Published on
Three types of artificial intelligence (AI) technologies are quickly moving to the urgent care exam room: real-time transcription; real-time coding; and clinical decision support systems.[1] They promise to reduce clinician burden per visit, improve documentation in the medical record, reduce cost per episode of care, and improve reimbursement. They may create challenges related to clinical ownership, medical decision making, and compliance, however.[2] Notes and code suggestions generated by AI for urgent care patients are likely to increase, but the clinician’s documentation will remain closely tied to their clinical reasoning and thinking as AI expands the clinician’s expertise.
Understanding how such AI-assisted documentation tools will work, ways to improve note effectiveness and efficiencies, and the need for human oversight are essential steps for urgent care clinicians to capitalize on the benefits of AI while maintaining the obligation of patient-centered care and accountability for their documentation.
The Documentation Dilemma in Urgent Care
Urgent care medicine is a high-volume industry, but the greatest burden to clinicians is documentation. Clinicians must provide an accurate, compliant medical record that reflects medical decision making, while supporting the coding and reimbursement and regulatory processes. All this must be captured in spite of the fact that the average visit is 10 to 15 minutes overall.
As payer requirements for documentation have increased, so have the ever-evolving evaluation and management (E/M) guidelines. The 2023 American Medical Association E/M documentation guidelines greatly eased the documentation of certain elements of the history and the examination, but also increased the emphasis on the medical decision making (MDM) component of the note, which is the most cognitive, complex part of the note.[3] The challenge centers around how to capture the clinician’s thought process as clearly and succinctly as possible, without writing an administrative essay in the chart.
With the continued rapid rise in the volume and range of services performed by urgent care centers, healthcare systems have looked to technology to help address some of the challenges urgent care clinicians face. One of the most promising and yet most controversial technologies is AI-assisted documentation.
The Rise of AI in Clinical Documentation
AI introduction into the urgent care space has quietly changed how encounters are captured and coded.[4] Early dictation and voice-to-text software functioned solely as transcribers. The natural language processing and machine learning systems now can detect context, structure documentation, and even assign the most likely E/M codes.4
Advanced ambient documentation products are now able to listen to and summarize the visit and then produce a draft within a few seconds. Some of these systems interface with the EHR, prepopulating the history, physical examination, and assessment fields. Others include real-time coding suggestions or can identify documentation deficits that are likely to negatively affect compliance.4
Urgent care clinicians could benefit substantially from AI-enabled documentation—in fact, pilot trials suggest that it could relieve up to 60% of after-hours charting and improve patient satisfaction with the in-person visit.4 As a result, clinicians can devote their full attention to their patients.
However, there are some caveats to the use of AI. Errors can be made if clinical language is misinterpreted, transcription errors are made, or coding recommendations are too aggressive, for example.4 As with any new technology, efficiency must also be weighed against risk.
AI and the Medical Decision Making Process
The MDM section of the chart details the reasoning behind the treatment plan. It describes the synthesis of data, differential diagnoses, management options, and risk profile that leads to decisions on patient care. For urgent care, the MDM section is more heavily weighted due to the sparsity of initial data and the high diagnostic uncertainty.
In addition to accuracy risks, AI-generated summaries impact the complexity of a visit. If an AI assistant correctly determines that the clinician ruled out a serious diagnosis, but the reasoning is not documented, then the complexity of the visit may be underestimated based on the documentation. Whereas, if the tool overstates risk or includes boilerplate phrases to justify higher acuity, it may create a potential compliance pitfall for incorrectly coding a higher-level E/M service than what was performed.[5]
The clinician must review, edit, and validate the generated clinical note to ensure the documentation accurately reflects and upholds their medical decision making, rather than endorsing the decision making of the algorithm. In this model, AI is a superpower scribe that can efficiently generate a summary, but the clinician must define the direction and supervise the output.
Coding and Compliance Considerations
Similarly, AI can improve coding accuracy and reduce coding variation.[6] AI can identify patterns of documentation and AMA E/M guidelines to generate a level of coding based on the complexity of the documentation. This may help avoid the problems associated with undercoding and overcoding.
