Street Smart: A Los Angeles Street Medicine Program Is Proving AI Works for Healthcare's Hardest Cases

April 2026 Authors: Karthik Murali, Akido Labs Dr. Rishi Patel, Akido Labs Jared Goodner, Akido Labs Prashant Samant, Akido Labs

Executive Summary

American healthcare faces a moment of unprecedented convergence. Medicaid cuts threaten to unravel safety net care across the country. A physician shortage crisis deepens, with projections showing a deficit of up to 86,000 doctors by 2036. And artificial intelligence has matured from experimental curiosity to clinical reality—with 66% of U.S. physicians now using AI tools, up from 38% just two years ago.¹˒²

The question is no longer whether AI will transform healthcare delivery, but how it will do so in the real world. Will it simply automate documentation—or will it force-multiply physicians and expand true care capacity, enabling providers to see more patients without sacrificing quality? 

The promise is not efficiency alone. It is the ability to deliver deeply personalized, specialist-informed medicine to everyone, not just the privileged few. That means giving patients real answers in the moment—surfacing clinical pathways a generalist might not immediately pursue, flagging the cardiac symptoms that warrant a closer look, and bridging the distance between a primary care visit and specialist-level insight. ScopeAI—the technology platform powering Akido's AI-native healthcare system—was built for exactly this purpose: guiding the intake process, investigating symptoms and surfacing relevant clinical information for rapid review by a licensed provider. It does not replace physicians. It extends their reach and transforms the physics of how care gets delivered. The future of AI in healthcare will be defined by whether it expands access, shortens time to resolution, and leaves every patient feeling fully heard and fully cared for.

This paper presents evidence from an unlikely place: the streets of Los Angeles. For nearly three years, Akido has operated an AI-native street medicine program serving one of America's most vulnerable populations—people experiencing homelessness. This is a population facing multiple chronic diseases, struggling with addiction, navigating serious mental illness, and living isolated from traditional healthcare access. These patients represent the hardest test case for any care delivery innovation. If AI-powered care can work here, it can work anywhere.

The results suggest it can.

Key Findings:

63% of patients see a medical provider by their first day of enrollment—in a population where traditional systems typically require multi-week waits and multiple failed attempts to engage

Patients average 2.6 visits per month in Los Angeles, with high-acuity patients averaging 4.2 visits per month (a visit per week)—engagement rates that exceed most traditional primary care settings

87% retention at three months and 67% retention at six months—in a population where loss to follow-up is typically the norm, not the exception

24-hour turnaround for prescriptions to patients seeking medication-assisted treatment for opioid use disorder, compared to the seven day average

Providers see 2-3x more patients and time savings that allows them to focus on the most complex cases, study the latest treatment research and coordinate care with other team members

1,300 patients treated for Substance Use Disorders, with 60% initiated on Medication-Assisted Treatment (MAT) like buprenorphine, which reduces overdose deaths by 50%

31% reduction in emergency department (ED) visits among enrolled patients, with peak ED utilization reduced by 46%

3 specialties (primary care, addiction medicine, and mental health) delivered by a single street medicine provider team, eliminating referral handoffs where patients are most likely to fall out of care.

These outcomes were achieved not through heroic individual effort or unsustainable resource expenditure, but through systematic integration of AI throughout the care delivery process. ScopeAI—the technology platform powering Akido's AI-native healthcare system—is an AI system supporting a complete patient visit by guiding the intake process, investigating symptoms, and surfacing relevant clinical information for rapid review by a licensed provider. ScopeAI transforms the physics of how care gets delivered.

Implications:

This case study demonstrates that AI-powered care delivery models can simultaneously improve access, engagement, clinical outcomes, and provider efficiency. For health systems facing Medicaid cuts, physician shortages, and mounting pressure to deliver value-based care, the Los Angeles program offers a proof point that the model works—and a roadmap for how to deploy it. Since launching its street medicine program in March 2023, Akido has served 5,500 vulnerable patients and enrolled 1,500 individuals into ongoing comprehensive care—demonstrating both clinical effectiveness and operational scalability. The future of healthcare is not being determined in academic medical centers or Silicon Valley boardrooms, but rather on our streets, where Akido's AI-powered street medicine teams transform a potential fentanyl overdose into sustained recovery care.

