What I built
A custom AI script that runs entirely on a pharmacy owner's laptop. It reads sensitive files — inventory data, patient-adjacent records, operational documents — and answers questions about them in plain English. Need to know which supplier had the best price on a specific item last quarter? Ask it. Need to cross-reference an inventory count against a purchase log? Ask it.
The difference between this and ChatGPT is where the work happens. Nothing gets uploaded. Nothing gets sent to a server. The AI model lives on the owner's machine, the data stays on the owner's machine, and the answers come back without ever touching the internet. For a pharmacy handling anything adjacent to patient health information, that distinction isn't a nice-to-have — it's a compliance requirement.
Why "local" matters for sensitive data
Most of the AI tools people reach for — ChatGPT, Gemini, Claude, Copilot — work by sending your input to an external server, processing it there, and sending the result back. That's fine for writing a social media caption or brainstorming a marketing plan. It's not fine when the input is patient records, financial statements, legal documents, or anything covered by HIPAA, SOX, or attorney-client privilege.
The moment sensitive data leaves your building and hits someone else's server, you've created a compliance exposure. Even if the AI provider promises encryption or says they don't train on your data, the data still left. That's the part regulators care about.
A local AI model changes the math entirely. The model runs on your hardware. Your files never leave your machine. There's no API call, no cloud endpoint, no third-party server in the middle. The data stays where it started — inside your building, on your computer, under your control. That's the difference between a useful AI tool and a compliance violation.
Who this is for
This isn't just a pharmacy thing. Any business that handles sensitive data and wants to use AI without creating a compliance risk is a fit:
- Medical practices and dental offices — patient records, treatment histories, insurance data
- Law firms — case files, contracts, privileged communications
- Accounting firms and financial advisors — tax returns, financial statements, client portfolios
- Insurance agencies — claims data, policyholder information, underwriting files
- Trades businesses — customer PII, contracts, proprietary pricing sheets, bid documents
If you've ever thought "I'd love to use AI for this, but I can't upload this data anywhere," a local tool is the answer.
What it replaces
This isn't a SaaS replacement in the traditional sense. It replaces the risk of using cloud-based AI tools with sensitive data, and it replaces the cost of hiring a development shop to build custom internal tools. Here's what the alternatives look like:
| Option | Typical cost | What you give up |
|---|---|---|
| Custom dev shop | $5,000–$50,000+ per tool, one-time |
Months of back-and-forth; you don't understand or control what they built; ongoing maintenance fees |
| Cloud AI + BAA | $20–$100/mo + compliance overhead |
Data still leaves your building; BAAs don't eliminate risk, they shift liability; ongoing subscription |
| Local AI via Bootcamp | One-time cost You own the tool. No ongoing fees. |
Nothing. Data stays local. You own and control everything. |
The cloud AI path deserves a closer look. Even with a Business Associate Agreement in place, you're still sending data to someone else's infrastructure. A BAA doesn't prevent a breach — it defines who's liable after one. With a local tool, there's no data transfer to protect against in the first place.
How it actually works
I build the script during the Owner Bootcamp. The client sits with me through the process so they understand what the tool does and how to change it. Here's what the setup looks like:
- A local AI model — runs on the client's own computer. No internet connection required once it's installed. No API key, no monthly bill, no account to manage.
- A custom script — written for the client's specific data and questions. It reads the files the owner points it at and returns answers in plain English.
- Cursor as the editor — the client learns to modify and extend the script themselves. Need to add a new file type? Point it at a different folder? Change how it formats the output? They can do it without calling a developer.
There's no server to maintain, no hosting bill, no software subscription. The tool and the data belong to the owner. If they want to stop using it, they just stop. If they want to change it, they open it up and change it. That's the whole model.
Want something like this?
This is an Owner Bootcamp build. Over 30 days, I work with you to build the tools your business actually needs — custom AI scripts, internal dashboards, client-facing sites — and I teach you to run and extend them yourself. Everything stays on your machine. Everything belongs to you.
If you're not sure whether a local AI tool makes sense for your situation, book a free 30-minute conversation and I'll tell you straight. No pitch, no pressure — just an honest answer about whether this fits.