ScopeAI Clinics: LLMs Lead Appointments While Doctors Review

A different kind of visit

Imagine booking a same-day appointment and spending a full half hour describing symptoms and your medical history without feeling rushed. At several clinics in Southern California operated by Akido Labs, that scenario is becoming reality. But the person listening carefully during that session may be a medical assistant guided by an AI system rather than a physician.

How the workflow works

Akido uses a proprietary system called ‘ScopeAI’ — a collection of large language models that transcribe and analyze conversations between patients and medical assistants. The assistant follows questions generated by the system. As the dialogue unfolds, ‘ScopeAI’ proposes follow-up questions, lists of likely conditions, and a concise visit note that highlights the top diagnosis, alternative diagnoses, recommended next steps, and justifications for each recommendation. A licensed physician later reviews and approves or modifies those recommendations.

Where this is being used

ScopeAI is deployed in cardiology, endocrinology, primary care, and Akido’s street medicine program serving people experiencing homelessness in Los Angeles. In that outreach work, caseworkers interview patients using the system and clinicians review recommendations asynchronously, enabling faster access to medications and referrals than was previously possible.

Efficiency gains and access benefits

According to Akido leadership, this model can amplify physician productivity by four to five times. Faster appointments and broader coverage can be particularly valuable for Medicaid patients and other populations facing long waits and limited provider availability. Clinicians involved report being able to reach more patients and, in some cases, start treatment within 24 hours — an outcome described as unprecedented by the street medicine team.

Risks and ethical concerns

Experts caution that shifting much of diagnostic cognitive work to LLMs raises safety and equity issues. There is a substantial expertise gap between doctors and AI-assisted medical staff, and closing that gap with automation could introduce diagnostic errors or other harms. Another worry is automation bias: clinicians reviewing AI outputs may be inclined to accept the system’s recommendations even when they disagree with them, especially if the review happens asynchronously and the clinician was not present for the visit.

Regulatory and legal questions

A system that effectively acts as a ‘doctor in a box’ could attract regulatory scrutiny. FDA approval might be required for tools that autonomously diagnose and treat, and state medical practice laws limit who may practice medicine. Akido argues that ‘ScopeAI’ is not autonomous because every recommendation is reviewed by a licensed physician, and the system was intentionally designed to avoid operating entirely independently.

Transparency and informed consent

At present patients interact with a human assistant and are not shown the ‘ScopeAI’ interface. Staff report telling patients that an AI is listening to gather information for their doctor, but detailed explanations about how recommendations are produced are often omitted. Critics argue that this could obscure the role of AI in care and affect patient trust.

Evidence and evaluation gaps

Akido tests ‘ScopeAI’ on historical datasets and monitors how often clinicians correct its suggestions, using those corrections to retrain models. The company requires that the correct diagnosis appear among the system’s top three suggestions at least 92% of the time on retrospective tests. However, Akido has not published randomized or comparative studies that measure patient outcomes versus traditional in-person or telehealth visits. Researchers say such trials are necessary to determine whether the approach preserves or improves safety and effectiveness, and to quantify the impact of automation bias.

Balancing innovation and caution

The promise of cheaper and more accessible care is compelling, especially amid provider shortages and funding pressures. But experts emphasize the need for careful evaluation, transparent patient communication, and regulatory clarity. For now, the Akido model illustrates both the potential of LLMs to scale medical services and the unresolved questions about safety, fairness, and oversight when AI takes on central cognitive roles in health care.