Where does our client information go?

That is the first question to ask any vendor, and the first thing settled in any setup I do. The default is inside your practice system and your own tenancy, with training switched off and a data processing agreement in place before client information becomes routine. I build from de-identified examples, so no client record leaves your clinic for me to do the job.

Do we need to tell clients?

Yes, and before AI touches their information, not after. I mapped my own clinic's client journey and found our disclosure sat in the service agreement, after AI had already been used on the intake call. It moved. Where it sits in your journey, and what your team says when a client asks or says no, is part of the setup.

Our team is nervous about AI. Is that a problem?

It is the normal starting point and healthier than the opposite. The training happens on your machines doing your work, so the first thing anyone sees is a real result on something they already had to do. Nervous people become careful users, which is exactly what a clinic needs.

Your clinic remains the author and the record keeper, and your professional and funding obligations do not change. What I set up supports your existing documentation standard rather than replacing it, and I will say so plainly if something you have asked for sits outside what I think is defensible.

Bring AI into your clinic for the right jobs, safely.

Scribes, AI receptionists, the AI feature every platform now has, ChatGPT, Claude. Everyone is selling clinics something with AI in the name. I run a clinic on it every day, so the advice is about what is worth doing in yours.

Get a second opinion→

What it is actually good for in your clinic.

The clinical work is not the target. The writing, reading and chasing around it is.

01
Session notes
A first draft in your template, from what the clinician said. Ten minutes back per note, and the biggest win in almost every clinic.
02
New enquiries
Read, sorted and a reply drafted in your voice before the day starts. Answered in hours instead of days, still sent by a person.
03
Reports and letters
Progress reports and referrer letters drafted from the notes that already exist. Reviewing beats retyping.
04
The admin that repeats
Reminders, filing, handovers between systems. Less AI than automation, and often the better fix.
05
Your own writing
The policy update, the job ad, the newsletter, the letter to a school. Drafted from what you already know.
06
Knowing your numbers
Utilisation, waitlist, referral sources, in plain words from the data you already have.

Not sure? The doubts, answered first.

Whether it is a feature in software you already pay for, Claude on your own accounts, or an AI receptionist on the phones, the same questions come up before anyone will consider it. They are the right questions. Here are the honest answers.

Is it safe for client information?: It can be, and the difference is settings, not luck. Paid plans, training switched off, a data processing agreement in place, and a straight answer from the vendor on where data is stored and processed. A free account on default settings is where clinic data quietly leaks.
Are we allowed to?: Yes, with two things in place. Clients told before AI touches their information, not after. And a clinician who is still the author, still reviews, and still signs off. Your professional and funding obligations do not change, and nothing I set up sends, saves or books on its own.
Someone on my team is already using the free version. Is that a problem?: It is the most common risk in a clinic and nobody intends it: a client's details pasted into a chatbot to tidy up a report. The fix is one page your whole team has read. Which tools are approved, what never goes in, and what to say when a client asks or says no.
The tool we are being sold looks great. Is it?: Sometimes. The demo is built to impress, so the questions to put to the vendor are the boring ones: where does the data go, does it leave Australia, will they sign an agreement, does a person review before anything is saved or sent, and what does leaving look like. A vendor who cannot answer those plainly has answered them.
Will my team actually use it?: Nervous is the normal starting point and healthier than the opposite. Trained on their own work, on their own machines, the first thing anyone sees is a real result on something they already had to do. Nervous people become careful users, which is exactly what a clinic needs.

Brandon broke down the practicalities of AI in a way that made sense for me and my specific OT business.

Sarina A'Vard, Lavender Kind

How I help, three ways.

Fixed fee, a date it finishes, and a person still reviewing everything at the end of it.

01 Guidance
A technology second opinion, before you sign.
You are weighing up a quote from a vendor, a demo that impressed, or a tool someone on your team loves. I read it the way a technology person does, not the way the brochure wants me to: what it actually does, what it connects to, what your data does, what it costs to leave, and whether it fits the outcome you are after. Sometimes the answer is yes, buy it. Sometimes the right tool is one you already pay for.
What that includes
- Product and vendor review against your actual use case
- The questions to put to the vendor, and what a good answer sounds like
- A plain recommendation, including when the answer is a feature you already have
02 Enablement
Claude in your business, and your team trained on it.
The tool most clinics should start with is Claude on their own paid accounts, connected to Microsoft 365 or Google Workspace. Set up properly, then your team trained on their own work: the report they were already writing, the enquiry they were already answering. They leave knowing what to reach for it for, and what never goes in.
What that includes
- Accounts, settings and connections done properly on your tenancy
- Hands-on training with your team, on real jobs from their week
- A one-page policy your whole clinic has read: approved uses, what never goes in, what to say to clients
03 Builds
Something built for your clinic.
The notes prompt in your practice system, the morning enquiry digest, the report draft, an automation with AI in the middle of it. Built with the person who will use it, run in a real week, tuned, then handed over with a cheat sheet. Built from de-identified examples on your accounts, so no client record leaves your clinic for me to do the job.
What that includes
- Built on your accounts and your tenancy, nothing of mine left behind
- A person reviews everything it produces, by design
- Handed over with a cheat sheet and no dependency on me

Already running in clinics

All the builds→

Built by Cadence

Built by Cadence3 min read

Supervision notes that build and file themselves

Transcribed supervision sessions turned into clean, branded notes and filed to the right clinician's folder automatically. Around 15 to 30 minutes back per session.

Built by Cadence

Built by Cadence3 min read

Documents refreshed without starting from scratch

Old policies, procedures, handouts and templates rebuilt into the clinic's current branding and formatting, so the team reviews rather than retypes.

Common questions

It depends on the job, which is the honest answer. For session notes, the AI feature in your practice system set up properly, because the data already lives there. For most of the rest, Claude on your clinic's own paid accounts connected to Microsoft 365 or Google Workspace. A dedicated scribe or receptionist is sometimes right, and I will say so.

Next step

Weighing up an AI tool? Get a second opinion.

Tell me what you are thinking of buying, subscribing to or moving to. One call, an outside technology view on whether it fits your clinic, and what I would do instead if it does not.

Get a second opinion→hello@bycadence.com.au

Bring AI into your clinic for the right jobs, safely.

What it is actually good for in your clinic.

Session notes

New enquiries

Reports and letters

The admin that repeats

Your own writing

Knowing your numbers

Not sure? The doubts, answered first.

How I help, three ways.

A technology second opinion, before you sign.

Claude in your business, and your team trained on it.

Something built for your clinic.