From Tools to Partners: The Rise of Agentic AI
summary
Google processes 1.3 quadrillion tokens/month and token costs fall from ~$30 per million to ~$0.08, driving explosive LLM use; model capability doubles task-handling every ~6–7 months. Enterprise agent deployment jumps from 11% to 33% to 42% in one quarter series, coding assistants create ~$10B ARR market and enable near‑100% benchmark recovery when wrapped with tool use. MCP lists about 17,000 servers as a unifying API, A2A enables authenticated agent-to-agent workflows, and hyperscalers plus Claude/OpenAI each release agent frameworks and SDKs.
AI-written summary · about
chapters (21)
- 0:00Agentic AI keynote intro
- 1:04Token processing growth
- 2:53Coding assistance first use case
- 3:55AI handling complex tasks
- 5:47Defining AI agents
- 8:34Workflows vs agent capabilities
- 10:21AI agent components
- 11:46Multi-agent system patterns
- 13:54Agentic AI frameworks trends
- 17:02MCP servers tool integration
- 20:14Agent communication protocols
- 22:38A2A protocol secure collaboration
- 25:26Agentic commerce payments
- 27:14AI coding assistants impact
- 30:23Why coding is AI focus
- 32:49Debunking AI project myths
- 34:36AI impact on jobs
- 37:47AI memory and learning
- 40:37World models challenges
- 42:23Rapid prototyping with AI
- 43:03AI co-workers future
related talks
34mThe Compute Economy of AI Agents
29mWhat exactly are AI Agents?
26mAPIs in Agentic world
14mReliable Agentic AI, the Open Way
28mAI and the state of APIs in 2026
15mThe Responsible Path to Hyperproductivity with AI Agents
29mAgentic AI: What It Is and Why It Matters
22mScaling AI for Developers: Lessons from the Inside