Andrew Miracle

I think I have a cold start problem.

I asked the agent to do something small.

I used to lead teams, where I’d role play CTO or Engineering Manager. or something like that I would have a vision, I write it down in the most elegant manner using technical speak, flowcharts, diagrams, OpenAPI specs, C4, Gherkins or something like that…

People ask me how I get agents to ship real work.

I think I finally figured out how to describe the AI analogy.

I think AI stole my notepad.

For years, software teams measured work with the same buried assumption: more scope means more time.

For the longest time, I thought the challenge of building a company brain was a knowledge problem.

I keep coming back to one question in this AI era: where is the leverage actually showing up?

AI exposes your bullshit.

I have a new product manager.

Being a flower in a desert costs something.

Someone joins your team.

You ask AI to pressure test your idea.

The most valuable thing you produce with AI is not the output.

I’ve been thinking a lot about leverage lately.

The recent phenomenon I’ve noticed, and honestly it’s funny when you think about it, is what working with AI agents now looks like inside collaborative environments, workspaces, organizations.

In 1835, around 75% of cotton mills in Britain were steam powered.

I have gotten used to reviewing 1,000 plus lines of code.

Every personal website is a time capsule.