Anand Chandrasekaran is the principal architect at Arya Health, a workforce management platform for in-home healthcare. Arya uses AI scheduling agents, plus employee onboarding and credential and compliance management, to take routine admin work off agency office managers. Anand runs Arya's AI architecture. He joined us in Episode 24 (December 12, 2025) alongside Arya's co-founder and CTO, Arun Kalaiselvan.
Arya's entry point is scheduling (20:23). When a nurse calls out sick, an admin can spend two to three hours finding someone who is available, nearby, credentialed and a good fit for the patient. Arya already knows all of that, so AI phones the right caregivers with the open shift and a bonus offer, and the shift gets filled right there.
Anand's part of that is the matching (24:21). Early on, Arya built its own heuristic eligibility system on top of customer data. Arun explained that its output is a tiered list: caregivers are scored per patient, and the top tier gets called first.
we kind of also have our own heuristic based eligibility system that helps us get this system going. And we've been very, very successful in matching caregivers only because of that.
— Anand Chandrasekaran, Episode 24
So from the first, our model has been, like, you can only use the models that's available in our cloud provider because only that is giving us the utmost security.
— Anand Chandrasekaran, Episode 24
That's like the biggest advantage. And also the prompt itself can be very precise. Instead of writing tons and tons of examples, you don't have to do that anymore.
— Anand Chandrasekaran, Episode 24
In the news segment (33:29), Anand defended Cursor, even though Arya dropped it after a six-month pilot one and a half to two years earlier; he said Cursor is now very mature. His argument: in two years most models will be good enough at coding, tooling and cost will decide the winner, and Claude Code is already expensive by comparison.
Sam built an AI start-of-care packet generator with Arya in mind. It turns an unstructured hospital discharge summary into a care plan, a four-week schedule by discipline, matched clinicians and drafted outreach. He got a PRD from ChatGPT, pasted it into Google AI Studio with Gemini Pro, and had it working in about 30 minutes.
Anand said (06:19) it already captured something Arya planned for its fourth version. The one thing off was that it assumed real time, and in practice the process is not. For more, see our AI in healthcare page.
Anand Chandrasekaran is the principal architect at Arya Health, where he leads the AI architecture. He joined Built This Week with Arya's co-founder and CTO, Arun Kalaiselvan, in Episode 24 in December 2025.
Arya Health is a workforce management platform for in-home healthcare. It offers AI scheduling agents, employee onboarding, and credential and compliance management, and it can have AI phone eligible caregivers to fill an open shift.
Arya only uses models available through its cloud providers under contract: Anthropic models through AWS Bedrock and Gemini through Google Vertex AI. Anand also said Arya's AWS and GCP environments talk only over a separate private tunnel.
Anand uses Pydantic AI dependencies, which act as input types, instead of examples and JSON types. He said that removed the need for examples in 90 to 95% of their agent work and cut hallucinations by 40 to 50% for him.
Anand said Arya built its own heuristic eligibility system early on. It scores caregivers for each patient and groups them into tiers, and the highest-scoring tier gets called first when a shift needs filling urgently.
