For creators and clinicians in the health sector, the era of passive digital tools is ending. The pressure on global health systems is mounting, driven by chronic underfunding, recruitment bottlenecks, and a surging demand from an aging population. The World Health Organization has warned that the global shortfall in health workers will hit 11 million by 2030. In response, more than two-thirds of healthcare providers have already integrated agentic AI into their operations, according to KPMG.
Unlike previous iterations of software that merely digitised paperwork, these autonomous agents are designed to handle complex back-office tasks, collaborate with medical teams, and triage patients. The goal is to reduce the cognitive load on staff and improve care quality just as the supply of human workers dwindles.
A shift from administration to augmentation
Historically, digitalisation in healthcare has often backfired. Staff frequently blame outdated technology for increasing administrative burdens rather than alleviating them. For instance, while US patient data was migrated to electronic health records (EHRs) in the early 2000s, the information remains fragmented and reliant on manual entry.
Ashis Barad, MD, chief digital and technology officer at the Hospital for Special Surgery (HSS) in New York, notes that early telehealth tools and remote monitors failed to replicate in-person care or earn patient trust. He argues that agentic AI is fundamentally different because it does not rely on rigid frameworks or manual inputs for edge cases.
Instead, these agents can navigate nuanced scenarios, make autonomous decisions, and iterate over time. As Dr. Barad explains: “Agentic AI takes your workflow and collapses it, augments it, supercharges it, and makes it more performant.”
Real-world impact at HSS
At HSS, AI agents have already transformed backend operations. Previously, processing insurance claims involved weeks of work by both HSS staff and third-party contractors. Now, AI agents process 1,100 claims per month. Since implementation nine months ago, the appeals stage has shrunk from 45 minutes to five, and the success rate has jumped from 65% to 100%. The hospital now handles all claims in-house.
This success has led to a patient-facing triage service developed in collaboration with enterprise agentic AI developer Ema Unlimited. Accessible 24/7 via web, text, or phone, the service uses conversational AI to clarify patient conditions and book appointments with the most suitable clinician, factoring in location, insurance, and availability.
“It completes the whole loop,” says Dr. Barad. The system is trained on HSS’s entire knowledge base, allowing patients to access highly specialist knowledge directly. However, given the high stakes, the service includes strict safeguards. Sensitive or complex cases are escalated to human specialists, and every AI decision is auditable. Ema states that this approach balances efficient automation with patient-first safety.
Systems-level transformation
As the technology proliferates, providers must embed guardrails into their systems. At HSS, all technology decisions are filtered through an AI subcommittee co-chaired by Dr. Barad and a senior nursing executive. Agents touching patient care face far more rigorous scrutiny than those handling backend processes.
Dr. Barad is also establishing a dedicated AI lab on the HSS campus to democratise access. Open to all staff, it offers training and classes to help employees build and understand AI agents. This mirrors research by Deloitte, which found that leading adopters are redesigning end-to-end workflows rather than relying on narrow, isolated use cases.
Dr. Barad emphasises that agentic AI is a general-purpose technology, analogous to electricity, rather than a collection of specific use cases. To realise its value, providers must build a unified data strategy. Currently, data is often siloed across departments with legacy IT systems, and metrics lack standardised definitions.
For example, Dr. Barad notes that hospitals often define “time to start surgery” differently, hindering AI agents from assimilating tacit knowledge. By improving data interoperability at HSS, patient-facing agents can now draw from a patient’s full clinical history and current symptoms to decide if a situation requires escalation before notifying the correct specialist.
The future of white-glove care
Dr. Barad envisions a future where 90% of non-clinical tasks are administered by AI agents, freeing clinicians to focus on what he calls “white-glove work”—the most complex, specialised, and sensitive cases. This optimism is shared by the industry; 84% of providers feel comfortable handing decision-making over to AI agents, per KPMG research.
“We’re spending so much time on keyboards and computers right now that we’re actually not doing what we should be doing,” says Dr. Barad. “This is going to rehumanize health care.”
Key takeaways
- Agentic AI moves beyond simple digitisation to autonomously handle complex, nuanced workflows, reducing administrative burden on clinicians.
- Successful implementation requires a unified data strategy and interoperability to allow agents to access comprehensive patient histories and make informed decisions.
- Providers are prioritising safety by keeping humans in the loop for sensitive scenarios and establishing rigorous governance committees to oversee AI deployment.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.
