Who actually owns your AI?
Quick idea this week, but it matters more than it sounds.
Almost every AI tool you use — ChatGPT, Claude, Gemini — is what's called "closed." You rent it. The company owns the actual engine, runs it on their servers, sets the price, and decides what it does next. You're a tenant.
There's another kind: "open source." Think of it as a recipe the company publishes for anyone to use. Meta's Llama, Mistral, DeepSeek — models like these are free to download and run yourself. You own the engine instead of renting it.
Here's the part I'd bet on. Running your own model used to take an engineer. That's changing fast. New tools like OpenClaw and Hermes are making it realistic for a regular business to run a capable open source model itself, cheaply and privately. Two or three years from now, I think that's normal, not fringe.
Even before you run anything yourself, open source is already working in your favor. It keeps real competition alive at the engine level, and that competition is the thing on your side. It means choice, so you're not betting your whole business on one company's prices or roadmap. It means control, because open models can run privately and your customer data doesn't have to sit on someone else's servers. And it means leverage, because when real alternatives exist, everyone has to keep getting better and cheaper to keep you.
Remember a few weeks back, when the "most powerful model ever built" got pulled offline days after it launched? That's the risk of renting everything. Open source is the hedge.
The CEO of Palantir, Alex Karp, put it bluntly this week: companies should own their "means of production"… if this stuff is so valuable, why are they just charging you for tokens?
You don't have to agree with all of it, but it's worth five minutes: watch the clip.
Own what matters. Rent the rest.
Learn how to actually use AI in your business.
Two upcoming sessions — one on reviews, one on open source models.
Replying to reviews isn't politeness, it's revenue. Womply analyzed 200,000+ small businesses: respond to at least 25 percent of your reviews and you earn 35% more than average.
By the end of this session, you'll know how to turn any review into a warm, on-brand reply in under a minute. You'll leave with the guide and the prompts to handle every review, good or bad.
You'll leave with the exact setup steps and the comparison checklist we used — so you can run the same test on your own business the same day and decide for yourself whether owning beats renting.
Here are the best insights our team found from across the web this week.
There is more AI news than any operator has time for. So we read it so you do not have to.
Our take: The 18,000 waters story is a feature, not a bug, in how you should think about deploying AI. You test. You find the weird edge cases. You pull it back. Then you roll it out right. That's not failure, that's the process. Most businesses quit after the first weird output. Taco Bell is now in 890 locations.
Our take: This is the other side of the reviews conversation. (That's exactly what we're covering on July 23rd.) Getting your review responses right matters even more when the regulators are watching who's gaming the system. Authentic replies — fast, human-sounding ones — are now both a competitive advantage and a compliance signal.
We're sharing this for two reasons. If you've got a kid, an employee, or a friend early in their career wondering where they fit in an AI world, this is a real, paid on-ramp — applications for the first cohort close July 17. And it's a glimpse of the bigger shift: the opportunity in AI isn't just for engineers anymore. Curiosity and a willingness to learn are becoming the actual qualifications.
That's it for today. See you soon.
Brian, Murphy, John, Beth, and Jake — some of the humans behind AI Front Desk.