What Happens in an AI Setup Workshop — and What You Own When It's Over
Workshop vs. course vs. consultant
Three things get sold under the "AI training" banner, and they leave you with very different objects:
| Format | What happens | What you're left with |
|---|---|---|
| Course | Someone teaches your team about AI | Knowledge — and homework nobody does |
| Consultant | Someone studies you and recommends | A slide deck and an invoice |
| Setup workshop | Someone installs and configures with you | A working system on your machines |
None of these is dishonest — courses and strategy have their place. But if the problem is "we know AI matters and nothing in our office actually uses it," the missing piece is an install, not another explanation.
What actually gets installed?
A serious setup workshop leaves five concrete things behind:
- An AI agent on your machines — a tool like Claude Code that can do work in your files, not just chat. (Here's what that setup looks like if you want to see the parts.)
- An office folder — one place where everything the AI reads and writes lives, in plain files you can open.
- A standing brief — who you are, how you sound (built from your real writing), what you sell, and what the AI must never do. This is the difference between generic output and output in your voice.
- A first automation, already running — typically the drafted replies or write-ups that were dying on someone's desk. Chosen to be visible, low-risk, and human-approved.
- The approval habit — the working rule that AI drafts and a person approves, demonstrated live until the team trusts it.
Who should be in the room?
- The owner. The brief is full of judgment calls — what you promise, what you never say — that only the owner can make.
- Whoever runs the day-to-day. They know where the real time goes, and they'll be the heaviest user.
- The skeptic (optional but valuable). A skeptic who watches the thing get built and drives it themselves converts harder than anyone.
What does it cost?
Honestly: it varies widely across providers, and the number matters less than the ownership question below. For reference, our own workshop + installation is a one-time $1,500 (starter) to $5,000 (growth, multi-seat) — one payment, no required retainer, no recurring fee to us. Others price differently; some quietly convert the install into a monthly platform bill, which brings us to:
The one question to ask any installer
"If you disappear tomorrow, does it keep working?"
In a well-structured install the answer is yes: everything is files on your machines plus your own account with the AI provider. If the answer involves the installer's platform, the installer's login, or the installer's monthly fee — you're not buying a setup, you're renting one. That's the line between an installation and a dependency, and it's worth more than any feature comparison.
How to prepare (whether you hire anyone or not)
- Pick the annoying task. The recurring chore you'd most like off your plate — that's the first automation.
- Gather your voice. Five to ten emails or messages you've written that sound like you. The AI learns your voice from real samples, not adjectives.
- List your never-dos. Claims you can't make, topics to stay out of, things that always need a human.
The honest alternative: do it yourself
You don't need to hire anyone to get started. Our free setup guide walks the whole install screen by screen at your own pace — it's genuinely the same foundation, self-served. The workshop buys you speed, your team trained in one room, and your voice and guardrails tuned by someone who's done it many times. Both paths end with you owning the result.
Questions people ask
Is this different from hiring an agency?
Yes, in the direction of the dependency. An agency does work for you forever, billed monthly. A setup workshop builds the capability into your office once, and you own it. (Some businesses want both; but know which one you're buying.)
Do we need new computers?
Almost never. These tools run as ordinary apps on the machines you already have — the heavy computing happens on the AI provider's servers.
What if nobody here is technical?
That's the target audience. Everything operates in plain English, and the workshop itself is the training — your team drives before the installer leaves.
What happens when the AI models change?
Your context — the brief, the voice samples, the guardrails, the memory — is plain files, and it carries forward. Models improving under your setup is the good scenario: same office, smarter employee.