Artificial intelligence is changing medical practice operations through gradual redesign rather than broad workforce replacement. Medical groups are adding or expanding tools in ambient documentation, scheduling, patient communications and revenue cycle management (RCM), while leaders continue to prioritise staff, retention and compensation in spending plans. A June recent MGMA Stat poll of 260 applicable responses found that 68% of practice leaders had not redesigned a role or adjusted staffing with AI in the past year, while 26% had and 5% were unsure. The operational shift is therefore uneven and still developing. Some practices are rewriting job descriptions, adjusting reporting lines and leaving selected posts unfilled, while most remain in assessment mode around the same administrative and documentation use cases.
Must Read: Physicians Need More Support to Use AI
Administrative Workflows Carry the First Shift
Administrative workflows are absorbing the earliest changes. Nearly 7 in 10 medical groups added or expanded AI tools last year, with activity concentrated in ambient documentation, scheduling, patient communications and RCM workflows. At the same time, 37% of leaders placed workforce investments at the top of new spending plans for 2026 and 36% named automation as the most important cost-cutting lever. The two priorities sit together in practice operations, where AI absorbs routine tasks while people remain central to capacity, supervision and patient contact.
Among organisations making role or staffing adjustments, most moves are small and practical. Front-office, RCM and call-centre functions carry much of the activity, especially scheduling, registration, prior authorisation, billing and routine correspondence. Practices automate tasks, shift staff responsibilities and leave some posts unfilled. Some cut or reassign full-time-equivalent posts, especially in RCM and assistant roles. Others place AI agents on phones, stop replacing departing scribes, add oversight tasks or rework reporting relationships. The dominant pattern is leaner operations rather than direct role elimination. The majority without formal changes are mostly watching and waiting. Many have no current plans, while others weigh RCM, scheduling, prior authorisation, call handling and documentation. When timelines are named, they tend to fall three to 12 months out. Uncertainty, governance questions and staffing concerns slow movement.
Human Oversight Moves into Higher-Value Work
Workflow targets show how role content changes under familiar job titles. In front-office AI priorities, scheduling accounts for 31%, calls for 27%, registration and eligibility for 23% and prior authorisation for 16%. These queues previously required staff-heavy first-pass handling. Increasingly, software performs the initial pass and people take cases that are low confidence, complex or patient-facing.
RCM shows the clearest version of this shift. Billing and scheduling are the two fastest-growing AI use cases across healthcare. Coding automation can analyse electronic health record data, assign diagnostic and procedural codes and route low-confidence outputs to human reviewers. Human review remains part of the design, especially when codes are flagged for additional assessment. In daily practice, coders and billers spend less time on data entry per claim and more time on work AI cannot resolve confidently. Denials, exceptions and patient-facing hand-offs become a larger share of the job. The title may stay the same, but the work behind it changes substantially. The pressure is particularly visible in prior authorisation, where 92% of practices had hired or reassigned staff solely to handle volume. Electronic prior authorisation takes an average of 11 minutes, and portal-based work takes 16 minutes on average, leaving time demands tied directly to patient volume unless workflows change.
Clinical Documentation Changes More Gradually
Clinical roles are changing more slowly than administrative roles, with documentation closest to the RCM pattern. AI use among physicians rose from 47% in March-April 2025 to 63% in November 2025-January 2026. During November 2025-January 2026, 29% of physicians used voice-based documentation tools, including ambient listening and AI scribes, while 94% were either using AI or interested in doing so.
Practice-level use in visits does not automatically alter staffing. Among practice leaders, 71% indicated some use of AI in patient visits, although 47% applied it in 25% or fewer encounters. Of practices using AI in visits, 39% saw reduced staff workload, 44% did not and 17% were unsure. For medical assistants (MAs), scribes and clinical support staff, effects remain narrower. Examples include less MA shadowing, fewer in-person scribes and isolated productivity gains such as higher imaging volume. Scheduling, reminders and paperwork generation may move further towards automation by 2026-27, with MAs and nurses redeployed towards patient education, follow-ups and other higher-value work. New roles may also emerge, including AI coordinators or informatics liaisons inside practice management. Narrow deployment can also exist without formal role redesign, with dictation, documentation, smarter scheduling, prior authorisation support and AI scribes already running in some workflows.
AI-driven workforce change in medical practices remains measured, practical and uneven. Early movers focus on administrative bottlenecks, selective redeployment and revised duties, while many organisations continue to evaluate near-term use cases. The central operational requirement is not job replacement, but explicit workflow design. Automated processes need defined human review points, clear metrics such as claim turnaround, days in accounts receivable, note completion time or no-show rates and planned upskilling. Without a clear workflow gain replacing the labour removed, automation risks shifting cost into overtime rather than reducing it, even when software appears to reduce manual effort.
Source: MGMA
Image Credit: iStock
