Most EHR AI tools are a single feature — a voice-to-note button, a chat window — sitting on top of software that wasn’t designed with AI in mind. They are limited by the architecture beneath them.
Alma was designed alongside PracticeStudio, not added to it. She knows every screen, every workflow, every data element in the system. She has persistent memory. She runs on infrastructure we control entirely. That means new capabilities aren’t additions — they’re extensions of something already fully connected.
Every AI on the market today waits to be asked something. Alma is being built to act. Guided Agency is the capability that elevates Alma from assistant to employee — give her a multi-step workflow, and she executes it from start to finish.
No step-by-step handholding. No toggling between tools. You describe the goal — Alma runs the reports, cross-references the data, and delivers a finished output. The work gets done whether you’re watching or not.
Guided Agency operates within a strict permission boundary — Alma can only take actions that have been explicitly permitted for that user in that practice. She does not act outside those boundaries. The practice defines what Alma is allowed to do. Alma does it.
Recurring workflows can be defined once and run on a schedule. The output is not a suggestion — it is a finished work product.
Example — monthly production review
“Alma, run the monthly production report, the outstanding claims summary, and the collections aging report. Combine them into a formatted executive summary in Excel and flag any claims over 60 days old.”
These are capabilities actively in development or in design — natural extensions of the foundation already in place. Not a wish list. A direction.
When a task completes, Alma texts you. When a claim needs attention, she notifies the right biller directly. No checking back in, no missed flags — Alma follows you out of the software. Staff will eventually be able to text Alma directly, and she’ll know exactly who they are by their number.
Alma already explains why a claim is being rejected. The next step is resolution. Alma works through scrubbing errors directly in the software — fixing what she can, and routing what requires human judgment to the task system with a notification to the right person. Flagged to fixed, with a handoff only when it’s genuinely needed.
Ask Alma about a patient and get a rendered card — photo, balance, open items, clinical flags — not a paragraph of text. Alma knows not just what information to surface but how it should be presented. Role-aware and security-aware: what you see is shaped by who you are and what you’re permitted to access.
Alma already knows your patients, your claims, and your schedule. The next layer is analysis — trends in outcomes, collections patterns, appointment flow, and the questions a practice manager asks but rarely has time to dig into. Alma as an analyst who never stops running the numbers.
The answer is almost always: not yet — but she could. Because Alma is woven into PracticeStudio rather than sitting on top of it, the path from idea to capability is short. Every new feature inherits everything she already knows about your practice, your patients, and your workflows.
That’s not a claim most software companies can make. It’s the direct result of 37 years of building the software from the inside and then building the AI the same way.
If you have a question Alma should be answering, a workflow she should be running, or a problem she should be solving — we want to hear it. This page will keep changing.