TL;DR
Built Polaris, an AI-powered search engine trained on Dermsquared's 20+ year content library, enabling dermatologists to get real-time answers to clinical questions without disrupting their workflow
Developed a hybrid AI and keyword search pipeline that categorizes clinical inquiries in milliseconds, routes them to specialized knowledge bases, and returns clinically relevant answers with supporting content
Designed an automated content ingestion pipeline that continuously indexes new dermatological content, including videos, podcasts, CME materials, case studies, and drug documentation, keeping clinical knowledge current
The Challenge
Dermatologists face a daily problem of information overload. Staying current with treatment protocols, drug interactions, access management workflows, and emerging research requires constant effort. When a clinical question arises mid-appointment or during documentation, there is no fast, reliable way to search a trusted dermatology knowledge base without context-switching to general search engines that surface generic, unvetted results.
Dermsquared had spent over 20 years building one of the most comprehensive repositories of dermatology-specific content, including expert videos, podcasts, conference proceedings, CME materials, case studies, and drug documentation. But that content was fragmented across the platform and not optimized for fast, conversational retrieval. Users could browse or scroll, but not simply ask a clinical question and get a direct, authoritative answer.
Traditional keyword search returns lists of results, but dermatologists need answers. A physician asking about systemic medication approval workflows or appropriate treatments for a teenage acne patient needs a synthesized response grounded in vetted clinical content, not ten links to sort through manually.
Dermsquared needed a search experience that could function as a digital clinical assistant, one that understood the intent behind clinical queries, pulled from verified expert content, and surfaced related resources in a way that felt native to how physicians actually work.
Key Results
Launched publicly March 2024 following staged rollout with pilot user testing
Content library spans 20+ years of expert dermatology content across articles, videos, podcasts, CME materials, case studies, and drug documentation
Query classification routes requests to specialized knowledge bases in milliseconds
Automated pipeline continuously indexes new content without manual intervention
Expanded in v2 to cover 100% of Dermsquared's public-facing website content
Powers the platform content recommender on video and podcast pages
