https://www.canva.com/design/DAHDRsJVKhw/Z_jglL3JlOXlb6UZJgpCzw/view?utm_content=DAHDRsJVKhw&utm_campaign=designshare&utm_medium=link2&utm_source=uniquelinks&utlId=he4518b5d15

Problem

We live in a world where connecting with others should be easier than ever. There are more events, communities, and platforms than at any other point in history. Yet for many people, actually turning intention into real-world interaction is surprisingly difficult.

Imagine someone who wants to attend a local tech talk, a music gig, or even just plan a dinner with friends. The process usually begins by searching across multiple platforms — event marketplaces, community newsletters, social media groups, and messaging apps. If they find something interesting, the next step is checking calendars, coordinating schedules with friends, and figuring out travel logistics. Sometimes they realise none of their friends are interested in that particular event. Other times they simply give up because organising everything feels like too much work.

This friction exists everywhere in our social lives. Platforms for events, communication, and scheduling all exist — but they rarely work together. A person might discover an event on one platform, coordinate through another messaging app, check availability through their calendar, and buy tickets somewhere else entirely. These disconnected systems make planning social experiences unnecessarily complicated.

Even when people do attend events and meet new people, the problem continues. After a great conversation, both people might exchange contact details with the genuine intention of staying in touch. But when they return home, the moment passes. They may forget to follow up, hesitate about what to say, or simply get caught up in their daily routines. A potentially meaningful connection quietly disappears.

Our team saw this as a systemic problem rather than a personal one. People do want to connect. The issue is that there is no infrastructure designed to support the full lifecycle of social interaction — from discovering experiences, to attending them, to building lasting relationships afterward.

HuddleUp is our attempt to solve this problem by reducing the friction involved in planning, attending, and following up on real-world

social experiences.

Idea explanation

HuddleUp is an AI-powered platform designed to help people discover experiences they care about and connect with others who share those interests.

Rather than acting as just another event platform, HuddleUp focuses on simplifying the entire lifecycle of social interaction. When a user joins the platform, they quickly create a profile that captures and builds upon:

• the types of experiences they enjoy

• the interests and communities they belong to

• the kinds of connections they hope to build

From there, HuddleUp creates a personalised feed of experiences by bringing together events from platforms such as Eventbrite, Luma, local community listings, etc. Instead of searching across multiple websites or scrolling through newsletters, users see events that align with their interests, location, and the times they are actually free. The system also considers signals such as preferred event types and past activity, helping surface opportunities that feel genuinely relevant rather than random.

Planning becomes easier because the platform connects directly with the user’s calendar. When someone expresses interest in an event, HuddleUp checks their availability, accounts for travel time, and suggests the best way to fit the experience into their schedule. If they want to attend with others, the platform can also help coordinate plans by identifying friends or compatible users who may be interested in the same event and have overlapping availability.

The experience does not stop at discovery. After events, HuddleUp uses Gemini to generate a personalised HuddleDebrief that highlights meaningful connections and suggests thoughtful ways to reconnect. Instead of leaving people unsure how to follow up, the system provides context about shared interests and drafts natural messages that users can edit or send.

To encourage genuine participation, the platform also introduces HuddlePoints. Users earn points for actions that strengthen the community such as attending events, sharing reviews, reconnecting with people they meet, and contributing to the reliability of event information. These contributions help improve the quality of recommendations while also rewarding users who actively participate.

By bringing together event discovery, calendar coordination, social matching, and thoughtful follow up support, HuddleUp transforms what is usually a fragmented planning process into a seamless experience that helps people turn intentions into real world connections.

Implementation

HuddleUp was built as a single-page web application using HTML, CSS, and vanilla JavaScript, with all AI functionality powered by the Google Gemini API. This stack allowed us to rapidly prototype the full user journey within hackathon constraints.

Mock Data and Simulation

The prototype uses structured mock datasets covering user profiles, event listings, and social connection graphs in place of a live backend.

Planned Production Integrations

•       Live event feeds via webhooks from Luma, Eventbrite, etc.

•       Google Calendar API for real-time availability and scheduling

•       Third-party platform data from LinkedIn, Spotify, and Strava...

•       Persistent backend with scalable storage and continuous behavioral learning

Google Gemini Integration

Gemini serves as the intelligence layer across the platform — an embedded reasoning engine that drives personalization, discovery, and follow-up throughout the entire user journey.

•       Personality Matching: Processes signals from connected platforms to build interest vectors and match users on complementary traits.

•       Event Reasoning: Ranks events against a user's interests, availability, location, and energy level to surface a relevant shortlist.

•       Event Scraping: Parses unstructured listings from Luma, Eventbrite, and newsletters, extracting vibe tags, social intensity, and commute difficulty.

•       HuddleDebrief: After events, identifies the connections most worth following up and provides context on why.

