Sompl

The Story Of My Product's Life

Comment

Inspiration

Software teams have Git. Product teams don't.

Engineers can replay every code change ever made, but the reasoning behind a product — the hypotheses, experiments, metrics, trade-offs, failed ideas, and lessons learned — is usually scattered across Slack threads, Jira tickets, Notion pages, meeting notes, and people's memories.

When team members leave or enough time passes, that institutional knowledge disappears. New PMs end up revisiting decisions that were already made years ago.

We wanted to build version control for product thinking — a system that captures not just what a product became, but how and why it evolved.

That's Sompl — Story Of My Product's Life.

What it does

Sompl is a living evolutionary record of a product. Every feature has a birth, growth phase, mutations, successes, failures, and sometimes a death — and Sompl captures, visualizes, and learns from that entire journey.

Product Museum

Explore rich evolutionary stories of products like Spotify, Netflix, Airbnb, Notion, Figma, and Linear, including major feature launches, pivots, and features that were ultimately killed.

Evolution Timeline

A chronological history of a product with 14 event types spanning the entire lifecycle — from Idea and Research to Launch, Success, Iteration, and Kill. Each event stores metrics, ownership, attachments, alternatives considered, and lessons learned.

Evolution Graph

An interactive lineage tree built with React Flow that visualizes how features branch, evolve, merge, or die. Shipped features appear in green, while killed features appear in red. Every node can be explored in detail.

Smart Commit

Users can write plain English such as:

"We killed the social tab because only 3% of DAU used it."

AI instantly converts it into a structured product event with the correct event type, metrics, rationale, alternatives considered, and lessons learned. Content can also be pasted directly from Slack, Jira, or Notion.

AI Historian

With one click, AI generates a narrative history of the product, highlighting recurring patterns, successful decisions, failed branches, and recommendations for future direction.

Time Machine

Users can move through time using a slider or replay the entire product evolution to see exactly how the product changed across its lifecycle.

Public Story Pages

Products can be published as beautiful documentary-style pages that showcase their evolution. These pages are ideal for product retrospectives, case studies, and PM portfolios.

Novus Analytics

Every meaningful interaction inside the platform is tracked and surfaced in a live activity feed, providing insight into how users explore and interact with product histories.

How we built it

Frontend

  • Next.js 15 (App Router)
  • TypeScript
  • Tailwind CSS
  • Framer Motion
  • React Flow (@xyflow/react)

AI Layer

We used Groq running Llama 3.3 70B for:

  • Structured tool-calling to convert natural-language product updates into typed events.
  • Streaming generation for AI Historian narratives.

Database

We built on Supabase Postgres using:

  • Row-Level Security (RLS)
  • JSONB columns for metrics and alternatives
  • Self-referential feature relationships for lineage tracking

The Product Museum is seeded automatically and idempotently on first load.

Analytics

Novus.ai powers event tracking across:

  • Product views
  • Timeline interactions
  • AI generations
  • Graph exploration
  • Time Machine usage
  • Public story shares

Reliability

Every external dependency has a graceful fallback:

  • No Groq API key → AI mock responses
  • Supabase unavailable → localStorage persistence

This ensures the application remains fully functional even if external services fail.

Deployment

  • Vercel

Challenges we ran into

Making the demo impossible to break

Hackathon demos often fail because of API limits, missing keys, or backend issues. We designed fallback paths for every critical dependency so the application always remains usable.

Maintaining lineage relationships

Our museum dataset initially used string identifiers while our production database used UUIDs. We implemented parent-before-child seeding with ID remapping to preserve feature relationships.

Instant interactions with cloud persistence

We wanted every action to feel immediate while still syncing to Postgres. We solved this through optimistic updates that reconcile real database IDs asynchronously.

AI-powered data entry

Building Smart Commit was challenging because human product updates are messy and ambiguous. We needed AI to reliably extract structured information while maintaining trustworthiness and consistency.

Accomplishments that we're proud of

  • Creating what we believe is a genuinely new category: Version Control for Product Evolution.
  • Building an AI-powered data-entry experience where one sentence becomes a structured, metric-rich product event.
  • Delivering an interactive timeline, lineage graph, and replayable Time Machine that create an immediate "wow" moment.
  • Generating documentary-style public story pages suitable for portfolios, retrospectives, and product storytelling.
  • Shipping a polished, startup-quality product that feels production-ready rather than a typical hackathon prototype.
  • Designing a system that remains functional even when AI, database, or analytics services are unavailable.

What we learned

  • The hardest part of preserving product history is not storing information — it's reducing the friction required to record it.
  • AI-first data entry dramatically increases the likelihood that teams will actually document important decisions.
  • Killed features often contain the most valuable organizational knowledge, yet most product tools ignore them.
  • Designing for graceful degradation improves both demo reliability and real-world robustness.
  • Product storytelling can become a powerful distribution channel when teams can easily share the narrative behind their decisions.

What's next for Sompl

Integrations

Automatically generate product events from:

  • Slack
  • Jira
  • Linear
  • GitHub Pull Requests

Team Collaboration

  • Shared workspaces
  • Multiplayer editing
  • Roles and permissions
  • Authentication

Pattern Intelligence

Cross-product learning that identifies recurring patterns and insights such as:

"Teams that killed features like this often learned X."

PM Portfolio Mode

Transform product histories into professional, shareable portfolios that showcase product thinking and decision-making.

Enhanced Editing

  • Drag-and-drop timeline reordering
  • Rich attachments
  • Embedded documents and media
  • Improved collaboration workflows

Our vision is for Sompl to become the system of record for product evolution — the place where teams preserve not just what they built, but why they built it.

Built With

Share this project:

Updates