I opened Claude expecting to type a question and get an answer.
That is not what happened.
The first thing it did was ask me who I was.
Not literally. There was no form to fill in. But within a few messages it became clear that it was not going to be useful until it understood some things about me. What I did. How I worked. What I was trying to build. Why.
I told it I was dyslexic. That I dictated everything. That I worked in partnerships and hated writing. That I had a rough idea of the system I wanted to build but no technical background to build it from.
It did not skip past any of that.
It asked questions. Not in a clinical way. More like a conversation where the other person is genuinely trying to understand you before they try to help you. I had not experienced that with any tool before.
I came in expecting a search engine. I stayed because it felt more like a thinking partner.
That first session went on longer than I planned. By the end of it I had not built anything. No script, no dashboard, no deliverable.
What I had was a system that now knew who it was working with.
I did not fully understand at the time how important that was. I know now. Everything I have built since has run on the context from that session. Every script, every automation, every agent. They all operate on a foundation that started with me just talking about myself for an hour.
That is what most people skip when they try AI and decide it is useless. They type a question. They get an answer. They decide it is a slightly better search engine.
They never had the therapy session.
If you are starting out: before you ask it to do anything, tell it who you are. What you do, how you think, what you are trying to build, what matters to you. Give it something to work with.
There is a section inside Claude's settings called Instructions for Claude. It is where you tell it who you are before you even start. Most people never touch it. I would start there.
The tool is only as useful as the context you put into it.