TAO Intel
Overview
TAO Intel is a production-ready Bittensor intelligence platform built for holders, validators, and subnet operators who need unified analytics without juggling multiple dashboards. The platform combines real-time market data (CoinGecko), live trading insights (Binance order books and volume analytics), and admin-curated educational content with context-aware AI chat across every section. Each tab delivers a specialized analyst persona—market commentary, trading flow analysis, or protocol education—powered by tab-specific data pipelines that pre-compute heavy aggregations (365-day price history, 38-metric trading analytics, knowledge base relevance matching) so responses stay fast and grounded in actual platform data. It is architected as a freemium SaaS with execution tracking, retry logic, and storage-cap enforcement built into the data layer from day one.
Features
- Built automated data ingestion pipelines with production-grade reliability (retry/backoff, job tracking, alerting) pulling from CoinGecko (5-min cadence), Binance (60-sec staggered collection), and future Bittensor RPC integration.
- Engineered a trading analytics pre-computation layer that samples 50k trades daily and derives 38+ metrics (liquidity, spread, momentum, order flow, time-of-day patterns) to enable instant AI responses without streaming raw data.
- Implemented AI context builders that assemble tab-specific data bundles (365-day market history, live orderbook snapshots, curated subnet documentation) and format them into structured prompts for Gemini 2.5 Flash.
- Designed an AI profile system (in progress) allowing users to switch between beginner/intermediate/expert personas per tab, with admin-managed prompt templates, temperature controls, and test preview workflows.
- Delivered a dual-table architecture separating 5-minute market snapshots (real-time stats) from daily closes (chart integrity) to prevent data pollution while maintaining live feel.
- Built an admin content management layer for subnet documentation, protocol guides, FAQs, and glossary with hierarchical organization (topics → subtopics → articles) and React Query-powered optimistic updates.