What works for America's most vulnerable populations can work for everyone.

Prologue: The Inflection Point

On a Tuesday afternoon in South Los Angeles, an Akido community health worker approaches a man in his thirties in a Home Depot parking lot. He's in active withdrawal from fentanyl—trembling, sweating, desperate. He’s about to steal a pair of bolt cutters to exchange for drugs. In the traditional healthcare system, his path to help would be long and uncertain: find a clinic that treats substance use disorders, navigate intake paperwork, wait for an appointment that could be hours, days or even weeks away, hope the motivation to get clean survives the bureaucratic gauntlet, not to mention the risk of getting caught and having to navigate the legal system.

Many people in his situation never make it through.

An Akido Community Health Worker pulls out a tablet. Within minutes, he's completed a comprehensive medical intake for a Substance Use Disorder (SUD). ScopeAI—the technology platform powering Akido's AI-native healthcare system—supports the full patient visit from intake through assessment. With the patient's knowledge and consent, ScopeAI captures the conversation, actively guides the intake process by prompting the health worker with targeted follow-up questions based on the patient’s responses, surfaces relevant clinical information in real time, and generates a structured clinical summary for rapid review by a licensed medical provider specializing in SUD. That same day, the provider calls the patient for a telehealth check, the patient receives a prescription for Suboxone, and a community health worker drives him to the pharmacy to pick it up. He's enrolled in ongoing care with the provider and a community health worker makes regular visits to check his health. Three months later, he's still engaged in treatment—one of hundreds of similar stories from Akido's LA street medicine program.

This is not a hypothetical future. This is happening now, every day, on the streets with one of America's most vulnerable communities.

The American healthcare system stands at an inflection point defined by three converging forces:

The Financial Crisis. Medicaid funding cuts threaten to devastate safety net care across the country. Hospitals that serve vulnerable populations face mounting revenue pressure at precisely the moment when demand for their services is likely to increase. The traditional response—do more with less and stretch already-thin resources even thinner—has reached its breaking point.

The Capacity Crisis. The Association of American Medical Colleges projects a shortage of between 13,500 and 86,000 physicians in the US by 2036.¹ The impact will be felt widely, but rural areas, inner cities, and safety net settings face the most severe shortages. The communities that need care most have the least access to it. No amount of medical school expansion or residency funding will solve this problem fast enough.

The AI Maturity Moment. In January 2026, OpenAI released data showing that 5% of all ChatGPT conversations globally now involve healthcare questions—representing more than 40 million daily active users seeking health information from AI. Nearly 600,000 healthcare-related messages per week come from rural "hospital deserts" where traditional care access is severely limited. And crucially, 66% of U.S. physicians now report using AI tools in their practice, up from 38% in 2023.

AI in healthcare has moved from experimental to operational.

These three forces create both unprecedented risk and opportunity. The risk is that safety net care collapses under financial and capacity strain, leaving millions of vulnerable Americans without access to basic healthcare. The opportunity is that we fundamentally reimagine how care gets delivered—using AI not to replace human clinicians, but to multiply their capacity for care.

The central question facing healthcare leaders today is whether AI can actually deliver on this promise. Can it improve real clinical outcomes for real patients? Can it increase access and engagement, not just documentation efficiency? Can it work for the populations who need it most—or will it primarily benefit the already well-served?

Akido’s Los Angeles Street Program offers answers.

¹ Association of American Medical Colleges, "The Complexities of Physician Supply and Demand: Projections From 2021 to 2036," 2023.

The Challenge: When the Safety Net Has Holes

The American safety net healthcare system was built for a different era. Federally Qualified Health Centers (FQHCs), county hospitals, and community clinics provide essential care to millions of vulnerable Americans—but the model struggles with fundamental structural limitations that no amount of dedication from individual providers can overcome.

