Infrastructure of AI
summary
Agentic trading stack ships as a self‑hosted product called Donut Zero that connects LLM-driven bots to Solana with low‑latency CLIs and a control panel; deployed agents already generate roughly $1M revenue across trading and yield‑farming experiments. A realtime Solana layer exposes native instructions for verifiable callbacks and private payments, targeting high throughput web‑2 latency. A post‑training orchestration service (early April) claims to run at 20% of web‑2 cost and a multi‑model API marketplace for inference.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 2:53AI infrastructure challenges
- 7:48AI trading use cases
- 11:19Real-time Solana transactions
- 12:26Model aggregation and orchestration
- 13:51Upcoming features
related talks
08mBrowsing the Agentic Internet
22mAI Investor Panel
23mLeveraging AI & Layer 2 Blockchain Solutions
15mBuilding AI Agents the Web3 Way
16mAI Without Limits: How ZeroG is Building the Future of Decentralized AI
16mTokenization and Agents: Why It Matters
06mThe Building Blocks of Agentic Al
15mHow AI is transforming possibilities into reality