9 · Each one is a small AI app that solves one real problem — with a 30-second demo and three ways to get value.
An experiment in building an AI "digital life" as an on-chain system: five Solana contracts govern rules and funds, a Python kernel runs heartbeat, memory and self-programming — every behavior publicly observable.
Track on-chain smart money: live monitoring, candidate signals, pre-trade AI assessment, unified risk control, automated execution — evolved from a solo tool into a multi-tenant control plane with billing, treasury and an admin console.
Four modules in relay: AI scans contract vulnerabilities, EVM simulation verifies exploitability, a tri-entry atomic contract executes in milliseconds, Flashbots routes privately. Executor contract deployed to testnet.
Upload one photo, meet your storybook hero in 30 seconds, read a personalized 24-page story in 3 minutes, then order the hardcover. Multi-model orchestration plus a layout system keeps the AI look out of a premium gift.
Four agent personas plus one Brain meta-agent: deploy a token in 67 seconds, write lore to IPFS, post on X after real x402 USDC payments, and run on an autonomous heartbeat — all verified on-chain.
A minimal self-editing HTTP API harness: the agent accumulates helpers as it works, and git is the long-term memory. Core LOC is CI-budgeted — line count is part of the product.
A 10-role agent team autonomously handles sourcing, copy, publishing, support and retros; the human does exactly two things: monitor and authorize. Guardrails cannot be bypassed; memory compounds across cycles.
Four specialized agents — Researcher, Creator, Reviewer, Archivist — relay through research, writing, QA and archiving: one sentence in, eight finished formats out in five minutes.
A 10-step standardized pipeline covering script, storyboard, keyframes, image-to-video, voice and editing — aiming at 100 one-minute AI anime episodes. First pilot in production.