Reliable Agentic AI, the Open Way
summary
AI shifts from chat to systems and the talk lays out three concrete pillars: training, inference, agents, with specific claims and examples. Infrastructure demand hits token processing 50–100x token growth, software yields the claimed 30–40x efficiency gains vs hardware alone, and enterprises already run many models — Uber trains 20,000 models/month. Open tools drive inference speed: Llama CPP sees NVIDIA 40% speedup upstream and VLLM ranks as the top open project in 2025, while kata containers sandbox risky agent-generated code.
AI-written summary · about
chapters (5)
- 0:00Introduction to open AI infrastructure
- 2:39Three pillars of open source AI
- 8:19Inference innovation in open source
- 10:23The rise of AI agents
- 12:49Key takeaways and future outlook
related talks
42m3 Pillars of Open Source AI: Training, Inference, Agents
21mDemocratizing AI with Open Source Multi-Agent Systems
45mFrom Tools to Partners: The Rise of Agentic AI
35mThe Fight for the Future of Artificial Intelligence
44mBuilding Agentic AI Systems: From Transactions To Conversations
54mThe Future of AI: Building the Most Impactful Technology Together
32mThe Rise of AI agents and open sources Ai models
34mAI Agents: Transforming Autonomous Decision-Making