Innovation Engine

Inspiration

There's increased opportunities in the medical space, and supporting the MedTech industry seems compelling.

What it does

Connects startups with stakeholders and resources using AI-powered matching to help them grow. There are two separate onboarding paths / login journeys, as a startup and as a stakeholder. On the startup portion of the app, there is an AI powered bot that can serve as a virtual concierge to answer questions and steer startups in the best direction.

How we built it

Gemini 2.5 to create architectural specs, engineering requirements, step-by-step prompts to use in a coding assistant (Cursor) and documentation. After getting stakeholder feedback, and the team verifying the technical accuracy of the docs, fed those docs into ChatGPT to evaluate. Then took the updated output and fed into Cursor (mostly via Claude )to do the initial build of a prototype. Iterated over this process multiple times after additional stakeholder feedback.

Challenges we ran into

Ambiguity about the "pain points" MUSC is facing, and what specific aspects of an app would provide a novel solution.

AI coding agents outputting faulty code as the project size grew (possibly context too large)

Integrated 2FA via Clerk - however the free version only allows around 20 2FA text messages per month.

"Scope creep" and being able to nail down core requirements given the longer duration of this particular hackathon

Accomplishments that we're proud of

How quickly we were able to leverage AI to create technical specs and requirements, and get a basic app running in a dev environment.

Frequent feedback loops with MUSC stakeholders and startups (potential customers if we were to commercialize).

What we learned

Google AI Studio (Gemini 2.5) and Claude 4 are awesome.

Need to verify code output of LLMs, and experience is still needed to evaluate that. Vibe coding is a powerful thing but need to be careful of "garbage-in-garbage-out".