An AI payments platform, built from scratch
The problem
A health tech platform serving providers across the country saw a structural gap in how patient AR was being managed. Traditional recovery and AR workflows were built for the world before patients carried high deductibles and the volume of small balances they now do. A meaningful share of patient balances had become too small for traditional recovery to economically pursue, so providers were writing them off. But the underlying problem was bigger than recovery. The entire patient-AR motion (payment plans, communication, offers, analytics) was still being run on infrastructure designed for paper, not for AI-native operations.
The engagement
Build an AI-native AR management platform that addresses the full patient-AR motion: self-service settlement, intelligent payment-plan tailoring, optimized communication cadence, and analytics-driven offer optimization. Lean, compliant by default, economically viable at small balances where status quo had broken down. Designed for one vertical first, architected for many.
What we did
We built the platform full-stack and end to end, with AI woven through every layer rather than bolted on after.
The patient-facing surface is a no-login settlement portal that presents personalized payment offers and flexible payment plans tailored to the specific balance and patient context. The offer logic learns from outcomes: what plan length, what discount tier, what monthly amount actually closes on which kind of balance. Multi-channel outreach runs across SendGrid for email and Twilio for SMS and voice, with cadence optimized per patient based on engagement signals, and tuned automatically to consent and contact preferences. Right channel, right moment, no fatigue.
For providers, a portfolio dashboard tracks outcomes at the portfolio level and the account level, with analytics built in to identify which offer strategies are working, which patient cohorts respond best, and where to push next. The dashboard is the operating surface for the AR motion, not a static report.
Compliance is built into every flow: disclosures, opt-outs, payment terms, full audit trail. The platform ships compliant by default rather than after. Payment infrastructure runs on Stripe Connect with ACH plans, auto-retry on payment failure, and dispute and hardship flows that route to admin review for human judgment. Every state transition logs an event, so the platform is auditable end to end. Multi-vertical architecture from day one: same platform core, vertical-specific configuration on top. The next vertical extends without a rebuild.
The impact
An AI-native AR management platform built lean and operating in pilot, where status quo had ceded the work to write-offs. The first provider was onboarded with $113K under management in the pilot cohort, against a status-quo recovery rate of effectively zero at that balance tier. Payment plans tailored by AI rather than rigid templates. Communication cadence optimized per patient rather than blasted. Provider analytics that show what's working and where to lean in. And the multi-vertical architecture extends to the next use case without a rebuild. Compliance, payments, communications, analytics, and audit all run on one substrate.