I stopped asking AI questions. I started giving it a job.

Most people use AI like a search engine. You type a question, you get an answer, you close the tab.

Tomorrow, you open a new tab. Different question, same blank slate. The AI has no idea who you are. No idea what you're building. No memory of what you asked yesterday.

That's useful. But it's also the most basic version of what's possible.

There's another way to use AI that most people haven't found yet. Once you make the shift, going back feels a lot like going back to paper maps.

Give it a job, not a question

The difference sounds small. It isn't.

A question is transactional. You ask, it answers, the conversation ends. A job is ongoing. You give it context — who you are, what you're building, how you like to work — and then you put it to work. Day after day. Project after project.

The question version of AI is a really fast librarian. The job version is something closer to an assistant.

I gave mine a name

His name is Winston.

That probably sounds strange. It did to me too. But here's what happened when I named him: I started treating the setup like I was actually bringing someone on, not just using a tool. I wrote down my business context. My preferences. The clients I was working with. How I like to communicate.

I gave him standing instructions. The AI equivalent of onboarding a new hire.

And then something clicked. Instead of starting every session with "okay, so I'm building an AI consulting business and I have this client who..." — he already knew. We could just work.

The memory piece

This is the part that changes everything.

AI assistants can hold context across sessions. Not automatically — you have to set it up. But once you do, the assistant carries forward what matters: your clients, your decisions, your ongoing projects, your preferences.

Winston knows I filed an LLC in March. He knows which client is waiting on a business plan. He knows I prefer bullet points for internal notes and narrative paragraphs for client deliverables. He knows all of this because I told him — once — and he kept it.

Compare that to a search engine. Google doesn't know anything about you that you haven't typed into the search bar in the last ten seconds. It can't. That's not what it's built for.

An AI assistant with persistent memory is a different category of tool. The longer you use it, the more useful it gets. That's not true of most software.

What it actually does

The big tasks are obvious. Research a client's industry before a call. Write a first draft of a proposal. Pull together a report.

But the thing that surprised me most is the small stuff.

The three-sentence email I kept putting off because it felt awkward to write. The formatted table I needed for a deliverable but didn't want to spend twenty minutes on. The task I needed to think through out loud with someone who could push back a little.

Those things add up. The small friction points are where most of the day goes. An assistant who handles the first draft of any of them — in a minute, without complaining — changes the math on your whole day.

What it costs

This is the part that stops people.

Twenty dollars a month.

Less than most software subscriptions you're already paying for and not thinking about. For that, you get something available at 11 PM when an idea hits. Never in a bad mood. Never too busy. Never forgets what you told it.

The business case is not complicated.

What actually changed

I used AI for a long time before any of this clicked. ChatGPT, Perplexity, all of it. Impressive, fast — but stateless. Every conversation a fresh start.

The shift happened when I stopped treating it like a tool I picked up when I needed it and started treating it like something I was actually working with. Gave it context. Gave it a name. Gave it a job.

The AI didn't change. What I was asking of it did.