Skip to content

Greenbottle (by Storika)

GreenBottle solves the long-term memory problem in AI.

Code Video Cerebral Valley

Core trick

Combines a NetworkX knowledge graph and ChromaDB vector search, weighting stored brand events by significance with exponential temporal decay.

Tags

agents

Description

GreenBottle solves the long-term memory problem in AI. Current LLMs lose coherence over long histories — context windows overflow, and temporal relationships get lost. We built a hybrid memory system that combines: - Knowledge Graph (NetworkX) — brand ontology with entities, relationships, and multi-hop traversal - Vector Memory (ChromaDB) — semantic search across 100 years of brand events, campaigns, and model history - Temporal Decay (EWA) — significance-weighted exponential decay so founding events stay relevant for decades while fleeting trends fade naturally - Gemini 3 Flash — LLM-powered memory enrichment, consolidation, and context injection via Google ADK agents
storika/gemini-3-seoul-hackaton-202602★ 20 forksPythonCreated Feb 28, 2026 · last push Feb 28, 2026

README excerpt

# Green Bottle: Temporal Knowledge Reasoning for Brand Intelligence

**Live Demo: [https://greenbottle.live](https://greenbottle.live/)**

**Gemini 3.0 Seoul Hackathon 2026**

---

## The Problem: AI Branding Is Shallow

Today's AI-driven brand campaigns are optimized for **momentary traffic** — click-through rates, viral hooks, and fleeting engagement. They chase the surface and miss the soul. The result? Brands that feel disposable. Campaigns that are forgotten tomorrow.

**AI brands fail to deeply move humans** because they lack temporal depth. They have no memory of why a brand succeeded 50 years ago, no understanding of how cultural shifts shaped consumer loyalty across generations, and no reasoning about the causal chains that built iconic brands over a century.

## Our Solution: Temporal Knowledge Reasoning Graph

We apply **Temporal Knowledge Reasoning Graph** technology to give AI the depth of a century of brand knowledge.

Instead of asking "what's trending now?", we ask: **"Why did this brand succeed in this era, given the cultural context, consumer sentiment, and competitive landscape of that time?"**

The system constructs a dynamic knowledge graph that evolves across a 100–200 year timeline, powered by:

- **Temporal Decay**: Nodes and edges carry time-weighted relevance — recent events matter more, but foundational heritage never fades
- **First-Order Logic (FOL) Reasoning Chains**: Formal causal reasoning that connects era context → public sentiment → brand/mo

Team