AI conversation bots are available 24/7, filling gaps in patient education, adherence, and follow-up care.

By Sree Roy

For years, cognitive psychologist Dan Gartenberg, PhD, struggled with the sheer volume of inquiries that users of SleepSpace, his sleep wellness app, asked about sleep hygiene, CPAP equipment, and other basics. When he released “Dr. Snooze,” an AI chatbot, into that app just under a year ago, Gartenberg discovered that the bot could do a much better job answering user questions than he ever did.

“I would take weeks to get back to people, so this is much better,” says Gartenberg, SleepSpace CEO. “It handles around 90% of my interactions with users that I used to have to do myself.”

Ironically, Dr. Snooze never sleeps, and being available 24/7 is a huge reason it and other artificial intelligence (AI)-powered voice and chatbots are slipping into the sleep world, helping to scale capacity and access to sleep wellness and sleep medicine.

Fast, Friendly AI Bots

Will Kaigler, founder and CEO of sovaSage, notes that the average CPAP patient receives a massive amount of information on diagnosis day—a time when they are often sleep-deprived and overwhelmed. “That’s not how people learn,” Kaigler says. “People learn by small, very focused, timely information that’s delivered in an easy way to consume, that answers their need at that point, and then perhaps reinforces it down the road.”

This is where AI virtual coaches, like sovaSage’s “Jeanie,” come in. By handling repetitive queries and monitoring patient data in real-time, these tools allow human clinicians to focus on complex problem-solving. Jeanie engages patients at pivotal moments, and when human intervention is necessary, escalates to a live coach. Jeanie’s training set includes sleep equipment manuals and manufacturer videos.

Jeanie can swiftly wade through, say, a 10-minute video on CPAP cleaning, to instruct a patient who experiences headaches on how to clean their exhalation port specifically.

Similarly, SleepSpace’s Dr. Snooze leverages a body of knowledge that includes peer-reviewed journal articles and manuals on cognitive behavioral therapy for insomnia (CBT-I). It even houses information on dream interpretation, a popular feature for a subset of power users who like to ask it what the meaning behind their dreams.

AI-Powered CPAP Troubleshooting

A significant portion of patient attrition in CPAP therapy occurs during the first few weeks of treatment.

Kaigler identifies anxiety as one of the most common CPAP troubleshooting issues handled by Jeanie. “The classic first step in helping a patient overcome anxiety is to say, ‘Try using your PAP therapy while you’re watching TV, relaxing, get used to the pressure and then try it again at night,’” he says. This type of behavioral coaching does not require a licensed respiratory therapist for the initial suggestion and, in many cases, can be handled by Jeanie.

Jeanie can also troubleshoot common equipment errors. For instance, showing a patient how to adjust humidifier settings to alleviate morning dry throat. Kaigler shares another common scenario: a patient receives a new mask but fails to remove a plastic piece from the old mask left in the tube. Jeanie can often identify this through guided troubleshooting but is also designed to flag a human coach if the digital interaction doesn’t resolve the issue.

Insomnia Assistance

For insomnia therapy, where CBT-I is the gold standard, AI is being positioned less as a replacement and more as a force multiplier. Access remains a persistent bottleneck—there are simply not enough trained clinicians to deliver live CBT-I at scale—and developers are explicitly targeting that gap. SleepSpace research presented at SLEEP 2024 suggests that chatbot-supported digital CBT-I may improve adherence and short-term sleep outcomes compared to standard digital programs alone.¹

Gartenberg’s vision for SleepSpace reflects a hybrid model. Patients could complete a structured CBT-I program with a clinician while using the app in parallel for reinforcement, and then continue engaging with the AI between visits or during relapses. “It’s providing additional support to the patient, not necessarily replacing a therapist,” Gartenberg says.

This longitudinal support model also addresses a common pain point in insomnia care: maintenance. Rather than patients re-entering live therapy for every setback, AI tools can provide on-demand reinforcement of core behavioral principles. For example, when users log sleep data, the bot can remind them to get out of bed if they’ve been awake for more than 15 minutes.

On the clinician side, by automating intake elements such as sleep diaries and integrating wearable-derived metrics, AI platforms can offload routine monitoring while surfacing actionable patterns—sleep efficiency, variability, circadian misalignment—for clinician review.

The SleepSpace roadmap includes features such as allowing behavioral therapists to upload their own content, perhaps guided relaxation or hypnotherapy tracks, effectively extending their therapeutic presence beyond live sessions.

“It can solve that access problem to a large degree,” Gartenberg says. “The way that I’ve envisioned this is instead of the therapist seeing, maybe 20 to 30 people per week, they could potentially see many more.”

Stepped Care Model

Rather than replacing live clinicians, many AI bot developers envision a stepped care model. “For certain lower-level issues, the AI can do a pretty good job and has advantages in terms of being accessible 24/7 and very scalable,” Gartenberg says, adding, however, that “it doesn’t have that human empathy touch, which is real.”

For clinicians, this also opens up new avenues for remote patient monitoring. Gartenberg is working toward making SleepSpace HIPAA-compliant so clinicians can patient data through a portal, facilitating treatment and billing under a medical doctor.

At sovaSage, if Jeanie detects that a patient’s apnea-hypopnea index remains high or that a mask leak is persistent despite digital coaching, it escalates the case to a team of experienced coaches or licensed RTs.

“Here’s the key: It can’t just be a pure technology play,” Kaigler says. “It has to be a hybrid where you’re enabling trained, talented coaches and respiratory therapists to intervene and problem solve at the right time. In the end, it’s always going to be people intervening and providing that point of care when the patient most needs it.”