The Traditional Safety Net Model: Built-In Barriers

Consider the typical journey for someone experiencing homelessness who needs medical care:

Step 1: Find the right place. In a fragmented system with separate providers for medical care, mental health services, substance use treatment, and social services, knowing where to start is itself a barrier. Many people need help across multiple domains simultaneously, but there's rarely a single door that opens to comprehensive care.

Step 2: Navigate intake. Traditional intake processes require extensive paperwork, insurance verification, identification documents that unhoused individuals often don't have, and medical histories that may be scattered across multiple systems or lost entirely. A comprehensive intake can take 30-45 minutes of staff time.

Step 3: Wait for an appointment. Even after successful intake, seeing a provider typically requires hours of waiting or scheduling an appointment days or weeks in the future. For someone living on the streets, keeping that appointment requires remembering the date and time, having transportation, securing their belongings, and maintaining the motivation to engage in care despite the delay. Each day of waiting is another day for competing priorities, crises, or simple despair to derail the process.

Step 4: Return for follow-up. If the patient makes it to the first appointment, ongoing care requires continued engagement—more appointments, more waiting rooms, more navigation of a system that wasn't designed for people whose daily reality involves finding food, shelter, and safety.

The research on what happens to vulnerable populations in this system is sobering. Follow-up rates are low. Time-to-treatment for urgent needs like substance use disorder can stretch to weeks. Patients cycle through emergency departments because they can't access primary care. The system is designed around the assumption that patients have stable housing, reliable transportation, phones that work, and the capacity to plan days or weeks ahead.

The Workforce Reality: Not Enough Hands

Even if we solved every process inefficiency, we'd still face a fundamental math problem: there simply aren't enough healthcare providers to meet the need.

The physician shortage is particularly acute in safety net settings. Private practices can offer higher compensation, better work-life balance, and lower administrative burden. Safety net providers who choose this work often do so out of mission and commitment—but they're still subject to burnout, still limited by the same 24 hours in a day, still constrained by the same documentation requirements and regulatory complexity as their colleagues in better-resourced settings.

The traditional response to capacity constraints has been to maximize efficiency within the existing model: shorter appointment times, higher patient panels, more advanced practice providers. But there's a floor to how much you can compress appointment times before quality suffers. There's a limit to how many patients one provider can manage. And critically, there's a limit to how much administrative burden providers can absorb—with studiesCKTK showing physicians spending nearly two hours on documentation for every hour of direct patient care.

The Post-Medicaid Cut Landscape

Medicaid cuts threaten to turn a chronic strain into an acute crisis. Safety net providers that have operated on thin margins will face impossible choices: reduce services, close facilities, or try to absorb greater uncompensated care burdens with fewer resources. In California, Medi-Cal reimburses primary care visits at rates 30-50% below Medicare, contributing to the 67% of California physicians who limit Medi-Cal patients.² This inadequate reimbursement, combined with the administrative complexity of serving patients with unstable housing and fragmented care histories, creates a structural deficit that undermines care quality and provider sustainability.

Traditional cost-cutting approaches—reducing staff, limiting services, restricting patient eligibility—all move in the wrong direction. They reduce capacity at precisely the moment when demand is likely to increase. They make access harder when access is already the fundamental problem.

This is the context in which healthcare leaders must answer a critical question: Is there a different model?

² Kaiser Family Foundation, "Medicaid-to-Medicare Fee Index," 2021; California Healthcare Foundation, "California Physicians' Acceptance of New Patients," 2023.

The Street Medicine Practice

Street medicine represents an extreme case of collective challenges to meeting care needs. Patients experiencing homelessness face every barrier the traditional safety net struggles with, amplified:

• Limited to no stable contact information - can't call to schedule appointments or follow up

• High mobility - might move encampments or neighborhoods frequently

• Complex, overlapping needs - medical, mental health, substance use, social services all simultaneously urgent

• Lack of documentation - often no ID, insurance cards, or access to medical history

• Trauma and mistrust - previous negative experiences with healthcare or authority systems create barriers to engagement

• Competing daily survival priorities - finding food, shelter, securing belongings, and safety often must come before healthcare

If a care delivery model can work for this population, it can work anywhere. Street medicine in LA is not just a service program, but a practice for showcasing how AI-powered care can solve the fundamental access and capacity challenges facing the entire safety net system.

