Yesterday I was talking to my buddy Micah about our product, and we had a really cool idea for a feature we wanted added. I made a note of it and had to go home. Made dinner with the kids. Did some programming, worked with a bunch of agents. Then later that evening I got an agent started on it — and the feature was done and added within probably half an hour, maybe an hour.
Today, here’s the conversation we had on Slack:
Micah: “I love how we can talk about some cool thing like this in the morning, and then it’s live (and really well done) by the end of the day.”
Me: “I was thinking the exact same thing.”
The old way
In the past, we would have waited a couple days to a week for product management to fully conceive of the feature. Think about all the things it needed, write specs and user journeys. Then make wireframe walkthroughs. We might have seen those in a week or two if this was a super high priority. But more likely this would be forgotten in a meeting, because it’s not on the main line of product dev for the release — in the grand context of everything we have to do for this product, it just doesn’t seem important enough to dedicate resources to. Too many other priorities.
But even if it weren’t forgotten, we might get that walkthrough in a month. And then discuss with everyone how it should work — endless boring meetings. We’d spend weeks on whether it should be full tool calls, or an interpretation of them. How many should there be in total? What does “live” mean versus not? How long do we store these? Blah blah blah blah. Then in six months or a year we might have seen it.
All because the act of writing code was expensive. And fixing it was even more so, and harder to do.
Forty people holding your paintbrush
Director and producer William Pritzker said this about making film: “Making a film is a little bit like you set up your easel out in the field to go paint, only you’re a hundred yards away from your easel and you’ve got 40 people holding your paint brush and you’re yelling at them with a bullhorn. More yellow in the lower — no, no, not there.”
And he was right. I’ve made film. And I’ve made software. Software is almost worse. No matter how many people it is, it’s the time. And the endless discussions. You can’t just make a decision yourself — every feature has to be economically justified. It’s insane.
The new world
Not anymore. Today, writing code is basically free. Sure, you spend a few bucks on tokens, and you have to wait a matter of minutes (maybe 60, or slightly more even!) to see it done. But that’s completely different from how I grew up in this industry, and in this world. Holy crap is it different.
This example here is not an exaggeration. I’ve started a bunch of companies, and I always could envision in my head what the product should be from the start — yet it took forever to get there. And it never came out the way I wanted.
Welcome to the new world — where you just open your agent and scream at Claude all day: “NO man! I said put the menu on the left!” But you get it. And you keep going. You respond to something in process — which is how most humans work best. Not a blank slate, but some shoulders to stand on. Something to build from.
And that’s why we built CatWrangler: to help this whole process along — so the thing you talked about in the morning is live, and really well done, by the end of the day.