Picture a referral coordinator at a community care desk. A veteran needs a cardiology appointment. The coordinator has the referral in one system, the provider network in another, availability in a third, and eligibility rules in a fourth. She toggles, copies, calls, waits on hold, and documents it all twice. The patient waits. On average, across the system she works in, that wait to get an appointment scheduled has been 28 days.
Deals that size generate a lot of headlines about AI transforming medicine. But read the announcement closely and the most telling detail is not a clinical breakthrough. It is a scheduling number. Building on the VA’s Unified Patient Scheduling initiative, which spans community care scheduling across more than 40,000 provider services nationwide, the stated goal is to reduce the average time to schedule an appointment from 28 days to minutes once fully deployed.
To be clear, that is a target, not a result. The deployment is just beginning. But targets are strategy made visible, and this one tells you exactly where the most scrutinized health system in the country believes agentic AI pays off first.
The ROI is hiding in the boring work
There is a persistent assumption that the value of AI in healthcare lives at the clinical frontier: diagnosis, treatment selection, drug discovery. Those frontiers matter. But they are also the hardest places to deploy, the slowest to validate, and the riskiest to get wrong.
Administrative friction is the opposite on every axis. Scheduling, intake, triage routing, benefits verification, eligibility checks, referral management: this work is high volume, measurable, and widely resented by the people doing it. It spans multiple systems by definition, which is precisely why humans spend so much time on it and precisely what agents are built to traverse. And when an agent gets it wrong, the failure mode is a rescheduled appointment, not a clinical event. You can supervise it, measure it, and improve it in production.
That is why the VA agreement leads with 24/7 virtual contact center support for patient triage, intake, and care coordination, with agents that surface knowledge during live calls, route cases instantly, and automate benefits verification. Nothing on that list replaces a clinician. Everything on that list gives one back their time. As Missionforce & Government Cloud, Salesforce, CEO Kendall Collins put it, every minute an employee spends navigating disconnected systems is a minute not spent serving a veteran. Swap “veteran” for “patient” or “member” and you have the business case for nearly every health organization we work with.
Why scheduling is the perfect first target
Three ingredients make 28 days to minutes credible as a goal.
First, the outcome is unambiguous. Days to appointment is a metric every board already sees and every patient already feels. When your first use case moves a number like that, it’s hard to argue whether the program is working.
Second, the workflow crosses systems. Scheduling data is scattered across platforms, provider directories, eligibility engines, and the EHR, and no amount of staffing solves a connectivity problem. That is why the VA deal pairs its agents with a data layer that pulls legacy platforms into a single trusted view of each veteran. The agent is the visible part; the unified data foundation is the load bearing part.
Third, trust was a requirement, not a feature. The program required a HIPAA-ready architecture and authorization to federal security standards. If governance can be satisfied at that altitude, compliance becomes a design requirement rather than a blocker.
Translating this to your organization
The pattern translates directly. For providers, it is referral management, prior authorization, and patient intake. For payors, benefits verification, member triage, and claims status. The strongest first use cases often fit the VA scheduling profile: a hated, high volume, cross system administrative workflow with a metric leadership already watches.
Start where friction is measurable, data can be unified for a narrow slice, and an imperfect answer is an inconvenience, not a harm. Prove the number moved, then expand. All new use cases will inherit the data foundation, governance, and trust the first one earned.
If 28 days sounds familiar, let’s talk
Every healthcare provider or payer has its own 28 day number: the referral that takes three weeks, the prior auth that takes eleven calls. You already know yours. What is usually missing is clarity on which workflow to automate first, what data has to be connected and governed, and how to get compliance to yes before the pilot.
That is the work we do. Lirik helps healthcare and life sciences organizations put governed agents into production on Salesforce, from the first use case to the data foundation beneath it. Bring us your 28 day number, and we will be honest about what it takes to turn it into minutes.
Vertex Tower, Plot no-
C-33, 4th Floor, Phase 2, Industrial Area, Sector 62, Noida, Uttar Pradesh, 201309