Production agents across claims, RCM, and patient engagement.

The challenge

Apollo's RCM team was scaling claims volume faster than headcount could absorb, and patient-engagement queues across multiple hospitals were stuck on shared inboxes. Any automation had to clear an internal compliance team that wanted line-item explainability on every clinical and financial decision.

How we approached it

Ajuni stood up three agents in sequence inside Apollo's existing Azure tenant: Vaidya on claims adjudication against the HIS, Asha on patient engagement across SMS and WhatsApp, and Veer signing every action into a tamper-evident ledger. The compliance team got a per-decision replay surface before any agent went live.

Outcomes in production

• 3 agents running production traffic across the group
• 84% of patient queries auto-resolved on first touch
• First agent reached production in 9 weeks
• Per-decision audit replay cleared compliance review
• Claims-cycle hours collapsed at every Apollo site

Stack & guardrails

Timeline

1. Week 0–2

   Compliance design

   Audit-replay surface designed with internal compliance team before build.

2. Week 3–5

   Vaidya in shadow

   Claims adjudication ran parallel to humans across two hospitals.

3. Week 6–9

   First agent live

   Vaidya promoted to production; Asha started shadow on patient queues.

4. Week 10+

   Three-agent rollout

   Asha and Veer joined production; daily compliance review cadence.

“We run agents across claims, RCM, and patient engagement. The audit trail is what made our compliance team approve it, every prediction, every change, signed.”