Akido Labs is testing a different shape for medical care: appointments where patients may spend little or no time with a doctor, while an LLM-based system helps collect symptoms, generate diagnostic possibilities, and recommend next steps.
The company says the model can expand access to care, including for some Medicaid patients and people served by its street medicine team. But the approach also raises hard questions about what counts as medical judgment, how much patients understand about AI’s role, and whether doctors reviewing recommendations after the visit can reliably catch mistakes.
How Akido’s appointment model works
At a small number of clinics in Southern California run by Akido Labs, patients can get specialist appointments on short notice. Some of those patients are on Medicaid, a group that often faces long waits and limited provider availability.
The visit does not necessarily center on a doctor. Instead, a patient speaks with a medical assistant, who listens, asks questions, and gathers information. The assistant has limited clinical training, while the deeper diagnostic work is handled by ScopeAI, Akido’s proprietary LLM-based system.
ScopeAI transcribes and analyzes the conversation between the patient and assistant. It then produces material for a doctor to review, including a concise note, the most likely diagnosis, two or three alternative diagnoses, recommended next steps such as referrals or prescriptions, and justifications for each diagnosis and recommendation.
Jared Goodner, Akido’s CTO, described the company’s goal directly: “Our focus is really on what we can do to pull the doctor out of the visit.”
Why the company says this can expand access
According to Prashant Samant, Akido’s CEO, the model allows doctors to see four to five times as many patients as they could previously. The access argument is straightforward: if doctors can review more visits and spend less time in each appointment, more patients can move through the system.
The source article places that claim against a broader pressure point. Americans are getting older and sicker, many already struggle to access adequate health care, and a pending 15% reduction in federal funding for Medicaid is expected to worsen the situation.
Akido is using ScopeAI in cardiology, endocrinology, and primary care clinics. It is also used by Akido’s street medicine team, which serves the Los Angeles homeless population. That team is led by Steven Hochman, a doctor who specializes in addiction medicine.
For the street medicine team, the workflow changes how quickly some patients can get treatment. Previously, Hochman had to meet a patient in person in order to prescribe a drug to treat an opioid addiction. Now, caseworkers using ScopeAI can interview patients, and Hochman can later approve or reject the system’s recommendations.
“It allows me to be in 10 places at once,” he says.
Since adopting ScopeAI, the team has been able to get patients access to medications to help treat substance use within 24 hours. Hochman calls that “unheard of.”
What makes ScopeAI different from common medical AI
AI is already present in medicine in several forms. The source article cites computer vision tools that identify cancers during preventive scans, automated research systems that help doctors sort through medical literature, and LLM-powered medical scribes that take appointment notes for clinicians.
Those tools typically support doctors inside existing medical routines. ScopeAI is more ambitious because, according to Goodner, it can independently complete cognitive tasks that make up a medical visit.
Those tasks include eliciting a patient’s medical history, generating follow-up questions based on what the patient says, creating a list of possible diagnoses, identifying the most likely diagnosis, and proposing next steps.
Under the hood, ScopeAI is not a single model. It is a set of large language models, each responsible for a specific step in the visit. The system mostly uses fine-tuned versions of Meta’s open-access Llama models, while Goodner says it also uses Anthropic’s Claude models.
During the appointment, medical assistants read questions from the ScopeAI interface. As the patient speaks, ScopeAI generates additional questions and continues analyzing the conversation.
The legal and ethical questions
Experts interviewed in the source article are not rejecting the goal of broader access. Their concern is whether this specific structure moves too much medical reasoning away from doctors and into an AI-guided process.
Emma Pierson, a computer scientist at UC Berkeley, says there is a large gap between doctors and AI-enhanced medical assistants. “I am broadly excited about the potential of AI to expand access to medical expertise,” she says. “It’s just not obvious to me that this particular way is the way to do it.”
Insurance rules also shape how the model can operate. Medicaid allows doctors to approve ScopeAI prescriptions and treatment plans asynchronously, both for street medicine and clinic visits. Many other insurance providers require doctors to speak directly with patients before approving those recommendations.
Pierson warns that this difference could deepen unequal care patterns. “You worry about that exacerbating health disparities,” she says.
Samant says the discrepancy is not intentional and reflects how insurance plans currently work. He also says patients can choose traditional doctor appointments if they are willing to wait.
Legal questions remain unsettled. Glenn Cohen, a professor at Harvard Law School, says an AI system that effectively acts as a “doctor in a box” would likely need FDA approval and could run into medical licensure laws. The California Medical Practice Act says AI cannot replace a doctor’s responsibility to diagnose and treat a patient, but doctors may use AI and do not need to see patients in person or in real time before diagnosing them.
Neither the FDA nor the Medical Board of California could say whether ScopeAI was on solid legal footing based only on a written description. Samant says Akido is in compliance because a human doctor reviews and approves all diagnostic and treatment recommendations, meaning ScopeAI does not require FDA approval.
The evidence gap still matters
One concern is whether patients understand how much the system influences their care. Patients do not see the ScopeAI interface. They speak with a medical assistant who asks questions in a familiar clinical style.
DeAndre Siringoringo, a medical assistant at Akido’s cardiology office in Rancho Cucamonga, says he tells patients an AI system will listen to gather information for their doctor. He does not tell them the specifics of how ScopeAI works, including that it makes diagnostic recommendations to doctors.
Zeke Emanuel, a professor of medical ethics and health policy at the University of Pennsylvania who served in the Obama and Biden administrations, worries that the setup may hide the extent of algorithmic influence from patients. Pierson adds: “That certainly isn’t really what was traditionally meant by the human touch in medicine.”
Another risk is automation bias. The source article notes that doctors using AI systems tend to accept system recommendations more often than they should. Pierson says that could be especially concerning when doctors are not physically present for appointments.
Akido says it has tried to reduce that risk. A company spokesperson says automation bias is a valid concern for any AI tool assisting doctor decision-making and that physicians are trained to use ScopeAI thoughtfully.
The company evaluates ScopeAI on historical data and tracks how often doctors correct its recommendations. Those corrections are also used to further train the models. Before using ScopeAI in a specialty, Akido ensures that, on historical data sets, the correct diagnosis appears in the system’s top three recommendations at least 92% of the time.
Still, Akido has not conducted more rigorous studies comparing ScopeAI appointments with traditional in-person or telehealth visits to determine whether patient outcomes improve or at least hold steady. Pierson sums up the issue: “Making medical care cheaper and more accessible is a laudable goal,” she says. “But I just think it’s important to conduct strong evaluations comparing to that baseline.”