A running log of what we've built, shipped, and learned. Posted as we go.
This repo is the work diary of a small team building at the intersection of AI, crypto, and professional services. We document experiments, ship notes, lessons, and the occasional rabbit hole — using modified names where client confidentiality applies.
The work spans: on-chain analytics, multi-agent AI systems, local LLM infrastructure, voice agents, credential systems, forensics tooling, agentic economy research, and whatever else comes up.
Projects: Structured write-ups of things we've shipped — what we built, why, what worked, what broke.
Ship notes: Short entries when something goes live. Date, what, one line on why it matters.
Experiments: Attempts that didn't fully land but taught us something.
Tools: Utilities we built to solve problems we kept hitting.
- AI: Anthropic Claude (main agent), local Ollama models (Qwen3, LLaMA), Grok voice agents, Frontier (Claude Code on NVIDIA DGX Spark)
- Infra: VPS + Docker + Nginx Proxy Manager, OpenClaw gateway, GitHub for async multi-agent coordination
- Chains: Ethereum, Base, Solana, NEAR — reading and writing
- Tools built on: Etherscan API V2, Supabase, HeyGen, ElevenLabs, Higgsfield, xAI Grok
- 2026 Q1 — ETHGlobal, OpenAgents, early multi-agent
- 2026 Q2 — Student Center, on-chain credentials, DeFi forensics
- 2026 Q2/Q3 — Local AI project: building firm intelligence from scratch
- 2026 Q3 — Voice agents: Main & Meta, ZIE, and the call that landed
- 2026 Q3 — On-chain forensics: LP position mapping
- 2026 Q4 — Multi-agent team: GitHub as async coordination
- 2026 Q4 — Agentic Economy: the thesis and what we built on it
Names of clients and confidential engagements have been modified. Everything else is real.