Program Overview: Street Medicine at Scale

Akido's Los Angeles street medicine program operates fundamentally differently than traditional safety net care. Rather than waiting for patients to navigate their way to a clinic, care comes to them—in encampments, under freeway overpasses, in parks, parking lots, and anywhere else people are unhoused and living.

But this geographic flexibility alone doesn't explain the program's outcomes. Plenty of street medicine programs operate with traditional care delivery models, and they struggle with the same capacity as clinic-based care. What makes the LA program different is that it was built as an AI-native operation.

The Pod Structure: The Operating Unit for AI-Enabled Care

The program is organized around "pods"—integrated care teams that serve as the foundational unit for care delivery, geographic expansion, and AI integration.

 Each pod consists of:

• 1 medical provider (Physician or Advanced Practice Provider) managing a panel of ~350 actively enrolled patients

• ~8 Community Health Workers (CHWs), each carrying a caseload of 40-45 patients

• 1 Program manager providing supervision and coordination

• Medical assistant support for intake and care coordination

Akido currently operates 8 pods across Los Angeles and Kern, with expansion underway in the Bay Area.

Why This Structure Works

Quality Care. The CHW-to-patient ratio is designed for intensive engagement. With 40-45 patients each, CHWs can maintain frequent contact, build trust, address barriers to care, and coordinate services in ways that would be impossible with larger caseloads.

Scalability. Expanding to a new geography means standing up new pods with the same team structure, the same ScopeAI integration, and the same operational playbook. This modular approach enables Akido to replicate its care model across regions without redesigning workflows or retraining from scratch.

AI integration. Every role in the pod — CHW, MA, provider — uses ScopeAI differently, but within a shared clinical infrastructure. CHWs use it for guided intake, care coordination and administrative support. MAs use it for field-based medical intakes. Providers use it for chart review and clinical decision support across specialties. The pod structure ensures AI augments the full care team, not just the provider.

ScopeAI: The Technology Foundation

At the heart of the program’s capacity multiplication is ScopeAI—the technology platform powering Akido’s AI-native healthcare system. ScopeAI supports a complete patient visit by guiding the intake process and surfacing relevant clinical information, including recent lab results, medication histories, social determinants of health, and longitudinal care plans. Using this context, ScopeAI helps structure the clinical conversation and generates a clinical summary for rapid licensed provider review in real time. ScopeAI’s multi-specialty capabilities allow a single Community Health Worker to run assessments across mental health, medical and addiction medicine in one encounter.

How ScopeAI Works in Practice:

When a Community Health Worker sits down with a patient—whether on the street, in a clinic, or anywhere else—ScopeAI is running on a tablet in real time, supporting the full patient visit from intake through assessment:

1. Conversation capture with consent: With the patient's knowledge and consent, ScopeAI captures the conversation as it happens, converting speech to text with medical terminology accuracy and structuring the encounter in real time.

2. Guided intake and real-time support: As the conversation unfolds, ScopeAI analyzes what is being discussed and prompts the Community Health Worker or Medical Assistant with targeted follow-up questions to further investigate symptoms. These prompts help guide the intake process and ensure relevant clinical pathways are explored in the natural flow of conversation. All clinical assessments and care decisions are made by a licensed provider.

3. Structured clinical information surfacing: ScopeAI identifies and extracts key clinical information—symptoms, medications, medical history, and social determinants of health—from the natural flow of conversation, even when patients share information non-linearly. This information is surfaced in a structured format to support rapid provider review.

4. Summary generation and documentation: Following the encounter, ScopeAI generates draft clinical notes, medical record updates, and structured recommendations across specialties. This is surfaced to the licensed provider in real time as a comprehensive clinical summary, increasing provider capacity while allowing more meaningful time with each patient.

5. Team-based care enablement: Because ScopeAI captures and structures comprehensive information from each encounter, other care team members—case managers, specialists, and community partners—can quickly understand the patient's situation without lengthy handoffs or repeated questioning.

