Gnosis AI: Onchain agents & longtail predictions markets
summary
Agents run fully on-chain, call a dynamic tool registry and use an AI Mech to fetch evidence, scrape ~13 websites and output a P_yes 0.2 for “China invades Taiwan by end of 2025.” Hundreds of agents trade with individual budgets of a few dollars, each research run costing about 4 cents, and the best agent systems already solve ~50% of level‑1 and ~18% of level‑3 benchmark tasks. Claim: these self‑funding, spawnable agents become a new lifeform by 2030.
AI-written summary · about
chapters (9)
- 0:00Prediction markets overview
- 1:41AI agents in action
- 5:22Market trading strategy
- 7:30General task agents
- 9:05Agent architecture
- 10:50Benchmarking AI performance
- 13:15Design evolution
- 14:16The future of AI life
- 16:13Implications and outlook
related talks
27mFrom Theory to Real-World Impact
18mNext-Generation AI Agents: Orchestrating Technological Evolution
30mHow AI Agents Will Change Everything (2026)
01h 01mKeynote Panel: Future of AI agents
34mAI Agents: Transforming Autonomous Decision-Making
14mReliable Agentic AI, the Open Way
34mThe Compute Economy of AI Agents
15mHow AI is transforming possibilities into reality