production AI needs a wider testing discipline.
The Next User of Your Software Might Not Be Human
APIs, CLIs and MCP are different doors into the same product. Understanding them is the first step toward building software that agents can actually use.
SaaS Is Not Dying. The Dashboard Is.
AI agents are making the graphical interface optional. The next generation of software will compete through capabilities, permissions and outcomes—not screen time.
Code Is Getting Cheap. Bad Decisions Are Getting Expensive.
AI coding agents are reducing the cost of implementation. The real constraint is moving to product judgement, architecture and deciding what deserves to exist.
AI Transparency Is Becoming a Product Feature
As new transparency obligations begin to apply, the strongest AI products will treat disclosure, provenance and human control as part of the experience.
The AI Agent Is Not the Product
An AI agent can perform tasks autonomously. A real product must give it context, tools, boundaries and a reliable path back to human judgement.
Where AI Actually Creates Business Value
AI creates value when it improves decisions, compresses workflows and enables better customer experiences—not when it is added without a clear business outcome.
The Hidden Cost of Bad Internal Software
Slow tools do more than waste time. They shape culture, decision quality and the speed at which a company can move.
Why Most AI Pilots Never Reach Production
A convincing prototype is only the beginning. Real AI products need trusted data, clear ownership and resilient engineering.
From insight to action