Manual Process: Ammar aimed to eliminate the manual prescription process, as tracking each visit was a major pain point for both prescribers and patients. Keeping, managing, and updating records was the bottleneck

PA Approval: A study revealed that both patients and pharmacies require complete patient history for PA approval related to specific diagnoses.

Media System (EMR): Ammar discovered the Electronic Medical Record (EMR), specifically designed to organize health-related information, supporting everything we need to streamline the process.

Prescribers Meeting: We integrate the EMR API to obtain the prescriber’s meeting schedules using their NPI number.

Gemini Integration: The prescriber recorded each conversation on his phone, auto-generating transcriptions and medical notes using an LLM. After reviewing, Moonlight saved the records back to the EMR for further processing.

Designing for an ever-changing context, including archetypes, behaviors, and user personas, was the most challenging part. At that time, AI struggled to understand pauses during conversations. Different linguistic styles, background noise, and third-person voices forced us to think broadly; if it didn't grasp the speaking style, it completely altered the language in transcript generation. Therefore, we had to consider these deep points and continuously improve.

Edit Notes: He introduced the edit option on previews; if anything goes wrong, the prescriber can edit it.

System Prompt Refinement: He made sure the LLM understands the user context and Different linguistic styles

Advocating for the prescriber: He created the guidelines to illustrate how the LLM perceives constraints and identifies best practices to enhance the success of AI transcript generation.

EMR allows patients to schedule appointments effortlessly, helping doctors manage their time better. Ammar enjoyed enhanced LMM recording, which now produces detailed transcripts and notes.