The result: What once required a 30-minute intake process with a provider is now a 5-minute conversation focused on patient needs and treatment. A 45-minute new patient visit that would traditionally require 20+ minutes of documentation afterward becomes a 15-minute conversation with near-zero post-visit charting time.

ScopeAI reduces administrative burden while enhancing clinical quality. As documentation time decreases, providers can focus their cognitive capacity where it matters most: patient engagement, complex clinical reasoning, and the human judgment that no AI can replicate

ScopeAI for Addiction: Specialized AI for Substance Use Treatment

Akido's program developed a specialized application of ScopeAI specifically for medication-assisted treatment (MAT) for opioid use disorder—one of the most urgent needs in the population served.

Traditional MAT intake is notoriously time-intensive. A comprehensive substance use assessment can take 45-60 minutes of provider time, covering detailed drug use history, previous treatment attempts, mental health comorbidities, social support systems, and medical contraindications. With limited provider capacity, this creates a significant bottleneck: patients who need urgent MAT access wait days or weeks for an intake appointment, and many never return.

ScopeAI for Addiction changes the equation:

“Jack,” a substance abuse counselor embedded within the street medicine pod, uses ScopeAI to conduct MAT intake visits. He has the clinical knowledge and relationship-building skills to engage patients effectively, but he's not a medical provider who can make clinical decisions or prescribe medications. ScopeAI supports the complete intake visit—actively guiding Jack through the encounter, prompting targeted follow-up questions, and surfacing relevant clinical information in real time—so that the encounter itself is more thorough and the output is ready for immediate provider action.

The workflow looks like this:

1. Any team member can identify a patient interested in MAT and connect them with Jack (or another counselor)—implementing a true "no wrong door" policy

2. Jack conducts the intake using ScopeAI, which guides him through the entire process, necessary clinical while building rapport with the patient

3. ScopeAI generates a structured clinical summary for rapid licensed provider review, routed directly to a licensed provider

4. The provider reviews the output and conducts a telehealth visit before initiating a prescription. Because ScopeAI has surfaced and organized the relevant clinical information in advance, this review typically takes 5-10 minutes rather than a full 30-minute intake

5. Patient receives treatment (such as a Suboxone prescription) within 24 hours—compared to the previous 7-day timeline

6. The initial team member (typically a CHW) transports the patient to the pharmacy to pick up the prescription—typically the same day it is processed

Provider feedback across the program reflects a consistent theme: ScopeAI enhances clinical judgment rather than replacing it, freeing providers to focus on the complex reasoning and relationship-building that drive outcomes in this population. Providers report processing 2-5 of these intake summariesassessments daily with substantially reduced time investment —work that would be impossible to fit into a traditional clinic schedule.

One provider described it as "something I never thought was possible after years of trying to figure out how to get people into MAT faster.”

ScopeAI for Addiction is a vital proof point of a broader capability —ScopeAI empowers a single Pod/Care Team to deliver care across specialties that traditionally require separate clinician visits. Today, Akido's street medicine providers deliver primary care, addiction management, and mental health treatment — three disciplines that would ordinarily require three separate referral pathways, three separate intakes, and three separate provider relationships that unhoused patients typically struggle to maintain.

The "No Wrong Door" Philosophy

Underpinning all of this is a design principle: any team member should be able to engage a patient and initiate whatever care they need.

In traditional healthcare systems, there are many wrong doors. Show up to the mental health clinic with a substance use problem, and you might be redirected to a different facility. Call the medical clinic when you need a psychiatry appointment, and you'll be transferred or told to call back. Need multiple services simultaneously, and you'll navigate multiple intake processes with different requirements.

The LA program eliminates this friction. A CHW encountering someone who needs medical care can initiate enrollment on the spot. A counselor working with someone on mental health can identify substance use needs and connect them immediately to MAT services. The field MA can conduct a comprehensive intake across specialties wherever the patient is comfortable.

