Goalmate

• 

  Main project screen

Inspiration

Big communities feel like noise. Posting in a 5,000-person Discord/Slack channel is basically shouting into the void — but most people don’t need 1,000 connections. They need the right 3 people who understand what they’re dealing with right now.

What it does

Goalmate is an AI-driven Micro-Community Engine that turns a user’s current focus into an instant Micro-Pod of 3:

• Evolve (goals): “hitting $10k revenue”, “1,000 Spotify streams”
• Enjoy (interests): “beginner bouldering”, “gaming nights”
• Support (struggles): “overwhelmed by finals”, “anxiety about interviews”

It matches people by meaning (not keywords), then shows a “Why We Matched” / Icebreaker Card so the pod immediately feels safe and clear.

How we built it

• Semantic matching with Gemini Embeddings: convert user text into vectors, then match by nearest neighbors (intent/context > keywords).
• Pod formation logic: cluster users into temporary groups of 3 based on similarity.
• Transparency Engine with Gemini Pro (2.5 Pro): generate an explanation + icebreaker (“You all mentioned X, you’re all aiming for Y…”).

• MVP structure:

  • Frontend: Vue.js
  • Backend: FlaskAPI
  • DB: MongoDB
  • SDK: google.generativeai for embeddings + chat completion

Challenges we ran into

• Matching quality vs speed: making matches feel “obviously correct” while staying instant.
• Avoiding keyword traps: “sad” vs “lonely” vs “burnt out” should still connect.
• Trust & explainability: users won’t engage unless they understand why they’re grouped.
• Pod dynamics: keeping it small (3) so it’s not awkward, but still useful.

Accomplishments that we're proud of

• A clear, high-impact concept: from feeds → to on-demand peer support
• Matching based on semantic intent, not surface words
• The “Why We Matched” feature that builds psychological safety fast
• A realistic 24-hour MVP plan with modern tooling (Supabase + pgvector + Gemini)

What we learned

• People don’t want more communities — they want better moments of connection
• Explainability isn’t optional: it’s what converts a match into a conversation
• Small group design (3) can massively reduce friction and social anxiety

What's next for Goalmate

• Build the full Pod View (chat + Profile Data card)
• Add time-limited pods (e.g., 48h timer) to increase urgency + focus
• Improve safety: moderation, reporting, and optional topic boundaries
• Expand beyond campuses: companies, creators, fitness groups, local communities

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Goalmate

Built With

• flask
• gemini
• mongodb
• vector-embeddings
• vue