What works reliably?
An open question
There is already enough AI content.
What are we still trying to understand?
I've spent the last two years learning AI by reading, experimenting and building real systems.
Before I start another learning series, I'd rather understand what isn't being answered by everything already available.
Six questions · About two minutes · No newsletter
The premise
Learning AI is easy.
Understanding it is harder.
There is a tutorial for almost everything
How to prompt.
How to build an agent.
How to use RAG.
How to automate a workflow.
How to call a model.
But building real systems creates a different set of questions.
What breaks?
When should AI make a decision?
When should it not?
How much context is enough?
What should an agent remember?
How do you verify something before acting on it?
When is an agent unnecessary?
Why does something work beautifully in a demo and fail in production?
How I've been learningRepeat / not linear
The question
If we had 60 minutes to talk seriously about AI, what would you want to understand?
Not what you think you should learn. What are you genuinely curious, confused or stuck about?
01 / 06