Ideas
Things I’ve been thinking about.
Most of them come back to two things: Business AI Architecture and the AI-DLC, how we design systems AI can work inside, and how we build them. Some are about AI. Some are about business. Some are about engineering, innovation, and why we keep making simple things unnecessarily complicated.
And occasionally, I write about something completely unrelated. I think that’s allowed. It’s my website. 😂
Read now
- Blank Canvas ThinkingBlank Canvas Thinking. Don’t ignore experts. Ignore inherited conclusions.The thinking behind everything on this site, and how to try it yourself.
- Building a practiceCan you really build a consulting business around AI?Knowing how to use AI isn’t a consulting business. Solving a problem someone will pay to fix is.
- Building a practiceWhat if you could build once and serve multiple clients?Stop starting from zero with every client. Learn how to turn a working solution into a repeatable consulting service.
- Business AI ArchitectureMaybe your business doesn’t need digital transformation.Before the expensive new platform, ask what’s actually not working. Sometimes the best fix is deleting an approval.
- Building a practiceHow do you price software that AI helped you build?If AI cuts the build from two weeks to two days, what is the client paying for? Price the value, the risk, and the work after the code.
- Blank Canvas ThinkingWhy do we keep teaching people how to use complicated software?If people need a three-day course to find the Save button, maybe the software is the problem, not the people.
- Blank Canvas ThinkingThe problem with the Hype Cycle.A curve of expectations can’t tell you whether AI will work in your business. A small, measurable test can.
- Human-in-the-leadI love AI agents. I just don’t trust them with everything.Summarizing a refund request and issuing a $10,000 refund are different jobs. Give your AI agents boundaries to match.
- Business AI ArchitectureYour client doesn’t need an AI agent. They need a closed system so AI assistants and agents can execute end-to-end processes that they can trust.Why the best AI projects start with the system, not a model.
- Human-in-the-lead“Human-in-the-loop” isn’t the same as a human in charge.Five levels of AI autonomy, and the one question that decides whether yours is safe to ship.
A note about these articles.
These are my observations, experiments, and opinions. They’re not universal truths.
Different businesses have different constraints. And sometimes I’ll change my mind after learning something new.
That’s part of the fun.
— JP