THE INTELLIGENCE IS MOVING INSIDE THE BUILDING

Last week my wife asked me a question while we were in the pool. She sees the AI clips everywhere now, and she asked, “What is all of this actually going to boil down to?”

I got my answer over the weekend, from a direction I didn’t expect.

A Chinese lab released a model called Kimi K3. I heard it broken down on the Moonshots podcast: the largest open AI model ever released, 2.8 trillion parameters, suddenly sitting at or near the top of a stack of leaderboards. But here’s the only detail I’d ask you to remember. The full model weights are set to release around July 27. That means anyone, anywhere, can download this thing and run it on their own hardware.

Most of the coverage is about the race. Who’s ahead, who’s behind, what it does to the labs. I think that’s the wrong signal.

Here’s the signal: for the first time, frontier-class intelligence is becoming something a business can own. Not rent. Own. Inside its own walls.

Think about what happens today when your team uses AI in the cloud. Two meters are running. The first is the bill: subscriptions, API calls, per-seat pricing. The second one is quieter: your company’s information is leaving the building to be processed. Your documents, your numbers, your customer context, flowing upstream to someone else’s computers. Nobody’s doing anything malicious with it. But the intelligence of your company is leaving the premises, every day, as the price of getting work done.

Run the model on your own machines and both meters stop.

Now, honest caveats, because I don’t sell fairy tales. This is not a laptop download. The panel discussing it estimated you’d need hardware in the range of a couple of maxed-out Mac Studios, around two terabytes of memory, to serve a model this size. And the bigger the company, the harder the move gets: compliance, integration, the whole apparatus.

But flip that around. A 25-person company can support that hardware budget without blinking. For once, the advantage of a technology shift lands on the small business first, not the enterprise.

And here’s the part I care about most, because it’s the part I live. It doesn’t actually matter whether this specific model wins. Costs are already dropping, and the next open release is always coming. What matters is that the two excuses I hear most from business owners, “we can’t afford to run AI seriously” and “our data can’t leave the building,” both just got an expiration date.

Which means what’s left is the real work, and the real work was never the model. It’s the working environment. Somebody in your business has to be able to take the intelligence, rented or owned, and wire it into the actual operation: the invoices, the follow-ups, the marketing, the reporting, all running from one window with a person steering. I do this every day. I run four businesses by myself, from one window, and the model I use matters less than you’d think. The environment and the operator matter more than anyone tells you.

The people who learn to do this first inside their companies are going to set the standard everyone else has to work to. That’s not a prediction about job losses. It’s a prediction about job descriptions.

The intelligence is moving inside the building. The question is whether anyone in your building will know how to run it.

If you’re wondering what that looks like in your business, let’s talk. 30 minutes, no agenda: https://kerzie.ai/schedule

Wade Kerzie

Founder, Kerzie AI Solutions

https://kerzie.ai