However, this could create potential issues with compliance, as the Centers for Medicare & Medicaid Services and commercial payers require documented services to substantiate the clinician’s work and decision making.[7] Because the language model is often embedded in the note through language or code generation, the clinician needs to take ownership of the generated note.
Auditors will look for charting that suggests personalized decision making for the individual patient. Template comments that repeat or contradict the note content can be a red flag, even if the comments are generated with an AI tool. During payer audits, the clinician is responsible for everything documented in the record.
Urgent care organizations can reduce compliance risk by establishing their own policies and procedures upon adopting the use of AI technology. These may include directives that:4
- AI-generated documentation is reviewed and signed by clinicians before submission.
- AI systems label all machine-generated text.
- Audits are completed regularly to assess the quality of the generated notes against audio transcripts or notes generated by a human.
- Clinicians will be trained in the responsibility for documentation and other ethical aspects of AI.
Early implementation of such directives allows organizations to reap the efficiency benefits of AI while maintaining compliance.
Ethical and Legal Implications
As AI is increasingly becoming integrated into clinical care while also emerging within legal and ethical questions, such as determining the authorship of AI-generated notes (ie, the clinician, the vendor, both). If clinical findings are misrepresented or risk is overstated, who is liable? In recent guidance, the clinician remains the author of a record, and their signature identifies that the information is accurate, complete, and a reflection of their clinical reasoning, irrespective of whether it was written by the clinician or a machine. The principle is consistent with existing documentation standards and relevant case law.4
Patient privacy presents another challenge: Most AI documentation tools record and process clinician-patient conversations through a cloud-based service. Clinicians should ensure that all vendors with access to personal health information meet the HIPAA standards and a business associate agreement is in place. Patients should be made aware of and consent when AI technology is being used in the encounter, as appropriate to the individual organization’s policies and procedures.
Bias is another concern, as AI algorithms trained on non-representative populations may simply replicate existing inequities in care. Developers have worked to manage this type of bias. AI operates objectively. However, the data it processes is subject to human bias. It is important to remember the 3 guiding principles in ethical AI: transparency; accountability; and clinical oversight. This way, AI can act as a partner for staff, rather than as an unseen risk.
The Practical Reality in Urgent Care
Clinicians working in urgent care see 30-50 patients daily and may experience documentation fatigue. Even small time savings within each encounter may provide substantial relief for clinicians trying to minimize documentation time. Some ways AI can help include:
- Ambient Scribing: The AI listens during the consultation and generates a note which is reviewed, edited, and signed by the clinician.
- Coding Assistance: Based on a gap analysis of the documented MDM, the system can recommend an E/M level and catch missing documentation (eg, data review, risk documentation).
- Decision Support: Some systems provide real-time prompts or references to remind the clinician of guideline-based management recommendations or differential diagnoses during the encounter.
- Administrative Efficiency: Systems automatically populate lab results, medication lists, and discharge instructions to save time and to reduce errors. Each of these features can potentially improve workflow, but clinicians must realize that AI documentation is less than perfect. Technical errors can occur when AI documentation drops a word or misunderstands a medical term. For example, “no tenderness” could be mistakenly transcribed as “note tenderness.” The last review of the clinician is important. Urgent care networks have run pilots with clinician champions—who are experienced clinicians and coders—who continuously learn and improve the performance of the AI, identifying potential problems as they arise and encouraging peer use of the AI system before it is widely adopted.
AI’s Influence on Education and Training
AI-assisted note generation may have educational value for early-career clinicians or recent graduates with limited experience with E/M documentation.[8] AI-assisted documentation could build on itself by indicating potential missing elements or suggesting potential risk categorization depending on prior knowledge and by reinforcing best coding/documentation practices. Additionally, there is concern that using AI tools to code cases may mean that clinicians become less familiar with E/M guidelines and that blindly accepting AI-generated text makes it more difficult for clinicians to understand that documentation and coding decisions are based on medical decision making. Continuing medical education must adapt as the profession begins to document and code using AI tools.
Case Example: The Promise and the Pitfall
A 45-year-old male patient presents with cough and breathing difficulty. An AI documentation assistant listens to the encounter and generates the following MDM statement: “acute bronchitis – low risk”. However, the clinician had strong concern for pulmonary embolism and had ordered a D-dimer for that reason. The AI summary did not incorporate this clinical reasoning and therefore misconstrued subtleties of the case.
