Agent 01
Referral management
Every inbound referral contacted, triaged, and tracked to a booked specialist visit.
Built for specialty practices, multispecialty groups, and healthcare operations teams.
A personal assistant for every provider.
TurnipCare is a self-learning AI operating layer for healthcare operations. It learns each organization's workflows, deploys specialized agents across referrals, scheduling, intake, prior auth, and follow-up, and keeps staff focused only on exceptions.
Backed by a clinical advisory team and engineers from Amazon, MIT, and beyond.
The handoffs between primary care, the specialist, and the patient are where care falls through — and where your team loses its day.
$300B+
of U.S. healthcare spend is operational, not clinical — the work that happens between systems, not in the exam room.
5:1
administrative teams outnumber clinical staff. Most of healthcare's headcount is spent moving information around.
Like a tech consultant for every provider
Software that adapts to your practice — not the other way around.
The orchestrator routes the work to the right agent. Every clinic's rules stay theirs.
Every workflow teaches the system how your organization operates, making each agent better over time.
A full-service team your practice needs.
Orchestrator
You type a request in plain English. The orchestrator reads it and hands the work to the right agent on your team.
6 agents online
Agent 02
Prior authorization
Assembles the packet, files with the payer, closes the loop. Denials surface as exceptions.
Agent 03
Patient scheduling
Books to each provider's real rules — slot lengths, double-booking, new-vs-follow-up.
Agent 04
Specialist coordination
Surgical scheduling, pre-op intake, records — the handoffs between primary, specialty, and OR.
Agent 05
Patient communication
SMS-first, multilingual, async. Patients confirm or reschedule at 9pm — no hold queue.
Agent 06
Patient Panel Outreach
Recall lists, follow-up panels, payer outreach — volume the front desk can't reach.
Case studies
Live in production, learning each practice.
Live in production
Orthopedic department
A multi-specialty clinic's orthopedic department. Four providers, three scheduling rule sets, 1,000+ patients handled in three weeks.
Provider 01
15-minute mornings. 20-minute afternoons.
Double-booking allowed for injection and gel-shot appointments. Morning slots reserved for new patients. Afternoon for follow-ups.
Provider 02
Alternating new and follow-up.
Across the day, the schedule rotates between new patients and existing follow-ups. No back-to-back of either type.
Provider 03
Ten-minute open slots.
First-come, first-served. The agent enforces no double-bookings and stops accepting once the day is full.
0+
Unique patients handled by 4 providers in under 3 weeks.
0%
Resolution rate on actionable requests.
0%
Self-resolved with no staff time.
0%
Reduction in average handle time.
Figures from a production deployment.
Illustrative — representative figures
Patient panel outreach
A primary-care group reactivates overdue patients across its panel — recalls, care-gap follow-ups, and payer-driven outreach the front desk never has time to work.
Workflow 01
Overdue recall sweep.
The agent works the recall list nightly — overdue physicals, lab follow-ups, and care-gap flags — texting patients to rebook.
Workflow 02
Payer-driven outreach.
Annual wellness visits and quality-measure gaps surfaced from payer feeds, contacted and scheduled without front-desk time.
Workflow 03
No-show recovery.
Cancellations and no-shows are automatically re-offered to the panel, keeping the schedule full.
0+
Overdue patients contacted across the panel in month one.
0%
Reactivated and rebooked from recall outreach.
0%
No-show and cancellation slots recovered.
0
Added front-desk hours.
Illustrative — representative figures, not from a live deployment.
One queue. Just the exceptions.
Your team stops working a stack of 150 callbacks. They work the exceptions — everything else is already handled.
See it on live data
turnipcare · exception queue
Today
3 items need a human
147handled
34scheduled
3exceptions
-
Insurance ambiguous M. Liu Orthopedic consult
Plan listed as "BCBS PPO" but member ID prefix unrecognized. Agent paused at scheduling; needs eligibility confirmation.
-
Provider preference J. Williams Spine eval
Patient asked for "Dr. Chen specifically." Agent confirmed availability for both Dr. Chen and Dr. Park — needs staff choice before booking.
-
Clinical flag S. Kapoor Joint pain, urgent
Patient described "swelling and limited range of motion since last night." Agent flagged for clinician triage before standard scheduling.
Illustrative interface · patient details synthesized
A note from the founder
It can't be someone's job to sit and stare at a phone all day, waiting for it to ring. That's the part we automate. People do the work that needs them.
Somani Patnaik · Founder & CEO
Book a demo