People. Process. Product.

All of the recent changes to Claude, OpenClaw, recent "wiggly robot behavior", and having y'all build with the robot — I need to form a better strategy for Slice. My laissez faire approach appears to be closing. For us to be successful—everyone here needs to play with the robot. And we need to play in the real world were there are consequences. Kids playing in a sandbox is adorable. Adults playing in reality is profitable.

That's the only way to learn and grow. Teams that are empowered win. Everyone here should have the ability to break things. My job is that you can only break a glass and not the house, front yard, street, city, earth, humanity.

There is not an obvious strategy for me to attain this goal while maintaining your agency. I cannot, will not, be a gatekeeper. The future is embracing the entropic nature of modern reasoning.

Staging Env

The first step is creating a staging environment for our software for unencumbered play. We'll have a staging URL you can view your work!

Promotion Flow

There will be a channel hosted by an agent whose job it is to deploy work done on staging if it passes currently undefined criteria! Ideally it coaches you, informs me, and ~~prevents~~ reduces the frequency of negative outcomes.

Theory of the Agentic Firm

There is an opportunity to reinvent the theory of the firm using AI. My thesis is the robot can uncouple coordination cost from organizational reach.

Software as the Expression of the Firm

"Software" (whatever that is anymore) is the cumulative expression of our work together. We are all going to create production software which means we are all responsible for the software we put out. Think of it as an expression of our work together. That means everyone here is a software developer whether you like it or not. To realize the promise of an agentic organization, everyone makes things. If you don't — the theory will decay and our outcome will not be as profitable.

This unique paradigm should inspire those that work with us.

This won't happen all at once.

There will be mistakes.

Ask for forgiveness not permission.

Be sure to tip your waitress.

Lots of software has gotten very cheap. A lot of it, actually. What used to cost real money to build can now be spun up by people who've never written a line of code. That's genuinely exciting!

There are more non-technical people contributing to the web right now than at any point in history. That's cool! And scary.

…But building something is only half the battle…

"I built a full app in 20 minutes." Cool. Who's fixing it in six months when the API changes? Who's patching it when a customer finds an edge case? Who's iterating on it week after week to actually solve the business problem it was supposed to solve?

Nobody's posting that video because it's boring!

Running business-critical software is a marathon, not a sprint. It's diligence. It's the slow, unglamorous work of maintaining, testing, and adapting systems over time to meet real customer needs. You can't sprint forever.

And here's the thing: maintenance is the opportunity.

Building got democratized. Great. That means the gap isn't in who can create anymore — it's in who can sustain. The people and teams who figure out the long game, who treat software like something you tend rather than something you ship and forget, are the ones who'll own the next decade.

I've been building complex stuff with AI lately. Things that would have taken weeks are taking minutes. You look at what you just made and think, yeah, that's pretty good. Look at all that!

It's a trap.

You're moving fast. You're riding this wave of productivity that feels incredible. And it is incredible. But there's this pervasive feeling of good enough… Look at how much we did! Look at how fast we did it!

But good enough is only good enough.

I've said it for years: the last 20% is the last 80%. That was true before AI. The first 80% arrives so fast that you haven't even warmed up yet to your own idea. Your brain is already moving on to the next thing. But the details, edge cases, the copy, the feel of the thing — that still takes time. Maybe two or three times longer than your first shot.

Speed is a gift, but be sure it doesn't make you lazy.