This is only possible because Akido's AI-native healthcare platform creates a shared clinical operating system across the care team. Everyone works within the same system, sees the same structured patient information (with appropriate access controls), and can initiate the same care processes by reducing the operational and documentation complexity that would otherwise require specialized training for each role. All clinical decisions are done by licensed providers.

Results: Outcomes That Matter

The Los Angeles program has now operated for over two years, providing sufficient time to assess whether AI-powered care delivery achieves meaningful clinical and operational outcomes. The results demonstrate that this model works—not just in theory, but in sustained practice with one of healthcare's most challenging populations.

Engagement Metrics: Breaking Through Traditional Barriers

• First-Day Provider Engagement: 63%

  Traditional street medicine programs often struggle with initial engagement, as patients experiencing homelessness may be skeptical of healthcare systems that have previously failed them. Akido's AI-enabled model achieves 63% first-day engagement with medical providers across 1,500 currently enrolled patients—a rate that demonstrates the program's ability to build trust and deliver immediate clinical value. This engagement rate translates to immediate clinical interventions: medication initiation, care plan development, and connection to community resources—all occurring at the first point of contact rather than after multiple outreach attempts.

• Six-Month Retention: 67%

  Perhaps more impressive than initial engagement is sustained retention. Akido achieves 67% retention at six months and 87% retention at three months across its population of approx. 1,500 enrolled patients—metrics that demonstrate the program's ability to maintain continuity despite the inherent instability of serving people experiencing homelessness. The team's ability to spend more time with each patient and be a 'one stop practice' for the patients improves the patient-care team relationship and prevents the fragmentation that typically undermines longitudinal care. This sustained engagement enables the program to address complex chronic conditions, stabilize medication-assisted treatment, and coordinate successful housing transitions—historically the most challenging inflection point where patients disengage from care.

• Engagement Frequency: 2.6 Visits Per Month

  The program maintains intensive longitudinal contact, averaging 2.6 visits per month in its South Los Angeles urban deployment. High-acuity patients receive even more frequent touch points, at 4.2 visits per month, ensuring that the most vulnerable individuals receive the proactive monitoring and rapid intervention their complex needs require.

Clinical Outcomes: Delivering on the Promise of Better Care

• Substance Use Disorder Treatment at Scale

  The program delivers measurable clinical improvements across key health indicators. Akido has treated 1,300 patients for Substance Use Disorders (SUD), initiating 60% of these individuals on Medication-Assisted Treatment (MAT) like buprenorphine, which reduces overdose deaths by 50%.³

  Beyond substance use disorder treatment, the program demonstrates effectiveness across multiple clinical domains. Akido has treated 246 patients for psychiatric disorders, addressing the high prevalence of co-occurring mental health conditions among people experiencing homelessness.

  ³ Sordo et al., "Mortality risk during and after opioid substitution treatment: systematic review and meta-analysis of cohort studies," BMJ, 2017.

• Reducing Emergency Department Reliance

  Preliminary findings from a research partnership with a managed care plan demonstrate the program's impact on acute care utilization. Among enrolled patients, emergency department visits decreased by 31%, with peak ED utilization reduced by 46%. Akido's enrolled patients maintain an average ED rate of 113.5 per 1,000 per month (compared to national average of 258 per 1000 for unhoused individuals)—a metric that reflects the program's success in stabilizing high-risk populations and shifting care from crisis-driven emergency interventions to proactive primary care management.

  Ref: https://www.cdc.gov/mmwr/volumes/72/wr/mm7242a6.htm

Serving the Highest-Need Populations

Akido's outcomes prove particularly significant given the complexity of its patient population. Among enrolled patients, 81% are experiencing homelessness—individuals whose medical needs are compounded by housing instability, food insecurity, and limited access to traditional healthcare infrastructure. This population typically faces multiple chronic conditions, behavioral health challenges, and social barriers that make sustained care engagement extraordinarily difficult.

The AI-powered Pod model's success with this demographic validates its potential for broader safety net applications. If the technology supports better care for patients experiencing homelessness—arguably the most challenging care delivery environment—it demonstrates readiness for deployment across other vulnerable populations with less severe barriers to engagement.

