The fastest-moving variable in practice operations today is the relationship between AI tools and the people doing the work around them. Nearly seven in 10 medical groups added or expanded AI tools last year, with adoption concentrated in ambient documentation, scheduling, patient communications and RCM workflows. When leaders set their 2026 budgets, 37% put workforce investments — staffing, retention, compensation — at the top of their new spending plans, while 36% named automation as the most important cost-cutting lever for the year. 

Both answers describe the same picture: practices are spending more on people while also expecting AI to absorb a growing share of the routine work those people used to do.  

The result, increasingly, is a quieter form of workforce change: Job descriptions are being rewritten faster. So are org charts, with role boundaries shifting under existing titles and new hires going into different roles than the ones a practice would have filled three years ago. 

What you told us 

Our June 2, 2026, MGMA Stat poll found that despite the increased use of AI in medical groups, most practice leaders (68%) say their organizations have not redesigned a role or adjusted staffing with the help of AI in the past year. Only about one in four (26%) say they have, and another 5% were unsure. The poll had 260 applicable responses. 

Practice leaders who reported AI-driven role or staffing changes noted their moves were small and practical. Most of the activity sits in the front office, revenue cycle, and call center: scheduling, registration, prior auth, billing, and routine correspondence. Here, respondents described automating tasks, shifting what staff handle, and leaving some positions unfilled. Several practices cut or reassigned FTEs (especially in RCM and assistant roles), put AI agents on the phones, or stopped replacing scribes who leave; others took on more oversight or reworked who reports to whom. Clinical effects were narrower — less MA shadowing, fewer in-person scribes, and isolated productivity gains such as higher imaging volume. Most of what respondents told us comes down to doing more with the staff they already have: automating routine work, adding capacity, and updating job descriptions. In short, AI is helped these practices run leaner without eliminating roles outright. 

The majority who reported no changes are mostly still watching and waiting. Many have no current plans, but a sizable group is weighing AI in the same areas: RCM, scheduling, prior auth, call handling, and documentation, including ambient listening and AI scribes. When they name a timeline, it tends to be near-term, three to 12 months out, though uncertainty, governance questions, and staffing concerns keep them from moving. These respondents can point to specific use cases but stay vague on dates. 

The "unsure" respondents acknowledged AI is already running in narrow workflows, even where no one has formally redesigned a role. Dictation, documentation, smarter scheduling, prior auth support, and AI scribes come up often, and at least one group built the staffing plan for a brand-new practice around AI from the start.  

Taken together, adoption is unfolding in stages: early movers go after administrative bottlenecks and selectively trim or shift staff; most are still evaluating the same near-term uses; and a trailing group is putting AI to work in spots without turning those gains into explicit staffing changes. 

How practice staff roles are changing 

A February 2026 report from Deloitte, based on a survey of 100 healthcare technology executives conducted in September 2025, identified workforce preparation as a core requirement for organizations scaling agentic AI. The report found 85% of healthcare leaders plan to increase agentic AI investment over the next two to three years, with the heaviest activity concentrated in back-office, administrative and payment workflows. 

That shift is visible in the specific functions practice administrators are automating first. Our Feb. 10, 2026, MGMA Stat poll on front-office AI priorities ranked scheduling (31%), calls (27%), registration and eligibility (23%) and prior authorization (16%) as the top operational targets for AI investment. Each of those queues had been a staff-intensive first-pass workload. Increasingly, that first pass runs through software, and humans take the cases AI flags as low confidence, complex or patient-facing. 

Revenue cycle is where the pattern is clearest: 

• A 2025 survey from the American Hospital Association and the Assistant Secretary for Technology Policy identified billing and scheduling as the two fastest-growing AI use cases across healthcare.  
• Coding automation can analyze EHR data, assign diagnostic and procedural codes and route low-confidence outputs to human reviewers.  
• Gartner research co-authored by senior principal analyst Robert Potts emphasizes that keeping humans in the loop on flagged codes is a defining design principle, not an afterthought.  

For practice administrators, the day-to-day reality is that coders and billers spend less time on data entry per claim and more time on the share AI cannot resolve confidently. Several recent MGMA Stat analyses, including our April 8, 2026, poll on evaluating AI for coding and charge review, point in the same direction: practice leaders are scrutinizing where automation can take a confident first pass and where human judgment still earns its hours. 

The clinical-side picture 

Documentation is moving along a similar track, though more slowly.  

• The Doximity 2026 State of AI in Medicine Report, based on 3,151 physician respondents across two survey waves, found AI use among physicians climbed from 47% in March-April 2025 to 63% in November 2025–January 2026. Voice-based documentation tools — ambient listening and AI scribes — were used by 29% of respondents in the latest wave, with 94% of physicians either using AI or interested in doing so.  
• An Aug. 5, 2025, MGMA Stat poll found 71% of practice leaders reported some use of AI in patient visits, although 47% said it was applied in 25% or fewer encounters. Of practices using AI in visits, 39% said it had reduced staff workload, 44% said it had not, and 17% were unsure — evidence that adoption alone does not produce role change. Workflow integration is what does. 

For medical assistants (MAs), scribes and clinical support staff, ambient documentation is the closest clinical analogue to what revenue cycle teams have already experienced. A February 2026 analysis from IntuitionLabs suggested that as scheduling, reminders and paperwork generation move toward fuller automation by 2026–27, MAs and nurses could be redeployed toward patient education, follow-ups and other higher-value work. The same analysis raised the prospect of practices adding new roles such as AI coordinators or informatics liaisons embedded inside practice management. 

What practice leaders are seeing on the front lines 

The cost pressure driving these decisions is real, and so is the staffing pressure running alongside it. A recent MGMA Annual Regulatory Burden Report found that 92% of practices had hired or reassigned staff solely to handle prior authorization volume, a stark indicator of how much capacity practices have already redeployed into administrative throughput even before AI tools matured. The most recent CAQH Index reported providers spent an average of 11 minutes per electronic prior authorization (ePA) and 16 minutes per portal-based one, time that scales linearly with patient volume unless the workflow itself changes. 

What to watch as AI-driven changes unfold 

Practice leaders should be wary of three patterns that may emerge: 

1. Redeployment without reduction: Practices growing patient volume or expanding services are likely to describe AI as enabling them to avoid hires they would otherwise have needed. The Sept. 30, 2025, poll already showed the heaviest AI expansion happening in growth-pressured areas (documentation, scheduling, patient communications), where the alternative to automation would have been new headcount. 
2. Tier-shifting within existing job titles: Coders, billers, schedulers and front-office staff increasingly work a second-pass: reviewing AI output, working denials, handling exceptions and managing patient-facing interactions AI hands off. The title on the job description often stays put while the work behind it changes substantially. 
3. New oversight and governance positions: A Jan. 20, 2026, MGMA Stat poll on AI governance documented practices building formal review structures around vendor selection, low-confidence flagging and audit trails. Practice leaders must decide: Do they have a need for an AI coordinator role outright, or can they fold the responsibilities into existing IT, compliance or operations positions? 

For administrators sizing up their own AI rollouts, it’s sound practice to pair every automated workflow with a defined human review point. Track the metrics AI is supposed to move, be that claim turnarounds, days in A/R, note completion time, and/or no-show rate. Budget for the upskilling required to move staff into the new work.  

As noted in our January 2026 cost-cutting analysis, an automation that cannot point to the workflow improvement replacing the labor it removes is rarely cutting cost; it is shifting cost into overtime.