Raintree Launches Agentic RCM — An AI Front Desk and Revenue Cycle That Works Without You

At TherapyCon '26, Raintree debuted Agentic RCM™ and Agentic PX™ — purpose-built AI agents for rehab therapy's hardest billing and scheduling work. An AI eligibility specialist navigates payer portals. An AI receptionist answers 98% of calls. The "Invisible EMR" just became an autonomous one.

The Setup: Rehab Therapy Has a Math Problem

57% of rehab therapy practices say they cannot meet demand in their local market. Physical therapist supply is running 8.2% short of what's needed just to maintain baseline care for a growing population. Inflation-adjusted PT pay has fallen roughly 10% over the past decade — which means fewer clinicians entering the field, and too many leaving it.

Raintree CEO Nick Hedges opened TherapyCon '26 with a clear thesis: the profession isn't being hollowed out by AI. It's being choked by administrative burden and thin practice economics. "The threat was never too few jobs — it is too few clinicians." Every product announced this week attacks one of those two problems.

Agentic PX™: The AI Front Desk

The first Agentic PX™ agent is named Marcy — an AI front-desk receptionist that runs 24/7, 365 days a year. Hedges opened the keynote by placing a live call to Marcy from the stage. She's bilingual (English and Spanish), handles complex multi-appointment plan-of-care scheduling, captures insurance details, answers authorization and copay questions, and calls patients who miss appointments.

The economics behind this are straightforward: roughly 70% of callers who hit voicemail never leave a message. They book with a competitor. Marcy's targets include:

• 98% of calls answered
• $90K–$185K annual revenue recaptured from missed calls and abandoned bookings
• 30% reduction in no-shows
• 12% reactivation of dormant patients

SchedulerIQ™, also debuting today, cuts multi-appointment plan-of-care booking from 10–12 minutes to one, reduces scheduling time by 60%, and removes roughly 10 times the front-desk clicks. The time savings alone are meaningful for high-volume rehab practices that book out weeks of POC appointments at intake.

Agentic RCM™: Lucy, the AI Eligibility Specialist

The more significant announcement for RCM professionals is Agentic RCM™. The suite is structured in three layers: the billing team sets goals and works complex escalations; an orchestration layer routes each claim to the right agent; and specialized AI agents each own a narrow component of the revenue cycle.

The first agent is Lucy — an AI eligibility verification specialist. What makes Lucy different from a basic eligibility API call:

• Navigates payer phone trees, including when they change
• Logs into and searches payer portals directly
• Processes clearinghouse data and holds complex benefits conversations
• Applies learned payer-specific patterns — known group IDs, policy quirks, routing logic
• Double-confirms what she's told before writing it back to the EMR

Raintree's value targets for Lucy: 91% reduction in per-verification touch time, a 90% reduction in visits rendered without confirmed coverage, and $2.20–$2.80 more recovered per patient visit.

This is the right place to start. Eligibility failure sits at the front of virtually every denial pattern — wrong payer, lapsed coverage, benefits not confirmed before service. Fixing verification autonomously doesn't just save staff time; it prevents the downstream denial volume that makes RCM expensive.

The Documentation Layer: NoteIQ™

On the clinical side, NoteIQ™ completes the documentation picture. Initial evaluation notes that took 15–45 minutes pre-AI now close in under two minutes. Against other AI-scribe tools, clinicians using NoteIQ™ sign initial evaluations 40% faster, capture over $5.00 more revenue per visit, and see after-hours documentation fall from 10–20% of visits to under 3%.

That $5+ per visit revenue lift deserves attention. It's not about upcoding — it's about documentation specificity. When AI captures the clinical picture accurately, the resulting note supports the right code without a coder having to extrapolate or query the clinician. In rehab therapy, where functional status and progress drive reimbursement, documentation quality is directly tied to what gets paid.

The Infrastructure: Spike Technologies Acquisition

The foundation for all of this is Raintree's July 2026 acquisition of Spike Technologies, an agentic AI voice company. The deal expanded Raintree's AI Center of Excellence fivefold and brought in machine learning PhDs with backgrounds from Google, Bloomberg, Oracle, and Amazon One Medical. Spike's models now power the voice-layer agents like Marcy and Lucy's payer phone tree navigation.

Raintree's AI stack is trained on more than 7.3 billion data points across clinical, front office, and revenue cycle workflows. Every model is embedded in a specific workflow with its own guardrails — not deployed as a general-purpose chatbot on top of a legacy system. That distinction matters when you're navigating payer portals and committing decisions back to the EMR.

The Bottom Line

Raintree is making a specific bet: rehab therapy's margin problem is an administrative problem, and autonomous agents are the fix. Not AI-assisted billing. Not AI that flags things for a human to act on. Agents that complete the work — phone calls, portal logins, eligibility confirmation, documentation — and hand off only the exceptions.

The Agentic RCM model Raintree is building is structurally similar to what Commure is doing in acute/enterprise and what Candid Health is doing in institutional. The difference is specialization. Raintree's models are trained specifically on rehab therapy's payer mix, care plans, and coding patterns — which means better accuracy where generic RCM AI misses.

For RCM leaders in PT, OT, and speech therapy: the 2027 roadmap — auth, claims, denials, payments — is the one to watch. If Lucy performs at the stated benchmarks, the rest of the suite becomes the compelling sell.