The chat is the least interesting part of AI
We’ve treated AI like a chatbot. The more important version works in the background and remembers what matters.
The most useful thing my AI assistant has done in the last two months wasn’t a fintech update or a last minute Liverpool injury alert before an FPL deadline.
It told me Project Hail Mary was in cinemas that weekend and to book before it sold out. I booked and it was full. It recommended it because it already knew I was reading the book. That’s what OpenClaw does at its best. It runs in the background, picks up on what you care about, and shows up with things you didn’t think to ask for.
I’ve been running it since February. Long enough to realize I was wrong about what it was.
OpenClaw started as a side project by Peter Steinberger, got dragged through a trademark fight, renamed twice, then exploded.
I set it up on an EC2 server in February because I thought it would be an interesting chatbot. I called it Frank. I spent the next month realizing I’d misunderstood what it was.
I’m not an engineer. Which is probably why this clicked later for me than it should have. I thought AI was a better chatbot. It turned out to be something closer to infrastructure.
My relationship with APIs is the same as my relationship with plumbing. I understand how it works until it doesn’t. Every AI tool I’d used before had one thing in common. You go to them. You open a tab, type something, get an answer. Start a new chat and they forget you exist.
OpenClaw is built differently. It has full access to my computer. It reads files, writes files, runs code, calls APIs, browses the web, sends messages, takes actions. It doesn’t wait in a tab for instructions. It works in the background. Once that clicked, everything followed.
I thought I was building a chatbot. I wasn’t. The chat turned out to be the least interesting part. Two months in, I have 19 cron jobs running on Frank.
Every morning at 8am a briefing lands in my Telegram. Liverpool fixtures, injury news, MENA fintech headlines, what happened in AI overnight, and an FPL deadline reminder if one’s coming.
My previous morning routine was finding out about things I’d already missed.
The ones I didn’t expect to value are weirder. Once a week Frank reads an article and sends me the key points. Once a month it runs a roast, reads the gap between what I said mattered and what I actually did, then writes something specific enough that I want to forward it to someone. Once a month it invites five guests to a fictional dinner party and writes the first five minutes as dialogue. The guests are chosen based on what I’m wrestling with, not what’s on my task list.
Underneath the scheduled jobs there’s a second layer most people skip. Frank has a vector database with everything I’ve fed it, daily notes, knowledge files, project states. When I ask about something with history, it searches that first. A few weeks ago it connected a forgotten video transcript to a project I was working on and named a data opportunity I’d missed. That’s when it stopped feeling like storage and started feeling like memory.
Frank also has named failure modes, all written into a file called SOUL.md. Sycophantic Frank agrees with everything. Essay Frank gives me a thousand words when I wanted forty. Hedge Frank qualifies a strong opinion until there’s nothing left of it. There’s one rule across all of it. Be the assistant you’d actually want to hang out with for a beer.
The mistake people make with AI is assuming better tools mean less human work. Usually it means more of the work becomes worth doing.
That pattern is old. In 1865, William Stanley Jevons noticed that when coal became more efficient, demand didn’t fall. It exploded. Cheaper and easier didn’t reduce usage. It made more things worth doing.
I think AI works the same way. Once it stops living in a tab and starts becoming persistent, contextual, and embedded in daily work, you don’t use it less. You use it everywhere. The constraint stops being access and starts becoming imagination.
Spreadsheets didn’t eliminate accountants. They made accounting faster, which meant more analysis got done, which created more demand for people who knew what to do with it.
That’s the part most people miss. The question isn’t just whether AI replaces tasks. It’s whether you understand what this version of AI makes newly possible.
You don’t need to be an engineer to run this. But you do need to think like one. Things break. One of my cron jobs hallucinated a Liverpool vs Bayern Munich fixture for three straight days. A Google Calendar integration never properly worked. What does work is debugging with another AI in parallel. Paste the error into Claude, ask what it means, try the fix, repeat.
Week one is frustration. By month one you’re experimenting. By the end of month two you know what you actually want this to be. I set up an AI agent called Frank. It roasts me, sends me a morning briefing, and acts like a second brain.
I’m bad at remembering birthdays, and Frank reminds me. That tiny use case tells you more about the future of AI than most product demos do.
The setup will be one click eventually. The people doing it the hard way now are learning what they actually want it to do. That part no product can automate for you.
The useful version of AI won’t feel like software you visit. It’ll feel more like a collaborator that remembers, notices, and occasionally interrupts you for a good reason.
See you out there.
Martin
P.S. My writing soundtrack Madonna Into The Groove (Live Aid 1985)







should I ask Claude or Opus how to create my Frank (named Seth) on my Mac mini?