You can't say you're falling behind in AI, and choose the slowest way to start
A common request I hear in AI consulting is: “Can we run the model locally?”
I understand the instinct. Privacy, control, and reliance on big technology companies are legitimate concerns.
But for most businesses, self-hosting a model before building a single useful automation is like deciding to fly yourself to Europe because you don’t trust commercial airlines.
Maybe someone gives you a private jet for free. Great.
You still need pilots, maintenance, insurance, hangar space, fuel, and a crew for when something breaks.
The jet may be free. Operating an airline is not.
That is what “running our own model” often means in practice: infrastructure, hardware, security, updates, monitoring, integrations, and people accountable for all of it.
Meanwhile, most businesses have not automated lead follow-up, customer support, reporting, or back-office work.
So why begin with the hardest version of the problem?
For $20 a month, tools like Claude Code can help you build a first working automation quickly. That is where the real discovery begins.
You learn what customers ask for, find the bottlenecks, and see where automation creates value.
🌶️ Hot take: You do not need the perfect vehicle before you take the trip.
Get to the destination first. Then decide whether you need your own jet.