Burnout across healthcare has become a systemic staffing crisis, and behavioral health is feeling it most acutely. In 2022, CDC survey data found that 46% of health workers overall reported feeling burned out often or very often, up from 32% in 2018. Behavioral health workers are experiencing something even more severe: a National Council for Mental Wellbeing survey found that 93% report experiencing burnout, with 62% calling it severe and nearly half saying they’re considering leaving the field. Administrative burden is one of the most consistently cited drivers of that attrition, and it’s compounding at a moment when the workforce can least afford to shrink.
At the same time, a clearer picture is emerging of where agentic AI is actually proving itself: in aiding, not replacing, workers. It’s in the administrative, coordination-heavy work that eats up hours without directly touching a patient: intake, documentation, scheduling, follow-up outreach, and referral tracking.
The Health Management Academy and Bamboo Health found that nearly every surveyed health system is already using GenAI for clinical documentation. Yet, few have translated that into measurable clinical or financial outcomes. That’s also where the outcomes gap lives. Most health systems are still deploying GenAI as a standalone documentation tool rather than embedding it within care delivery workflows that support care teams’ actions. That gap is exactly why adoption hasn’t yet translated into measurable results.
That distinction matters because agentic AI’s proven use case today remains mostly administrative, and care teams already stretched thin need to know how to responsibly integrate these solutions into direct care workflows, not just back-office ones. Care navigation for high-cost, high-need populations is a particularly difficult context in which to do that well. Organizations with real experience in embedding human-led, tech-enabled care navigation have already shown what’s achievable. Bamboo Bridge®’s human-led, tech-enabled model has demonstrated a 20% reduction in utilization among high-risk patients, 42% lower readmission rates, and an 8x faster time to care. Those results didn’t come from AI replacing the care team. They came from technology and human expertise working together, and they now serve as the benchmark that AI-assisted, human-in-the-loop models are working to reach, with clinical judgment still doing the work only people can do.
A coordination layer should excel beyond just serving as a patient outcomes tool. It’s retention infrastructure. When AI absorbs the administrative load that’s driving people out of the field, and human judgment stays exactly where it belongs, organizations get a workforce that can practice at the top of its license instead of drowning in paperwork. As one healthcare leader put it recently, AI’s job here is boring on purpose. It handles the paperwork so people can focus on the human part of the work.
None of this works if AI is asked to replace clinical judgment. The evidence points the other way: 87% of health systems surveyed by The Health Management Academy anticipate scaling GenAI over the next two years, but integration complexity, unclear ROI, and workforce readiness remain the top barriers to getting there. Solving for those barriers means designing AI to support the coordination layer, with humans firmly in the loop, rather than layering another disconnected tool on top of an already fragmented system.
The organizations that get this right won’t be the ones with the most AI. They’ll be the ones that use it to protect the people doing the work, while keeping clinical judgment exactly where it should be: with humans.
Learn more about how Bamboo Health supports coordinated, human-in-the-loop care teams to reduce admin burden or connect with us today to discuss what this could look like for your organization.
