Allow doctors
be doctors
Current ambient AI assistants, which gained popularity in 2023, are already able to record, organize and summarize patient encounters in real time. This frees doctors from the time-consuming process of writing notes, allowing them to fully engage with their patients. “For complex patients, it may take up to 45 minutes to complete documentation. At the end of the visit, I click, and NABLA produces a concise, carefully worded record of what happened,” says Lee, who places the accuracy of the NABLA system in the “high 90s” in percentile. The physician is always responsible for reviewing and signing off on the final record.
“For complex patients, it can take up to 45 minutes to complete documentation. NABLA makes this task infinitely better and allows me to give each patient my full, undivided attention. At the end of the visit, I click, and NABLA produces a concise, carefully prepared record of what happened.”
Dr. Ed Lee, chief medical officer at Naples
This type of constant patient engagement can lead to better eye contact and higher-quality interaction. For example, clinicians tend to verbalize their thought process more when there is an alternative notation during a patient assessment. “We initially thought patients would be anxious about listening to an AI device, but they are actually very excited,” says Alexandre Lebrun, co-founder and CEO of Nabla. “They get the full attention of their doctor during the visit, and they love when they hear the technical language because they feel like they are getting better care.”
According to LeBrun, Nabla’s system can further support doctors by automating advance planning, reviewing and organizing patient information in the electronic health record before an appointment, and coding medical data for use in areas such as billing. Nabla has also expanded its platform with built-in dictation capability, bringing clinicians closer to a unified experience. These types of AI assistant tasks can help streamline and enhance clinical workflow and contribute to reducing institutional administrative costs.
The promise
Amnesty International agent
Agentic AI, which companies like Nabla are currently working to integrate into their systems, promises to take the success of existing AI assistants another step forward. LeBrun envisions a future in which doctors interact with an agent platform that connects to all the tools they already use and simplifies multi-step interactions, such as reading patient data, working within the electronic health record, and adapting to workflows in real time.
“Instead of forcing doctors and nurses to click through dozens of separate systems, our platform will provide dedicated, customizable and composable agents that turn separate tools into one continuous workflow,” says Lebrun.
“Imagine a cardiologist preparing for his morning clinic. After a few voice commands to guide the system, one client pulls the latest vitals, lab results, and imaging reports from the electronic health record, another creates a clear patient summary, and a third flags an echocardiogram for missed follow-up. All before the patient even enters the room,” LeBron explains.
“Instead of forcing doctors and nurses to click through dozens of separate systems, our platform will provide specialized, customizable and composable AI agents that turn separate tools into one continuous workflow.”
Alexandre Lebrun, Co-Founder and CEO of Nabla
Lee says the near-term scope of agent AI includes standardized, protocolized non-clinical tasks, but he sees promise in areas such as treatment options and other types of clinical decision support, where AI can safely work with doctors always “in the loop.”
To get to this point, education is essential, Lee says. “The beauty of medicine is that it is a lifelong learning process,” he explains. “It’s not just about learning about the science behind drugs, diagnostics and treatments; it’s about adapting to using new tools that will ultimately improve the care of the patients you treat.”
“We need to start with the basics of AI,” Lee says, “and make sure everyone understands what it is and how it works. Not how it is programmed, but more about what it can do, what it can’t do, the risks and pitfalls, and then understand where it fits best in patient care.”
He adds that leadership must look forward strategically and ensure the entire organization is moving forward in its use and understanding of AI. “Part of this journey is engaging frontline users to be part of the process, co-designing whenever possible, and piloting new solutions so the organization can learn,” Lee says. Additionally, he says, “A culture of inclusivity, authenticity, and transparency must be in place so that you can be best positioned to succeed in transformative efforts like integrating and embedding agentic AI into the ecosystem.”
“Part of this journey is engaging frontline users to be part of the process, co-designing whenever possible and piloting new solutions so the organization can learn.”
Dr. Ed Lee, chief medical officer at Naples
Integrate safely
In the workflow
Applying AI to high-risk sectors such as healthcare requires finding a delicate balance between productivity on the one hand, and accuracy on the other. “Trust is everything in medicine,” Lebron says. “Earning this trust means giving doctors confidence through accuracy, transparency, and respect for their expertise.” Nabla uses techniques such as adversarial training models to verify results, and by default relies on conservative responses. “We’re working on improving accuracy,” says Lebrun. “If we have even a slight doubt, we prefer to remove something from the output by default.”
“Trust is everything in medicine. Earning that trust means giving doctors confidence through accuracy, transparency, and respect for their expertise.”
Alexandre Lebrun, Co-Founder and CEO of Nabla
New tools must also intertwine with existing workflows and platforms to avoid adding additional complexity for clinicians. “Any product can look great, but if it doesn’t fit well into your existing workflow, it’s almost useless,” Lebron says.
In sectors like customer service, it’s easy to create a new interface or platform, but this approach isn’t possible — or desirable — in healthcare. “It’s a complex web of dependencies with many workflows and processes,” says Lebrun. “Everyone wants to get rid of these things, but it’s not possible because you would need to change everything at once.” Effective AI methods offer great promise for sectors like healthcare because they can “optimize the process without eliminating legacy infrastructure,” explains Lebrun.
By simplifying complex systems, automating routine tasks, and taking on more time-consuming administrative workloads, agentic AI holds great promise in augmenting peripheral AI assistants. Ultimately, the potential of technology is not to make medical decisions or replace doctors, but to support healthcare workers to devote more of their time and attention to their main priority: their patients. “AI should focus on supporting decisions and automating everything,” says Lebrun. “The first role of AI is to put doctors back in the situation where they make medical decisions.”
Discover more ideas from Nabla here.
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by the editorial staff of MIT Technology Review. This content was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes writing surveys and collecting data for surveys. The AI tools that may have been used were limited to secondary productions that passed comprehensive human review.
