AI won't replace your clinical reasoning — but it will replace repetitive admin work. Here's how private practice owners are using it right now. Read more.
Most practice owners have opened ChatGPT at least once. Some have built workflows around it. A few have already replaced a full human function with an AI agent and didn't look back.
The question is no longer whether AI belongs in private practice. It's whether you're being strategic about where you let it in.
I sat down recently with Erin Clemens, a pediatric therapy practice owner who has spent the last two years testing, adopting, and building AI into her daily operations. What she shared cut through a lot of the noise that surrounds this conversation.
Most practice owners approach AI by asking, "What tool should I use?" Erin flips it.
Start with the problem you're trying to solve. Identify where you or a team member is spending time on repetitive, low-judgment tasks. Then find the tool that fits the problem, not the other way around.
This sounds simple, but it saves you from the trap Brandon and Erin both described: finding a product you love, committing to it, and realizing six months later it doesn't actually match your practice's workflow or specialty.
For daily writing, content creation, and duplicating your own voice and reasoning style, Claude and ChatGPT lead the conversation. Erin noted that Claude has pulled ahead recently, particularly in writing tasks where maintaining a consistent voice matters.
Beyond those two, here are the tools Erin uses on a weekly basis:
The takeaway isn't to adopt all of them. It's to know what each one is built for and match it to a specific gap in your practice.
This is where the conversation gets more nuanced, and Erin was honest about where she draws her own line.
She doesn't currently use ambient scribe tools during patient sessions. Her concern is privacy and security, specifically whether protected health information is truly secure in those platforms. She's waiting for clearer demonstration that security layers meet the standard that clinical environments require.
But her hesitation about ambient scribing doesn't mean she's against AI-assisted documentation in principle. She's 100 percent in favor of it, under the right conditions.
Her framing is worth keeping: AI gets you to about 70 to 80 percent. The clinician closes the gap. It's your job to review what the tool produced, correct it, and sign your name to something you actually stand behind.
According to ASHA's practice resources, clinical documentation standards for speech-language pathologists require clinician accountability for what goes into the record. AI doesn't change that standard. It changes how quickly you can get to a draft worth reviewing.
The concern Brandon raised in the episode is the real one: not that tools like Erin will misuse AI, but that the Susie New Grad down the hall will sign off on AI-generated documentation without reading it carefully. Clinical reasoning cannot be automated. The assessment, the judgment call, the recommendation that comes from two years of working with a specific child's family — that still has to come from you.
Erin is actively demoing AI phone agents for scheduling and answering calls. The technology, she noted, has gotten to a point where you genuinely cannot tell you're not talking to a human.
Brandon's prediction on the billing side is more specific: eligibility verification, prior authorization research, and denial management follow-up are the first administrative functions that AI agents will fully absorb. The technology already exists. Right now, the pricing models haven't caught up to make it financially sustainable for most specialty practices, particularly in the occupational therapy, physical therapy, and speech therapy space. But within 12 to 18 months, that math changes.
If you're building your practice's operational model right now, that's worth accounting for. The functions you're hiring humans to do today may not require a human in two years. That's not a threat to your team; it's a signal about where you should be directing their energy.
Erin said something at the end of the conversation that I keep coming back to.
Nothing is ever going to replace the human connection that therapists have with children and families. That's not a hope — it's a structural reality of what pediatric therapy actually is. The child who finally communicates a need, the parent who cries at the progress report, the relationship built over months of sessions — AI has no access to any of that.
The same is true on the business side. The practice owners who will thrive through the AI transition are the ones who innovate, create, and build human connection at the level no automation can touch. Repetitive tasks get automated. Judgment, creativity, and relationship don't.
For practices that are scaling, this is the moment to get clear on what belongs in each category. What work requires you specifically? What work just requires someone to do it correctly and consistently? The first category is where your energy belongs. The second is where AI, and the right operational partner, can carry the load.
Erin's final checklist for AI tool evaluation is practical and worth applying directly:
And one more thing Erin added that most evaluations miss: contact the developer directly. If a tool doesn't have use cases in your specialty yet, that's not always a dealbreaker. It might be a partnership opportunity to get early access, better pricing, and direct input into how the product evolves for your market.
AI is not coming for your practice. It's already inside it, whether you've intentionally invited it or not. Your team members are using it. Your competitors are using it. The question is whether you're using it with a strategy or without one.
The practices that will grow through this period are the ones that stay in charge of the judgment calls while letting technology absorb the repetition. That's not a passive position. It requires deliberate decisions about where automation earns its place.
If you want to talk through where AI fits into your practice's revenue cycle and operational model specifically, that's exactly the kind of conversation we have on a Discovery Call. Let's see if we're the right fit.
