The live simulation needs WebGL.

The browser couldn't start 3D rendering — but the platform is very real:
AI agents triaging referrals across thousands of SNFs, right now.

This world is our training ground.We specialize in agentic automation of healthcare administration.

Our technology connects providers, payers, and patients so we can automate healthcare holistically — starting with the most complex patients. While our customers only see seamless workflows, behind-the-scenes agents are training in our proprietary reinforcement learning environment to automate all these workflows.

Read the story ↓

4,218 training episodes 250,000+ referrals watching the environment…

drag to orbit · scroll to zoom

The problem — and the prize

Healthcare's back office is a staggering burden. Post-acute care carries the worst of it.

Roughly a quarter of US healthcare spending goes to administration. In skilled nursing, thin margins meet the heaviest paperwork: referrals by fax, eligibility by phone, claims by hand, schedules in spreadsheets, care plans retyped between systems. Whoever automates that work doesn't sell software to the industry — they become its operating layer.

01 — WHY SNFS FIRST

15,000 buildings, one loop

Skilled nursing is concentrated pain: every facility runs the same referral → admission → claim cycle on 1980s tooling. High volume, high stakes, brutally repetitive — perfect ground for agents.

02 — WHY AGENTS

Work done, not workflow software

The industry has bought dashboards for a decade; people still do the work. Our agents do the work itself — reading, deciding, messaging, filing — and are measured on outcomes, not logins.

03 — WHY NOW

The models finally read the fax

Extraction and judgment over messy documents crossed the usable threshold. The bottleneck is no longer the model — it's the environment to train and prove agents in. We built that environment. It powers the world above.

Our moat — the RL environment

Trained in simulation. Proven in production.

Before an agent ever touches a live referral, it has run the job thousands of times in a simulated world of hospitals, facilities, payers, and patients. The world above is real training infrastructure, not a marketing toy.

01 · SIMULATE

Thousands of episodes

Each episode is a full patient journey — referral to discharge home — replayed with real-world messiness: denials, missing prior auths, tight bedboards.

02 · REWARD

Census, revenue, compliance

Agents are rewarded for outcomes across the care journey: beds filled fast, claims paid first-pass, patients home safely. Policy updates follow the reward.

03 · DEPLOY

Live in thousands of SNFs

Only proven policies graduate to production, where they run referrals, census, and RCM across the real network every day.

04 · KEEP LEARNING

Every event is feedback

In production, every referral, claim, payment, denial, and discharge keeps refining the policy — the same loop shown in the simulation.

Under the hood

One pipeline from fax to payment.

INGEST

Everything, everywhere

eFax, Epic, 11+ discharge portals, uploads — even that old fax machine.

EXTRACT

AI reads the packet

Diagnoses, meds, payer, level of care — structured in seconds, not scanned by hand.

DECIDE

Best-fit ranking

Every referral scored against facilities' beds, capabilities, and economics.

WRITE BACK

Into clinical and billing systems

Responses, bedboards, billing data — pushed downstream automatically.

Deployed today

The agents already at work

Not a roadmap — production. Across thousands of SNFs the fleet reads admission packets, answers hospitals, runs bedboards and census, and defends claims: ~19 million automated runs every 30 days.

Referral Intelligence

All referrals, one dashboard. AI extraction from admission packets, best-fit rankings, and a queue admissions teams clear in minutes instead of afternoons.

ScoringRecommendationsAI chatReports

Census Growth

Live bedboards and census anywhere. Forecasted discharges, projected open beds, and referrals matched to openings before competitors even reply.

BedboardFloorplanPayer mixReadmissions

Revenue Cycle Management

Automation at the front of the revenue cycle: cleaner admissions, denial prevention, and coding capture — so census growth becomes collected revenue.

Clean claimsDenial preventionCoding capture

Two-way Messaging, incl. Epic

Real-time coordination with hospital discharge planners inside the tools they already use — bed availability, questions, and acceptances without phone tag.

EpicBulk replyNotificationsMobile

Agentic Automation

A fleet of portal bots working 24/7 — ~19M runs every 30 days — logging in, pulling packets, and posting updates automatically.

Portal bots24/7Write-back

Compliance

Growth with continuous compliance. Documentation, level-of-care transitions, and billing rules tracked automatically as part of the same loop.

DocumentationLOC transitionsBilling rules

The data flywheel

Every referral makes every agent better.

MORE FACILITIES

More real episodes

Each new building adds thousands of genuine referral-to-payment episodes a year to the training corpus.

BETTER POLICIES

Higher win-rates

Policies retrain on real outcomes — acceptance speed, denial causes, payment timing — and ship back to every facility at once.

BETTER OUTCOMES

Easier expansion

+$525k average additional annual revenue per facility is a case study every operator understands — portfolios adopt building by building.

THE NETWORK

A dataset nobody else has

Referral patterns across hospitals, payers, and facilities — structured operational data of post-acute care that compounds with scale.

The road ahead

From admissions to the operating layer of post-acute care.

Healthcare administration breaks into operating loops, and every loop is a building block: model it in the environment, set the reward, train the agent, deploy. Each block funds and trains the next.

NOW

Census + RCM for SNFs

Referral-to-claim automation live in thousands of SNFs, on simple per-facility SaaS that lands whole portfolios.

Full revenue-cycle autonomy

Eligibility, prior auth, billing, appeals — agents owning the dollar from admission to remittance.

THEN

The post-acute network

Hospitals, facilities, and payers coordinating through Magicare agents on both sides of every referral.

THE VISION

Agents for healthcare administration

The RL environment generalizes: simulate any operating loop in healthcare — financial, staffing, clinical — prove the agent, deploy it at scale.

The building blocks

Healthcare administration, one trainable loop at a time.

Census and revenue cycle were the first blocks — proof that a simulated loop becomes a production agent. Everything else an owner-operator manages is the same shape: a loop the world already models, a reward an operator already tracks.

BLOCK 01 · LIVE

Census & referrals

The block that proved the method: agents trained on simulated referral episodes now run intake, bedboards, and census in production.

BLOCK 02 · LIVE

Revenue cycle

Eligibility, auth, claims, denials. The reward is a claim paid first-pass — and every payment and denial keeps updating the policy.

BLOCK 03 · IN TRAINING

Staffing & scheduling

Same environment, new reward: every shift covered as census and acuity move, with agency as the last resort — trained on the same simulated census that already drives the world above.

BLOCK 04 · IN TRAINING

Cost management

Reward the number owners actually watch — cost per patient day, held down at constant quality — and let the agent search the levers: agency mix, supplies, length of stay.

BLOCK 05 · IN TRAINING

Clinical operations

Plan of care and MDS as a loop: clinical signals in, next-best action out, documentation current and survey-ready — proven against simulated residents before it ever touches a chart.

Traction

Live, revenue-generating, loved.

Thousands of SNFs run on Magicare today — portfolio-wide deployments, results operators repeat to each other, and an environment that gets smarter with every one of them.

+$525,000

Average additional annual revenue per facility

40 min → 3 min

Referral review time, before and with Magicare

1,000s

Post-acute care facilities live on the platform

Trusted by 100s of leading healthcare organizations

With Magicare, all our homes are now full.

Mary Jean CappabiancaInspira Health Group

Magicare increased my facility's revenue by $1M.

Jonah WeissTacoma Nursing and Rehab

With Magicare we are never alone. They are true partners.

Anaili RendonExcelsior Care Group

Connected across the healthcare ecosystem