HUDDLED WITH

HEALTHCARE HUDDLE

The Future of AI in Medicine: Meet the Agents

Every day I’m in the hospital or clinic, I catch myself thinking: how could this be easier?

Usually, the answer has something to do with AI. Lately, I’ve been especially interested in AI agents—basically, smart assistants that take annoying, repetitive tasks off our plates.

In this article, I’ll break down what agentic AI actually is, walk through how Sully.ai is building it the right way, and share my take on how this kind of tech impacts patients, doctors, and health systems.

What Are AI Agents, Anyway?

Most people think of AI in healthcare as one big brain—one model answering questions or transcribing notes. But when we—or, at least, I—think about the future of AI, I’m not thinking about one all-knowing model. Rather, it’s a team of AI agents—each built to handle a specific job.

I’ve been relating these AI agents to a pit crew in a Formula 1 race.

Each crew member has a defined role:

• One changes tires

• Another refuels

• Another checks the engine.

Individually, they’re focused specialists. But connected together, they make the system faster, safer, and more efficient.

That’s what agentic AI is bringing to medicine. Each agent is optimized for a specific part of the clinical workflow.

Behind the scenes, these agents are powered by large language models (LLMs), clinical databases, automation rules, and integrations with your EHR. When they’re designed well, they hand off tasks fluidly.

Sully.ai is one of the few companies building these agents, or pit crews, the right way.

The Deets: Sully.ai

Sully.ai offers a modular system of AI agents designed to assist across every stage of the clinical workflow. A “modular system” just means each AI agent works independently but can plug into a larger workflow, allowing clinics to adopt and scale only what they need.

The Agents

Sully’s platform provides this end-to-end ecosystem of agents that span the entire clinical interaction:

• Check-in Agent

• Receptionist Agent

• Scribe Agent

• Medical Coder

• Nurse Agent

• Medical Consultant

Now, I want you to close your eyes and just imagine what these agents can do to help you.

Pre-visit, a Check-in Agent verifies insurance and collects patient data, while a Receptionist Agent handles scheduling and inbound triage. During the visit, the Scribe Agent transcribes and drafts documentation in real-time at over 98% accuracy, while Interpreter Agent supports multilingual conversations across more than 20 languages. A Coder Agent assigns the correct ICD and CPT codes, and a Medical Consultant provides real-time treatment suggestions, medication information, and reference-backed insights, providing embedded clinician support throughout the care process. Post-visit, the Nurse Agent automates follow-up tasks: ordering labs, placing referrals, sending prescriptions, and generating draft discharge instructions.

Boom. Done.

X-factors

Sully is rethinking how AI can support their physicians through Clinician Support 2.0. 

While many tools claim to assist clinicians, they often introduce additional complexity. More clicks, more tabs, more systems to learn outside of established workflows. Sully’s approach is different. Their agents act as collaborators, designed to anticipate and automate info physicians need now. 

This includes features like: 

• Automated chart review

• Real-time differential suggestions 

• Treatment plan drafts

• Meds and prescription insights

• Drug pricing comparisons

• Safety checks and alerts

These processes all happen within the workflow. No need to bounce between apps or PDFs. It’s all integrated directly into the EHR, eliminating the need for copying and pasting. Orders and instructions are posted automatically. 

Outcomes

Of course, clinical AI always raises the question: does it actually work in the field? Early data from Sully suggests the answer is yes.

• To date, their agents have completed over 2.4 million medical tasks.

• Clinics using Sully have reported an 85% reduction in onboarding time for new clinicians and a 100% physician adoption rate—no small feat in an industry often resistant to workflow change.

• One clinic reported an 11.2% increase in revenue within the first month of using Sully, largely due to more accurate documentation and fewer billing errors.

• On average, Sully saves 2.8 hours per clinician per day, and over the past three months, they’ve maintained a 0% customer churn rate.

Impressive results, sure. But I’m less interested in metrics and more interested in impact.

Dashevsky Dissection

When I think about the future of AI in clinical practice, I’m focused on impact—on patients, on physicians, and on the operational systems that connect them. Sully’s agent-based model, while still early in its deployment, is one of the more promising attempts I’ve seen to create real, scalable change.

From the patient perspective, the value is clear. Clinics using Sully have reported over a 20% increase in visit capacity, which helps address a growing and well-documented problem: long wait times for appointments. In a recent piece, I explored why patients wait months to be seen and how much of that delay stems from inefficient workflows and administrative drag. Agentic automation—particularly in pre-visit intake and scheduling—has the potential to create meaningful throughput gains without compromising quality.

During the visit, patients benefit from something much harder to measure: presence. When documentation is handled by an AI scribe, and orders and labs are queued up in real time, physicians can re-engage with the patient in front of them. And with agents like Sully’s Medical Interpreter, language barriers—long a source of inequity in care—are no longer a limiting factor.

For physicians, the value prop is equally strong. Across many of the workflows I’ve written about—whether it’s improving discharge summaries, enhancing after-visit instructions, or managing complex patients under time pressure—AI has already shown its ability to streamline redundant tasks and reduce cognitive burden. What Sully does differently is unify those improvements into a single platform. Their agents cover the entire clinical journey, from check-in to follow-up, eliminating the need for multiple siloed tools and manual workarounds.

That said, I don’t view any of this as a silver bullet. There are areas where I’d like to see Sully push further. Specialty-specific use cases—particularly in high-acuity environments like inpatient medicine, emergency departments, or subspecialties like oncology and pulmonology—remain relatively underdeveloped. These are areas where the documentation burden is even heavier, and where agentic support could have an outsize impact.

Overall, Sully isn’t trying to change the things we do as clinicians, but is changing how we get those things done. I believe they’ll help restore the time, space, and clarity we clinicians have been missing.

🔒 PREMIUM CONTENT THIS WEEK

What you’re missing

While you’re reading this, Huddle+ members are diving deeper into the topics that matter most.

Here’s what they saw this week:

• The Chaos Inside RFK Jr.’s HHS Takeover: What happens when political ideology meets public health? I break down the budget cuts, agency restructuring, and potential fallout from RFK Jr.’s first six months at the helm of HHS. Spoiler: it’s a warning sign for science, safety, and the future of regulation.

• How AI Can Help Doctors Manage Complex Patients Faster (Sponsored by Navina): Managing complex patients is time-consuming—and AI might be able to help bystreamlining patient data, reducing prep time, and potentially improving outcomes by giving clinicians a more complete picture, faster.

• How Healthcare Really Works: Launching in two weeks: a new HuddleU course that unpacks the hidden forces shaping healthcare—hospital finances, insurance games, drug pricing, and more. (Access via The Huddle Community).

If you’re serious about understanding the business behind the practice—and want to grow your impact—upgrade here.