Robot Dojo is an open-source personal AI system that creates durable chat memory and builds apps on top of your data. A set of named specialist agents (an orchestrator, researchers, an architect, a builder, and an adversarial QC reviewer) run a deterministic pipeline from framing through build and quality control, each working in its own isolated context. A content-addressed, append-only memory log lets the system compound context across sessions instead of starting cold every time. I build it in my spare time to learn where AI value actually accrues: in owning the full stack of agents, orchestration, context, and memory, not in calling a model.
robotdojo.ai Personal AI system
Elephant Protocol mines property data directly from county recorder systems and writes it to an immutable onchain base layer. The system tokenized 10.5M Florida properties across 67 counties at under $0.01 per property per year, roughly 100x cheaper than incumbent providers. The architecture uses a TypeScript ingestion pipeline for county-level data extraction, Solidity smart contracts for onchain record storage and validation, and a burn-and-mint token model (150M fixed-supply MAHOUT) that prices data access at marginal cost. The protocol enables programmatic collateral validation, real-time property pricing, and automated title verification without centralized data dependencies.
github.com/elephant-xyz TypeScript / Solidity