However, signing that note makes the encounter a Level 3 visit. While a straightforward case of bronchitis has been addressed, the risk stratification of possible embolism presents a higher complexity of MDM. The clinician would have to add and sign the MDM to attach a Level 4 code to the encounter. This example demonstrates that AI can document what is spoken but not what is considered in the mind of a provider. The clinician’s narrative remains foundational.
The Future of AI in Urgent Care Documentation
In the future, clinicians could leverage AI algorithms to optimize documentation and workflows of urgent care medicine, such as:[9]
- Predictive analytics that identify patients at risk for adverse outcomes.
- Automated quality metrics, measuring adherence to clinical guidelines.
- Population health perceptions based on aggregate, de-identified data to optimize community health.
As these tools mature, documentation and coding may become standardized across the urgent care industry. AI has the potential to reduce variability, prepare clinics for audits, and improve accuracy in the revenue cycle.
However, these advantages are only realized with planning that involves an interdisciplinary team of clinicians, coders, compliance officers, and IT experts with regular monitoring, feedback loops, and transparency about the evaluation of outcomes.
Rather than being touted as compliance and coding shortcuts, AI should be touted as a clinical partner that makes clinicians more efficient without replacing their judgment. The focus should be on the fact that AI does not replace the clinician but gives the clinician more time with patients and less paperwork.
Conclusion: Maintaining the Human Element
AI represents the next evolution of urgent care medicine documentation as well as an opportunity to document, organize, and summarize the urgent care visit and potentially overcome one of the biggest pain points for all urgent care clinicians. Used appropriately, AI can help the clinician document in a clear, concise, and accurate manner while improving patient care and coding compliance. One principle must not change: the clinician remains the author and the chart remains the story of their medical reasoning. AI can assist the writing of that story. As the field of urgent care evolves, AI must be harnessed with intentionality, moral clarity, and the affirmation of the human voice in every note. At the same time, efficiency must never come at the expense of authenticity. When operating in partnership with AI, urgent care clinicians can help support a future in which technology increases and does not drown out the art and science of medicine.
References
- [1]. Ayers AA. Two AI Trends That Will Change Urgent Care. J Urgent Care Med. 2025; 19(8):43-45
- [2]. National Library of Medicine. Toward Alleviating Clinical Documentation Burden: A Scoping Review of Burden Reduction Efforts. https://pmc.ncbi.nlm.nih.gov/articles/PMC11152769/
- [3]. American Medical Association. CPT Evaluation and Management (E/M) Office or Other Outpatient (99202–99215) and Prolonged Services Code and Guideline Changes. AMA; 2023.
- [4]. Patel V, et al. Artificial Intelligence in Clinical Documentation: Implications for Quality, Safety, and Compliance. J Am Med Inform Assoc. 2023;30(7):1231–1240.
- [5]. Sexton S. Three essential strategies for coding excellence in the era of artificial intelligence. J AHIMA. Published June 3, 2025. Accessed December 10, 2025. https://journal.ahima.org/page/three-essential-strategies-for-coding-excellence-in-the-era-of-artificial-intelligence
- [6]. Medwave. How AI is Improving Medical Coding Accuracy and Efficiency. Medwave; Sept 29, 2024. https://medwave.io/2024/09/how-ai-is-improving-medical-coding-accuracy-and-efficiency/
- [7]. Centers for Medicare & Medicaid Services. Evaluation and Management Services Guide. CMS; 2024.
- [8]. National Library of Medicine. Generative Language Models and Open Notes: Exploring the Promise and Limitations. https://pmc.ncbi.nlm.nih.gov/articles/PMC10797501/
- [9]. National Library of Medicine. Improving Clinical Documentation with Artificial Intelligence: A Systematic Review. https://pmc.ncbi.nlm.nih.gov/articles/PMC11605373/
Brad Laymon PA-C, CPC, CEMC is a physician assistant with over 26 years’ experience in Urgent Care medicine. He is also a Certified Professional Coder and Certified Evaluation and Management Coder for the past 11 years. He is the founder of Coding Excellence, LLC, which is a medical coding consulting business.
