view case
.png)
Pippen AI is a healthcare tech company founded by the owners of a Family Medicine Clinic in Toronto. Witnessing firsthand the administrative burden on physicians, they envisioned an AI-powered solution that alleviates these tasks and empowers doctors to focus on patient care. Their goal was to build a Minimum Viable Product (MVP) that could be commercialized across Ontario, validating its impact on the province's healthcare provider crisis before scaling nationally.
Start-up
Healthcare
MVP
Canadian family doctors report spending an alarming 19 hours per week on administrative tasks. This burden overwhelms 94% of our family doctors, consuming nearly 40% of their time and contributing significantly to the shortage of family doctors in both Ontario and throughout the rest of the country. In 2023 alone, over 2.2 million Ontarians lacked a family doctor. Pippen AI wanted to create a platform that tackles these challenges head-on by capturing clinical encounters through speech-to-text technology, offering intelligent support for clinical questions, streamlining administrative tasks, and providing billing assistance.
The biggest hurdle was ensuring the product adhered to strict privacy regulations like PII, PHIPA (Canada), and HIPAA (US). This meant finding a speech-to-text solution that anonymized patient information and integrating a large language model in a way that remained compliant.
The platform needed to accommodate diverse workflows, allowing doctors to choose between real-time documentation and post-visit catch up. As this was an MVP, we did not want to spend time building an EMR (Electronic Medical Records) integration. However, we had to ensure that doctors could easily generate and copy SOAP notes, differential diagnoses, treatment options, referrals, billing and diagnostic codes from Pippen to their preferred EMR. While large language models can be powerful, they lack the specific knowledge needed to generate accurate billing and diagnostic codes in Ontario. Fine-tuning the model to address this wasn't feasible due to time and resource constraints.
During the technical discovery, the team validated the feasibility of a secure, cost-effective speech-to-text solution with PII redaction capabilities. We evaluated various large language models (LLMs) and API assistants, exploring OpenAI and Azure offerings alongside other LLM models. Prompt engineering techniques were also explored. During the product discovery, we conducted user research, competitive analysis, and market research to ensure product-market fit and identify potential limitations. Our final deliverables for this phase included user personas, user journeys, a Product Requirements Document (PRD), a high level roadmap, information architecture diagrams and user flows.
During the design phase, we created and validated wireframes before finalizing hi-fidelity designs. In the name of keeping this project at an MVP level, we used Google Material Design’s UI Kit to speed up design and development.
Finally, we built a product meeting HIPAA/PHIPA compliance guidelines that leverages Deepgram's on-premise deployment capabilities, PII redaction features, and HIPAA compliance. We used Azure’s offering of OpenAI GPT4-Turbo with elaborate prompt engineering as our LLM of choice for several documentation options, and built custom databases for billing and diagnostic codes to ensure accuracy. We also used Stripe to administer monthly subscriptions with a free trial and integrated it with our token-based system to bill based on platform usage. Retool was used to expedite the development of the admin panel.
Finally, our rigorous Quality Assurance process ensured a smooth launch, with zero bugs identified during the User Acceptance Testing window.
The developed MVP is currently being tested at the founding family medical clinic and rolled out to beta users before wider commercialization. Crowdlinker is continuing to support Pippen with their user’s technical support inquiries as well as building their roadmap features.
To learn more about Pippen AI, visit pippen.ai.
As clients of Crowdlinker Inc, we are thrilled to share our experience with their exceptional team and the outstanding AI-powered physician assistant they helped us develop. From the very beginning, Crowdlinker impressed us with their robust process and clear, consistent communication. Their team's expertise and dedication were evident at every stage, ensuring that the MVP not only met but exceeded our expectations. The collaboration with Crowdlinker was seamless. They took the time to understand our needs, provided valuable insights, and adapted quickly to any changes or challenges. Pippen is already making a significant impact in our own clinic operations. We are particularly grateful for the excellent support and professionalism of the Crowdlinker team. Their commitment to our project's success was evident, and they worked tirelessly to deliver a product we are genuinely proud of. We highly recommend Crowdlinker to anyone looking for a reliable and innovative software development partner.
News & Media
