Rehumanizing global health care with agentic AI

The global healthcare sector confronts an unprecedented operational crisis that threatens the stability of medical systems worldwide. As of 2024, health systems across developed and developing economies face a convergence of structural pressures: decades of underinvestment, chronic recruitment difficulties, and surging demand driven by aging populations. The World Health Organization projects a shortfall of 11 million healthcare workers by 2030, a deficit that cannot be addressed through traditional workforce expansion alone. In response to this existential challenge, healthcare providers have begun deploying agentic artificial intelligence at scale, with KPMG research indicating that 68 percent of healthcare organizations have already integrated AI agents into operational workflows. This represents a fundamental shift in how institutions approach the physician shortage and administrative paralysis that has come to characterize modern medicine. The technology is being mobilized not as a supplementary tool but as a core strategy to preserve patient care quality while the human workforce contracts relative to demand.

The trajectory of healthcare digitalization over the past two decades illuminates why agentic AI represents a qualitatively different intervention. Earlier waves of technological adoption, including the mandatory transition to electronic health records in the United States during the early 2000s and subsequent telehealth expansion, delivered disappointing results despite substantial capital investment. These systems relied heavily on manual data entry and human workarounds, creating additional administrative friction rather than eliminating it. The fragmentation of patient data across incompatible EHR platforms persists as a foundational problem, requiring clinicians to navigate multiple systems simultaneously while maintaining paper-based documentation as backup. Telehealth platforms similarly failed to achieve anticipated benefits, removing geographical barriers to access but sacrificing diagnostic nuance and clinical trust. Healthcare staff came to view technology implementations with skepticism, attributing increases in burnout to systems that automated nothing while generating new layers of documentation requirements. Agentic AI addresses this historical failure pattern by introducing autonomous decision-making capacity rather than merely accelerating manual processes. The distinction proves crucial: previous digitalization efforts attempted to computerize existing workflows, whereas agentic systems fundamentally restructure workflows through iterative learning and autonomous execution.

Hospital for Special Surgery, an academic medical center in New York specializing in musculoskeletal health, provides the most concrete evidence of agentic AI's operational impact. The institution deployed AI agents in its insurance claims processing function, a backend process that previously consumed weeks of labor across internal staff and third-party contractors. The results demonstrate measurable transformation: AI agents now process 1,100 claims monthly with substantially improved success metrics. Most notably, the appeals phase has contracted from 45 minutes of human review per claim to five minutes, while the success rate of appeals has improved from 65 percent to 100 percent across the nine-month implementation period. This advancement enabled HSS to internalize claims processing entirely, eliminating dependence on external vendors and achieving what represents a fundamental reimagining of a standard healthcare administrative function. These figures illuminate why healthcare executives view agentic AI not as a speculative technology but as an immediately productive tool capable of transforming institutional economics while reallocating human cognitive resources to higher-value clinical activities.

For practicing clinicians and healthcare administrators navigating immediate operational pressures, agentic AI offers relief from the specific malady that has driven documented increases in physician burnout: administrative overload detached from patient care. The historical digitalization failures created compounding frustration because new systems increased rather than decreased clerical demands. Agentic AI reverses this dynamic by automating entire categories of work that previously consumed clinician attention without improving patient outcomes. Claims processing exemplifies the category: this function generates zero clinical value yet commands significant physician time indirectly through documentation requirements, insurance correspondence, and approval processes. When an AI agent assumes autonomous responsibility for claims adjudication, appeals, and follow-up communication, the cognitive load on clinicians diminishes demonstrably. Ashis Barad, chief digital and technology officer at HSS, characterizes the technology as collapsing and supercharging workflows rather than merely accelerating them. This distinction matters operationally because it suggests that agentic AI can address burnout without requiring additional hiring or increasing patient caseloads, allowing healthcare systems to maintain care quality despite workforce constraints. The immediate relevance lies in the specificity of impact: not theoretical future benefits, but documented reductions in task duration and improvements in process success rates happening within current healthcare operations.

The broader pattern suggested by expanding agentic AI adoption signals a fundamental restructuring of healthcare labor economics and the relationship between human expertise and computational capacity. Rather than attempting to replace physicians or clinical specialists, agentic systems handle the boundary-spanning administrative functions that prevent clinicians from engaging in high-value diagnostic and therapeutic work. This division of labor addresses a specific failure of previous healthcare models: the requirement that credentialed medical professionals spend substantial portions of their time on tasks that require organizational expertise rather than medical expertise. As healthcare systems face the prospect of even greater workforce shortages, the margin for inefficiency compresses to zero. Agentic AI fills that margin by assuming responsibility for complex but non-clinical decision-making, freeing the diminishing number of available clinicians to focus on genuinely clinical activities. The pattern also reveals how healthcare providers have begun abandoning the assumption that digitalization must be physician-centric, instead building systems that serve physicians by removing obstacles rather than requiring behavioral change. This represents a pragmatic acceptance that technology adoption in healthcare succeeds only when it demonstrably improves the working conditions of staff while improving patient outcomes.

Healthcare organizations and technology providers will determine success or failure of this transition through measurable milestones in the coming eighteen months. HSS is continuing expansion of AI agent deployment into non-clinical patient-facing settings, with announcements regarding specific applications expected as implementation progresses. Simultaneously, the broader healthcare technology sector requires demonstration that agentic AI can scale beyond back-office claims processing into more complex clinical domains such as patient triage, preliminary diagnostics, and care coordination across fragmented provider networks. The WHO's 2030 worker shortage projection creates a deadline for implementation, suggesting that healthcare systems must move beyond pilot programs to organizational transformation by 2026 to address the earliest phases of predicted workforce gaps. Industry observers should monitor whether agentic AI adoption correlates with measured reductions in clinician burnout and whether the technology proves capable of maintaining care quality metrics as patient-to-clinician ratios decline. The stakes extend beyond individual healthcare institutions to the viability of healthcare systems themselves: if agentic AI cannot effectively assume administrative labor while human workforce constraints intensify, the fragmentation and access problems already evident will accelerate dramatically. The next two years will determine whether this technology can deliver on its operational promise or whether healthcare systems will face the collapse that current trajectories predict.