Skip to main content
Back to A NEW AGENT - Hackathon

Project Gallery

41 Projects

Submissions Closed
Tech
Winner
Tryal AI
GitHub

Tryal AI

by Tryal AI

We're building the patient-side infrastructure for all clinical research: one profile, instant matching, and AI-guided applications at scale. This isn't just a better search tool it's rearchitecting how millions of Americans could access cutting-edge treatments instead of the few who do today. Our multi-agent AI system accelerates patient recruitment and optimizes operations throughout the entire trial lifecycle: Prescreening Agent conducts conversational eligibility assessments via voice and text, scoring candidates against trial-specific criteria from Trial Intelligence JSON and seamlessly transitioning qualified patients to scheduling. Scheduling Agent coordinates appointments by integrating with site calendars, offering available slots, confirming bookings, and sending automated reminders across SMS, email, and voice channels. Clinical Trial Coordinator Agent acts as a 24/7 digital assistant, answering protocol questions, visit preparation, and reimbursement inquiries while maintaining compliance audit trails and escalating complex cases to human coordinators. Retention/Engagement Agent proactively maintains participation through visit reminders, medication adherence nudges, and side-effect check-ins, detecting disengagement patterns and alerting coordinators for timely intervention. Impact: Transform clinical trial access from exclusive to inclusive, enabling faster recruitment, higher completion rates, and giving patients unprecedented access to life-saving treatments.

TracksElevenLabs+2
Built withCSS+5
TypeScript90.9%PLpgSQL7.5%+41
E
R
J
4 members
Winner

ESPN-like commentary for kids sports

by PlayVoice

We use AI (Gemini, Eleven Labs) to generate broadcast-style, play-by-play commentary over on any footage of amateur sports, like watching on ESPN. No editing skills required. Targeting youth sports parents who film their kids' games and want to share clips online.

TracksElevenLabs+1
M
-
Johnny Yu
4 members
Winner
Polimarket
GitHub

Polimarket

by Polimarket

Our platform transforms public politician trade disclosures into actionable, AI-driven investment insights for everyday investors. Today, lawmakers’ trades — legally disclosed under the STOCK Act — sit buried in PDFs, weeks late, and impossible for retail investors to use. Yet these trades often precede significant market moves, revealing where political and economic power intersect. We aggregate, clean, and structure this data in real time, then use AI agents to interpret the patterns such as which politicians are consistently outperforming. From there, users receive clear, ranked trade signals with full transparency into the source, timing, and rationale behind each move. The result is an investing co-pilot that turns obscure government filings into institutional-grade intelligence. Users can simulate or mirror trades, follow curated “Congress Portfolios,” or receive personalized alerts based on their own risk profile. In short: we bridge the transparency gap between political power and public markets — democratizing access to information that was always public, but never usable. Every investor gets the tools, data, and context once reserved for insiders — converting transparency into traction, and insight into opportunity.

U
A
M
3 members
Winner
Donna AI
GitHub

Donna AI

by Team Donna

This project builds an AI-powered voice agent designed to help lawyers prepare for courtroom trials through realistic mock interview simulations. The agent acts as an intelligent opposing counsel, questioning, challenging, and counter-arguing based on the details of a given case file. It helps lawyers refine their reasoning, argument structure, and response under pressure before actual court appearances - all without the dependency on senior attorneys or lawyer colleagues. Trained on diverse legal dialogue structures and case precedents, the agent can conduct scenario-based interrogations, identify logical gaps, and offer feedback on the strength of arguments. Users can input case briefs, witness statements, or evidence summaries, and the agent dynamically adjusts its questioning to mirror real courtroom exchanges. The system integrates voice interaction, natural language reasoning, and adaptive questioning logic to provide a realistic and rigorous preparation experience. By simulating both aggressive and neutral questioning styles, the agent enables trial lawyers to enhance clarity, confidence, and adaptability in legal argumentation. Ultimately, this AI mock interview assistant serves as a virtual courtroom coach, providing accessible, personalized, and repeatable preparation sessions along with comprehensive feedback that sharpen lawyers’ advocacy skills and improve overall trial readiness.