•       Follow-Up Messages: Drafts personalized messages grounded in the shared experience, ready to edit and send.

•       Review Summaries: Condenses venue reviews into short atmosphere summaries so users know what to expect.

•       Agentic Actions: With user authorizaion, registers for events, purchases tickets, and handles logistics via external APIs.

The system simulates integrations with external platforms such as event providers, calendars, and interest based services. In a full production environment, these integrations would operate through APIs and backend services.

Key Product Features

HuddleUp is designed to eliminate the friction inherent in social planning by acting as an intelligent intermediary between discovery, scheduling, and connection.

• Unified Event Discovery: HuddleUp aggregates fragmented event data from across the web into a single, personalized feed. By analyzing user interests, goals, location, and calendar availability, the system surfaces events that align with the user’s actual lifestyle, significantly reducing search time.

• Calendar-Aware Coordination: The platform integrates directly with user calendars to ensure social plans are feasible. By accounting for travel time, schedule buffers, and availability, HuddleUp proactively recommends events that fit seamlessly into a user's week and identifies opportunities for friends to attend together.

• Social Compatibility Matching: HuddleUp transforms large, intimidating gatherings into opportunities for genuine connection. By synthesizing structured interest profiles—including HuddleList goals, personality types, and shared passions—the platform helps users identify compatible people, fostering deeper relationships based on high-signal compatibility rather than proximity alone.

Challenges

• Simulation vs. Reality: The compressed hackathon timeline necessitated the use of mock data to represent real-world functionality, as full live integrations could not be completed within the 24-hour window.

• Monolithic Instability: The initial reliance on a monorepo structure created a significant single point of failure; bugs in one module frequently impacted the stability of the entire application.

• Prompt Engineering Complexity: achieving consistent, high-quality outputs from Gemini required continuous iteration, as minor adjustments to system prompts significantly altered the reasoning accuracy.

• Data Security & Privacy: As a platform aggregating sensitive calendar, location, and social data, HuddleUp faces substantial security challenges that must be addressed to ensure user privacy at scale.

• Integration Friction: Managing authentication, data normalization, and state synchronization across multiple external services introduced significant overhead and technical fragility.

• Feature Prioritization: Navigating the vast potential of social coordination required strict scoping. Balancing a feature-rich experience with a clean, intuitive UI necessitated difficult trade-offs regarding which capabilities to defer.

Accomplishments

Here is your Accomplishments section, structured to highlight your technical growth, product vision, and team execution.

• Successful End-to-End Prototyping: Despite this being our first hackathon, we designed and executed a fully functional prototype mapping the complete social lifecycle—from initial onboarding and interest discovery to event exploration, attendee matching, and AI-driven post-event follow-ups.

• Mastery of the Google Gemini Ecosystem: We moved beyond basic API calls, integrating a multi-modal AI stack:

  • Gemini 1.5 Pro: Used for high-dimensional reasoning and social compatibility matching.

  • Gemini Nano Banana: Leveraged for high-fidelity image generation within the app.

  • Veo: Utilized for cinematic video content generation.

  • AI Studio: Employed for rigorous prompt engineering and iteration to ensure output quality.

• Transparent & Accountable Gamification: We architected the HuddlePoints system to serve as more than just an engagement tool. By clearly defining how points are earned and rewarded, we created a transparent feedback loop that builds user trust and ensures community accountability—a core requirement for any platform handling sensitive social data.

• Agile Execution & Team Velocity: We adopted a formal Agile methodology for the first time, utilizing iterative sprints and daily stand-ups to manage scope, handle technical blockers, and maintain momentum under extreme time pressure.

• Vision-Driven Social Tech: We successfully transitioned from a abstract problem statement to a tangible solution that challenges the status quo. HuddleUp reorients social technology away from passive digital consumption and toward active, meaningful real-world interaction.

Next Steps

HuddleUp’s current prototype serves as a proof-of-concept for intelligent social orchestration. To reach production scale, the platform will evolve through three primary focus areas:

• Advanced Gemini Integration: Transitioning from manual debriefing to autonomous, agentic workflows. This includes automatic post-event follow-ups, semantic intent-based search (e.g., matching "vibes" to availability), and AI-driven profile building that infers interests directly from linked applications.

• Enterprise-Grade Infrastructure: Shifting to a robust AWS-powered backend. The architecture will utilize AWS Lambda for serverless compute, Cognito for secure social identity, and DynamoDB for persistent data management. Infrastructure will be defined via AWS CDK to ensure scalability, security, and version control.

• Ecosystem Expansion: Building a comprehensive community trust layer through organizer verification, cross-platform API synchronization, and smarter group-coordination tools. This will transform HuddleUp into a universal coordination layer that bridges the gap between digital discovery and real-world connection.