CampfireCan I customize it, add my own ideas, or use it for something outside my clinic?
Yes. Every role we install is a plain-language handoff file you own, so you can add ideas, change how a role behaves, or point the same system at a second brand or an app. The only real limit is AI usage, which rises with how many roles run. Validate each new idea in a chat before building it anywhere.
The question behind the question
Six of my last 58 sales calls asked some version of this. It shows up a few weeks in, once the first roles are running and the owner starts seeing what else the same machine could do. It sounds like "can I add extras if I have ideas, is there a limit?" or "is it fine to advertise something outside physical therapy?" or, from the more technical owners, "should we build this in Claude Code?"
Underneath all three is the same worry: did I buy a fixed thing, or did I buy a thing I can grow? If you have paid an agency before, you already know what a fixed thing feels like. Every new idea was a change request, a new quote and a two-week wait. So when I say you can customize it, I want to be specific about what that means and where it stops.
Why you can: what a role actually is
Each role we install, whether it is the follow-up role that texts a new lead, the content role that drafts your posts, or the Friday reporting role, is a plain-language handoff file. It says who the role is, what it is allowed to do, what it must never do, what good output looks like, and where it gets its information. It is a document, not code. If you can write instructions for a new front-desk hire, you can edit a role. The instructions on the page you own are the product; the AI is just the employee reading them.
That is also why it travels. A second business line, a fitness program, an online course or a product idea does not need a second system. It needs a second set of instructions, and usually a second place to publish from. On one call this year the owner asked whether a fitness offer should run from a separate page or the clinic's page; the system does not care, but your patients and Meta do, so we keep separate brands on separate pages and let the same roles serve both.
The pattern I have watched on calls is that the owner who asks this question is usually the one who ends up building the most. One client, a clinician with a product idea, had spent about ten hours in a project-management tool mapping out how the thing should work. We described it in one paragraph to the AI on the call, it produced a working prototype in front of us, and the reaction was "this is sick." Ten hours became ten minutes. That is the cheapest developer you will ever have, and you own it.
Where it stops
Three limits, and none of them is a wall.
- AI usage. Every role that runs every day uses tokens or credits, which you pay for in your own accounts. Add five roles and that bill goes up. We set a cap before launch and check it mid-month, so adding is a decision, not a surprise. The full breakdown is on what AI usage costs.
- Patient data. Anything that touches health information stays inside the fence we built for the clinic roles. A new marketing idea for a second brand can be as wild as you like; a new idea that reads patient records goes through the same PHI-stripping step the originals use, or it does not run.
- Your time. The system is not the bottleneck anymore, you are. Every new role needs someone to review its first week of output. Add them one at a time and the review stays small.
The one rule: prove it in a chat first
This is the answer to "should we build this in Claude Code?" and to every tool decision after it. Before anything gets built in a database, a no-code tool, an app builder or a code editor, describe it in a single chat and make the AI produce the result by hand. If a working version appears in one paragraph, you have validated the idea and you know exactly what the role's instructions should say. If it does not, you just saved yourself the ten hours.
Owners who skip this step end up with what three of my recent calls described: too many tools, a feeling of being behind, and time sunk into the wrong one. The fix was always the same. One silo per job, one role per tool, and a prototype in a chat before anything is built anywhere else.
How to add your first idea
Write down the job the way you would brief a new hire: what triggers it, what it produces, who checks it, what it must never do. Open a chat, paste that in, and ask for one finished example. Adjust until the example is what you want. Then that text becomes the handoff file, and the role gets scheduled next to the ones already running. The whole thing usually fits inside the twenty minutes you already spend on the Friday check.
If the idea is bigger than one role, a second brand or a product, bring it to a call. We will do the one-paragraph prototype together and decide what it needs before you spend an hour on it.
Questions around the fire
Is there a limit to how many roles I can add?
No fixed limit. Each role adds a little AI usage and a little review time, so we add them one at a time and watch the mid-month usage check. Most clinics run two or three roles for the first twelve weeks and add from there.
Can I use it to market a second business that isn't physical therapy?
Yes. The roles do not care what you sell. Keep the second brand on its own Facebook page and landing page so Meta and your patients see two clear businesses, and keep any patient-data roles scoped to the clinic.
What if I break something while editing a role?
A role is a text file, so you edit a copy, test it in a chat, and swap it in when the output looks right. The previous version is still there. Nothing you type into a handoff file can spend your money or send a message on its own.