TracksEudia+2
Built withCSS+4
TypeScript80.1%CSS19.8%+1
Saloni Parekh
Purav Desai
H
4 members
Winner
AgentFacts
GitHub

AgentFacts

by AgentFacts

Point and click anything on the internet and AgentFacts will fact-check for you within seconds. Research reveals that most people instinctively trust what’s in front of them, often without pause, AgentFacts checks for you.

Built withCSS+3
JavaScript70.6%CSS12.5%+21
Evangeline
1 member
Finalist
Peazy Labs

Peazy Labs

by Peazy Labs

Peazy is an AI-first onboarding platform that conducts interactive training workshops to new employees just like a human expert. Current onboarding is passive and time-consuming with either 1. employees watch videos or read docs without real practice, leading to poor retention, low engagement, and slow ramp-up. 2. Business is paying a lot for expert led onboarding session that break their banks Peazy is for SaaS and professional teams with cost per training session at fraction of human expert costs. Additionally Peazy boosts skill retention, engagement, and measurable productivity. Peazy bridges the gap between knowing and doing, helping teams learn faster, perform better, and stay ahead in the flow of work. In short: training that actually trains you. Peazy MVP submitted is built using react+typescript, uses Speechmatics+Vapi for ASR+STT elevenlabs for TTS and Gemini 2.5 Flash as the agent brain It also has Clio as one of the workshops to demonstrate how it can train lawyers on contract management softwares

TracksElevenLabs+3
K
1 member
Hormone Harmony
GitHub

Hormone Harmony

by HLI

Hormone Harmony bridges the gap between hormonal awareness and action, giving users evidence-based to implement into a busy schedule. Hormone Harmony helps women understand and act on their body’s hormonal patterns and learning style. After a user logs their daily symptoms, energy, and mood, an Intervention AI Agent detects their hormonal phase and recommends personalized, research-backed interventions to manage symptoms and support your hormonal shifts. When users open an intervention, a Synthesizer Agent creates an actionable guide tailored to their preferred learning style text, audio, or visual using AI research synthesis, voice generation (ElevenLabs), and visual diagrams. Each guide is short, science-based, and designed for real-world usability. Users can complete a practice, reflect briefly, and track what works best for them over time.

TrackElevenLabs
Built withCSS+5
TypeScript93.8%MDX3.9%+227
P
1 member
Spec-Driven Development Tooling for Codegen Agents
Slides

Spec-Driven Development Tooling for Codegen Agents

by SpecMate

Current problems with codegen agents and vibecoding: The Underspecified Prompting If prompt is vague, codegen dev goes in wrong direction Each correction burns valuable context window After 40% context usage, performance tanks dramatically LLM forgets earlier requirements The Codebase Bloat Problem Each iteration adds new code without potentially removing old code Dead code accumulates, compounds over time Bloated codebase exceeds LLM's context window, becomes unmanageable and unrefactorable We need better ways to brainstorm and fully spec the project for codegen agents

D
1 member
HLI - Hormone Harmony
GitHub

HLI - Hormone Harmony

by HLI

Hormone Harmony bridges the gap between hormonal awareness and action, giving users evidence-based to implement into a busy schedule. Hormone Harmony helps women understand and act on their body’s hormonal patterns and learning style. After a user logs their daily symptoms, energy, and mood, an Intervention AI Agent detects their hormonal phase and recommends personalized, research-backed interventions to manage symptoms and support your hormonal shifts. When users open an intervention, a Synthesizer Agent creates an actionable guide tailored to their preferred learning style text, audio, or visual using AI research synthesis, voice generation (ElevenLabs), and visual diagrams. Each guide is short, science-based, and designed for real-world usability. Users can complete a practice, reflect briefly, and track what works best for them over time.

TrackElevenLabs
Built withCSS+2
TypeScript99.3%CSS0.5%+1
P
1 member