Jolt | Devpost

This is the landing page of our website

Inspiration

Prior authorization remains one of the most inefficient parts of the healthcare system. Despite advances in clinical technology, approval workflows still rely heavily on manual documentation review, fragmented EHR data, and unclear insurance policy requirements. Seeing how much time clinical staff spend preparing and resubmitting prior authorizations motivated us to build Jolt, a system focused on reducing administrative friction and improving approval efficiency.

What it does

Jolt is an AI-powered prior authorization assistant that automates the most-time consuming parts of insurance approval. It reads insurance policy rules, scans patient medical records for supporting evidence, identifies missing requirements, and generates a fully cited clinical summary.

How we built it

We used LLM to parse complex insurance policy PDFs into structured medical requirements. By using RAG, Jolts searches unstructured EHR notes to extract relevant clinical evidence. Then, we build a confidence-scoring system to estimate approval likelihood and highlight risks, and an auto-generation pipeline for submission prepare documents.

Challenges we ran into

Insurance policies are inconsistent, vague, and often written in non-clinical language. EHR data is fragmented and messy, making reliable evidence extraction difficult. Ensuring every AI-generated claim could be traced back to a specific clinical note was also a major technical challenge.

Accomplishments that we're proud of

We reduced prior authorization preparation time from over 45 minutes to under 2 minutes. Jolt achieves full source traceability for extracted clinical evidence and provides transparent approval-risk insights instead of black-box decisions.

What we learned

Healthcare automation requires trust, explainable, and precision. Speed is not enough, clinicians need to see why something will be approved or denied. Building with real clinical workflows in mind is critical for adoption

What's next for Jolt

Next, we plan to integrate directly with EHR systems using FHIR, expand policy coverage across major insurers, and incorporate denial feedback to continuously improve approval accuracy and confidence scoring.