Implications for Healthcare Stakeholders

The Los Angeles results have implications far beyond street medicine. They demonstrate a care delivery model that addresses the fundamental challenges facing the entire safety net system: insufficient capacity, inadequate engagement, fragmented care coordination, and unsustainable administrative burden.

For Health Plans: A Scalable Model for High-Risk Populations

Medicaid managed care plans face increasing pressure to improve outcomes and reduce cost for high-cost, high-need populations while managing medical loss ratios. Akido's model offers a path forward: AI-enabled care delivery that achieves clinical outcomes superior to traditional models while maintaining economic sustainability. Preliminary research findings demonstrate a 31% reduction in emergency department visits among engaged patients, with peak ED utilization reduced by 46%—translating directly to reduced acute care costs while improving patient outcomes.

The program's 67% six-month retention rate proves particularly valuable from a health plan perspective. Sustained engagement enables preventive care, chronic disease management, and medication optimization—the interventions that drive long-term cost reduction and better health outcomes. Traditional programs struggle to achieve this continuity for vulnerable populations; AI-powered care coordination provides the infrastructure to maintain longitudinal relationships despite housing instability and life chaos. For Medicaid plans serving high-utilizing populations experiencing homelessness, this model offers a path to bend the cost curve while fulfilling care quality obligations.

The business case extends beyond direct medical cost savings. Plans struggling with quality metrics—particularly measures like follow-up after hospitalization, medication adherence, and preventive care completion—can leverage AI-enabled teams to achieve performance that traditional care models can't match.

For Safety Net Providers: Extending Capacity Without Sacrificing Quality

Community health centers, public hospitals, CBOs and other safety net providers face an impossible equation: growing patient populations, shrinking reimbursement, and workforce shortages that show no signs of abating. The Association of American Medical Colleges projects a shortage of up to 86,000 physicians by 2036, with primary care and psychiatry—the specialties most critical to safety net care—facing the most severe deficits. AI offers a path to extend clinical capacity — even across specialties— without compromising care quality and oversight.

Akido's experience demonstrates concrete capacity extension. ScopeAI's guided intakes, clinical decision support, and investigative capabilities enable providers to manage higher patient volumes without sacrificing encounter quality or clinical decision-making. Administrative time decreases, allowing providers to dedicate more time to direct patient care. Clinical decision support reduces cognitive burden, preventing the burnout that drives workforce attrition in safety net settings. The result: existing clinical teams can serve more patients while maintaining—even enhancing—care quality.

The workforce implications are particularly important for safety net organizations struggling with recruitment and retention. Providers burned out by documentation burden and administrative complexity leave safety net care for less demanding settings. AI that eliminates evening charting, streamlines intake processes, and supports rather than burdens clinical decision-making can make safety net practice sustainable again. One Akido provider’s observation—that his wife noticed he's not spending two hours every evening on documentation—captures a quality-of-life improvement that could help safety net organizations compete for scarce clinical talent.

The benefits of ScopeAI go beyond volume alone. This technology empowers a single care team to deliver primary care, addiction medicine, and mental health without referral handoffs — which eliminates the structural leakage that fragments care for high-need populations. ScopeAI supports providers as they navigate across these areas, building confidence in their treatment decision-making, while operating well within their scope of practice.

For Technology Vendors: What "Healthcare AI" Actually Means

The Los Angeles program offers lessons for technology companies building AI solutions for healthcare. Most healthcare AI focuses on narrow use cases: reading imaging studies, predicting readmission risk, optimizing scheduling. These applications have value, but they don't fundamentally change the physics of care delivery.

ScopeAI works because it's built as a full-stack solution integrated throughout the care delivery process. It doesn't just transcribe—it prompts, extracts, documents, coordinates. It doesn't just serve physicians—it enables CHWs, counselors, and medical assistants to perform tasks that would present a challenge without AI support. It doesn't bolt onto existing workflows—it enables entirely new care models.

