You need an accountant. So you do what you've always done: you post the role, you look for someone with the right training, and you hire a strong recent grad. Top of the class. Solid fundamentals. A credential that tells you they can do the work.
That's the ceiling of what a resume can tell you. You still have to teach them how your business works.
They know GAAP. They don't know that one client always pays 45 days late and it's fine. They don't know which vendor invoices need a second look. They don't know the shorthand your bookkeeper has used for a decade. The value they generate compounds as they learn your context — your exceptions, your customers, your quirks. Nobody thinks this is exotic. It's just how hiring works.
This is also how you now hire AI.
The open-weight models available to a small business today are that recent grad. Credentialed. Capable. Generic. A model pre-trained on accounting knows the field the same way your new hire knows the field — thoroughly, in the abstract, and not at all about you. What it can do out of the box is real. It is also bounded, in exactly the way you'd expect a smart stranger's work to be bounded.
The engagement that follows isn't deployment. It's onboarding. You teach the model your process, your exceptions, your language — and the people who do that teaching are the same people who would train a new hire, because they're the ones who know. Your institutional knowledge stops being the bottleneck and becomes the asset.
And then the part that sounds like a metaphor but isn't: the model keeps getting smarter. Every correction from the people using it — actually, we do it this way here — makes it better at being your accountant, specifically. The compounding curve is the same as a human employee's. It just moves faster.
Six months in, the grad you hired is the person you go to first. The model follows the same trajectory. What you have isn't a tool you bought. It's an employee you trained.
The parallel holds because it isn't really a parallel. It's the same skill your business already runs on — the one you use every time you bring someone new onto the team. You already know how to train an employee. Point that same instinct at a model, and you're not adopting a new technology. You're making a hire.
Curious what your AI hire would look like? Every business has a different first role to fill — the accountant, the operations lead, the customer service veteran who knows every account by name. Let's talk about which one is yours.