Overview
End-to-end insurance prior-authorisation automation: extracts treatment codes from doctor notes, submits PHCS-formatted PA requests to 14 insurers via API, and tracks approvals in a real-time dashboard. PA turnaround 5 days → 4 hours; rejection rate drops from 22% to 6%.
The challenge
Majd's billing team of five was manually typing ICD-10 and CPT codes from handwritten doctor notes into 14 different insurer portals, each with its own form structure. The 5-day average PA turnaround was delaying elective procedures and causing patient dissatisfaction. A 22% rejection rate due to coding errors and missing supporting documents was generating rework that consumed 30% of the billing team's capacity.
The solution
Lirevon trained a Gemini extraction pipeline on 2,000 anonymised Qatari medical notes to reliably extract ICD-10, CPT, and DRG codes with confidence scores. A rules engine applies each insurer's specific coverage logic before submission, flagging likely rejections for human review before they are sent. Approval tracking across all 14 insurer APIs is consolidated into a single dashboard showing PA status, average insurer response times, and pending actions.
Outcomes
PA turnaround time fell from 5 days to 4 hours, eliminating delays on 90% of elective procedures
Rejection rate dropped from 22% to 6%, cutting rework that had consumed 30% of billing team capacity
Billing team overtime eliminated — staff reassigned from PA data entry to patient liaison roles
Revenue from recovered previously-rejected claims exceeded QAR 480,000 in the first six months
Key metrics
Tech stack
Services
Client testimonial
“We used to chase approvals for days. Now our coordinators get notified the moment a PA comes back and the rejection rate has become almost negligible.”
— وليد النصر، مدير العمليات الطبية
Related work
Last-mile delivery operator (80 drivers) · Kuwait City
Automated route-optimisation and proof-of-delivery pipeline: aggregates Talabat and Salla orders, generates optimised multi-stop manifests, and sends Arabic SMS delivery notifications with live tracking links. Failed-delivery rate drops from 14% to 3.1%; driver idle time falls 28%.
Result:Failed deliveries 14% → 3.1%; driver idle time −28%
Ghost-kitchen operator (6 brands) · Abu Dhabi
Multi-aggregator order routing and inventory automation: normalises Deliveroo, Talabat, and Careem Now webhooks, auto-assigns prep stations by brand SLA, and triggers low-stock WhatsApp alerts. Order-error rate falls from 9.4% to 0.8%; kitchen throughput rises 31%.
Result:Order errors 9.4% → 0.8%; kitchen throughput +31%
Residential property manager (1,100 units) · Abu Dhabi
Automated tenant-communications and maintenance-request pipeline: WhatsApp bot triages requests, generates work orders, routes to contractors, and tracks SLA compliance. Manual PM hours per unit fall from 2.1 h/mo to 0.4 h/mo; tenant NPS rises 29 points.
Result:PM hours 2.1 h → 0.4 h/unit/mo; tenant NPS +29 pts
Book a free 30-minute audit, or message us on WhatsApp. We'll send a fixed-price proposal in two days.