Rasi Bhadramani/istock via Getty

How AI agents scale personalized diabetes care

Salesforce Agentforce allows Precina Health to scale personalized support for type 2 diabetes patients and enable physician access to clinical data across multiple systems.

At Precina Health, a medical clinic that operates in California, Oregon and Texas, the goal is to keep patients' blood sugar levels under control so they stay out of the hospital and ensure that patients feel supported as they restore their health.

John Oberg, the CEO and cofounder of Precina, started the company after his mother was diagnosed with diabetic ketoacidosis, a condition in which the body doesn't produce enough insulin and a buildup of acids called ketones occurs.

For patients, the costs of insulin as well as supplies, such as glucose meters, test strips and syringes, add up. The CDC reports that $1 out of every $4 in the U.S. healthcare system is spent treating diabetes.

Offering virtual-only visits, Precina helps patients better manage A1c levels as well as their overall diabetes care and encourages them to ask pertinent questions about their condition. To personalize and scale this type of support, the organization is using AI agents from Salesforce's Agentforce platform.

Oberg hopes to help a billion people and expand access to personalized diabetes care,  particularly in underserved, rural and low-income areas. With Agentforce, an agentic AI layer that's part of the Salesforce platform, Precina aims to boost outreach and improve the quality of care at scale.

How Agentforce works

At Precina, Agentforce allows both clinicians and patients to interact directly with AI agents. It powers interactions taking place in Precina's mobile app as well as its browser experience. Some Agentforce AI agents are chatbots on Precina's website. Precina staff benefit from this omnichannel experience of accessing Agentforce functionality on any device, the provider said.

Agentforce allows organizations to build, test, deploy, manage and orchestrate AI agents. AI agents in Agentforce perform sales development, clinician coaching and customer service. They recruit patients, much like sales agents, and help book appointments with coaches. AI agents use machine learning and natural language processing to answer questions from potential patients, providing an opportunity for Precina to scale its business.

AI agents supported by large language models create a "much more mature customer interaction," said Madhav Thattai, COO of Agentforce at Salesforce. "You don't feel like you're talking to a robotic bot of some kind. It feels more fluid."

Thattai notes that Agentforce is built on interaction, business process, logic and data layers.

"We want to make sure that the agents have access to all that data so that the actual interaction is personalized, specific, accurate, and ultimately going to deliver on whatever outcome that patient or that customer wants whenever they're interacting with Precina," Thattai said. "So that data layer is really important."

How Precina uses AI agents

Not only does Agentforce help Precina recruit patients and book appointments, but it also provides automation capabilities that allows the organization to resolve complicated issues, such as fragmented systems. Precina wanted to integrate its systems so it could gain access to patient data from lab results, prescriptions and telehealth records. Agentforce unites data from EHRs, lab reports and Health Cloud and Data Cloud.

In Agentforce, Precina clinicians can use a voice copilot to allow providers to access patient charts. An AI tool helps with patient documentation as well as medical record classification and summarization and provides access to inbound records.

The provider also uses its own proprietary AI platform with Salesforce's Sales Coach Agent and Service Cloud to evaluate the well-being of providers as they engage in empathetic patient conversations, Oberg shared. Sales Coach offers providers an AI summary of their performance with the patient.

Salesforce AI agents also handle lead generation for the provider's inbound social channels, according to Precina.  

"In some ways, the lead generation is a growth strategy while you've got a service capability that is really more of a customer experience and making sure that customers are getting a richer experience," Thattai said.

In addition, Precina has created an AI copilot that's programmed to follow specific protocols and help providers read patients' medical charts to prepare for visits. Precina's cofounder and chief medical officer, Dustyn Williams, M.D., programmed the copilot with specific medical protocols and logic reasoning, and providers currently use it.

Training clinicians and paraprofessionals using Agentforce

Agentforce handles some of the onboarding and training for healthcare practitioners and paraprofessionals, like customer service representatives, at Precina. Agentforce provides training with role-playing, therapeutic support calls and clinical visits in the style of sales coaching. It provides guidance on what to do when clinicians get on the phone with patients and ensures that they take the right approach, Thattai said.

During role-playing, a Sales Coach Agent acts as the prospect, asking questions of a Precina customer service rep about the service. The agent then offers an AI summary of the rep's performance.

Agents evaluate and score a provider's pitch to a potential patient. When employees are too busy to perform human training, AI takes over and it's a richer training experience, according to Thattai. Clinicians learn rapport building, while paraprofessionals gain guidance on maintaining HIPAA compliance and clinical accuracy.

"We can use that coach to have them role-play and practice, meeting the patient where they are," Oberg said. "And so, while that's fairly opaque to the patient, the provider feels like they have a resource they can talk to, and it's tirelessly giving them coaching feedback help in a judgment-free environment to improve their clinical skill set, and they can use that as often as they want."

AI agents also work with patients on medication adherence as well as implementing lifestyle changes around diet and exercise. The AI coaching helps providers create meal plans for patients in a couple of minutes rather than taking hours, according to Oberg.

Not only that but the AI agents also provide feedback to providers on whether they appeared empathetic in communications with patients. It might provide feedback on whether the provider requires motivational interviewing.

Future growth with AI agents at Precina

In the future, Precina plans to expand the use of Agentforce to include goal setting and habit tracking for patients, and incorporate AI ambient voice agents for providers.

Longer term, Oberg also envisions patients and volunteers, or peer counselors, inviting AI agents into their conversations to provide relevant clinical interventions when medically necessary. Peer counselors are not medical professionals, but they can offer emotional support to patients. The support from AI agents and peer counselors could allow Precina therapists to achieve broader reach in the community. Oberg estimates that AI agents combined with peer counseling could reduce the cost of providing mental healthcare by 90% or more.

"These are some of the ideas that can credibly be talked about today that five years ago we couldn't credibly talk about that," Oberg said.

Though the prospects for AI agents are exciting, Thattai added that health results are most important when adopting AI agents.

"What really matters is, what outcome are you driving for your customers?" Thattai said. "And so these types of agentic experiences that Precina is building now are tied to that outcome."

Brian T. Horowitz started covering health IT news in 2010 and the tech beat overall in 1996.

Dig Deeper on Artificial intelligence in healthcare