The lesson: healthcare AI that matters must be designed around care delivery workflows, not just added to existing processes. It must enable team-based care models that expand the effective capacity of scarce clinical resources. And it must integrate deeply enough that it becomes infrastructure, not just another tool competing for attention in already-complex workflows.

Conclusion: Proof of Principle, Promise of Scale

The question facing American healthcare is not whether we'll use AI—that ship has sailed, with 66% of physicians already using AI tools. The question is whether AI will actually solve the fundamental access and capacity challenges facing the safety net, or whether it will simply automate documentation while underserved populations remain underserved.

The Los Angeles street medicine program provides an answer:

 AI-powered care can work, and it can work for the populations who need it most:

63% of patients see a provider by their first day in a population where multi-week waits are typical

67% retention at six months in a population where loss to follow-up is usually assumed

1,300 patients treated for Substance Use Disorders, with 60% initiated on Medication-Assisted Treatment like buprenorphine, which reduce overdose deaths by 50%

31% reduction in emergency department visits, with peak utilization reduced by 46%

Providers serving 2-3x more patients with higher quality documentation and more time coordinating care per patient

These aren't projections or pilot results. This is operational reality, sustained over more than two years, serving 1,500 actively enrolled patients out of 5,500 touched. Since launching in March 2023, the program has demonstrated both clinical effectiveness and operational scalability.

If AI can achieve this on the streets of LA—with patients who have no stable housing, complex trauma, and overlapping needs—imagine what it can achieve in traditional clinical settings with basic infrastructure advantages.

The implications for health systems facing Medicaid cuts are profound. The traditional response to budget pressure—cut services, reduce staff, see fewer patients—moves in the wrong direction. The AI-enabled response—redesign workflows, multiply provider capacity, serve more patients more efficiently—offers a path to maintain access and improve outcomes despite financial constraints.

But realizing this promise requires more than buying AI software. It requires:

• Workflow redesign around AI capabilities, not just adding AI to existing processes

• Team-based care models that enable support staff to perform clinical tasks with AI assistance and provider oversight

• Full-stack integration so AI connects seamlessly with operations and care delivery

• Commitment to serving vulnerable populations and measuring outcomes that matter to them

• Policy and payment reform that supports proactive, team-based, AI-enabled care

The Los Angeles program is a proof of principle. The next chapter is scaling these lessons across the safety net system—from the streets to the clinics, from the margins to the mainstream.

The path forward for American healthcare runs through its most vulnerable populations. Akido has demonstrated that AI-powered care can work in the most challenging environment, with the most complex patients, delivering outcomes that matter. With 1,500 individuals engaged in ongoing comprehensive care and over 5,500 patients served since launch, the Akido's street medicine program provides concrete evidence of AI's potential to transform safety net care delivery. The question now is not whether this model can scale—the technology exists, the clinical protocols work, the outcomes speak for themselves.

The model works. The opportunity now is to transform the healthcare system by bringing it to scale.

---

REFERENCES

1. Association of American Medical Colleges. The Complexities of Physician Supply and Demand: Projections From 2021 to 2036. Washington, DC: AAMC; 2023.

2. OpenAI. AI as a Healthcare Ally: How Americans Are Navigating the System with ChatGPT. San Francisco, CA: OpenAI; January 2026.

3. Sordo L, Barrio G, Bravo MJ, et al. Mortality risk during and after opioid substitution treatment: systematic review and meta-analysis of cohort studies. BMJ. 2017;357:j1550.

4. Sinsky C, Colligan L, Li L, et al. Allocation of physician time in ambulatory practice: a time and motion study in 4 specialties. Ann Intern Med. 2016;165(11):753-760.

5. Kaiser Family Foundation. Medicaid-to-Medicare Fee Index. Menlo Park, CA: KFF; 2021.

6. California Health Care Foundation. California Physicians' Acceptance of New Patients. Oakland, CA: CHCF; 2023.

7. Centers for Disease Control and Prevention. Emergency department visits among persons experiencing homelessness. MMWR Morb Mortal Wkly Rep. 2023;72(42).

Note: All Akido program outcome metrics reflect internal program data collected March 2023–February 2026.