Peter Battaglia of Google DeepMind discusses how AI is revolutionizing weather forecasting. From predicting Category 5 hurricanes like Melissa with unprecedented accuracy to enabling hourly global forecasts with WeatherNext 3, machine learning models are leveraging decades of physics data to provide critical early warnings, optimize renewable energy, and transform agricultural planning, fundamentally reshaping our understanding of atmospheric dynamics.
Alex Hancock discusses the need for a standardized protocol for client-to-agent communication in AI, similar to how MCP standardizes agent-to-tool interaction. He introduces the Agent Client Protocol (ACP), born from Zed and JetBrains, as a JSON RPC-based solution for client software to control agent harnesses, send tasks, and receive updates. Hancock demonstrates ACP's flexibility with multiple clients driving a single agent and highlights how its new remote transport capability, combined with MCP and model APIs, enables independent placement of all agentic stack components, fostering a competitive ecosystem that will significantly enhance user experience.
Brandon Waselnuk of Unblocked discusses the critical need for robust context engines in AI-driven software development. He explains how AI agents' lack of company-specific understanding leads to compounding costs like wasted tokens and rework. Waselnuk outlines common pitfalls in providing context and details six essential characteristics of an effective context engine, demonstrating its impact with a 50% token reduction and significant time savings. He also introduces three open-source tools from Unblocked, including a workshop to build a relational context engine, emphasizing that "access to information is not understanding" and that context, not intelligence, is the current bottleneck for AI effectiveness.
Independent model evaluation is crucial for AI progress, as public benchmarks are saturated and models optimize for them. Rayan Krishnan discusses VALS's role in providing continuously evolving, third-party evaluations, highlighting the Llama 4 'disaster' and the need to measure increasingly agentic systems. The conversation covers the complexities of evaluation beyond capability to include cost and latency, the looming financial impact of token spend on enterprises, and the critical role of evaluations in informing AI policy and international coordination for aligning AI with human intent and managing geopolitical risks, especially concerning recursive self-improvement.
LinkedIn revolutionized its approach to AI agents by addressing the inherent limitations of LLMs in complex enterprise environments. Facing issues like hallucinations due to a lack of internal context, they developed a system of "playbooks" — structured, self-contained instructions published as tools. To scale beyond typical MCP limitations, they introduced three meta-tools (search, get schema, execute), enabling agents to dynamically discover and utilize thousands of internal tools and playbooks. This self-improving, context-aware infrastructure empowers agents to handle tasks from incident response to code generation, significantly boosting developer productivity and ensuring reliability.
Kevin Madura introduces Recursive Language Models (RLMs), a paradigm shift where an LLM treats its context as an interactable object within a REPL, rather than mere tokens. This allows for direct computation, recursive decomposition of problems by delegating to sub-models, and significantly mitigates context rot, outperforming traditional RAG, agents, and tool calls. Use cases range from complex data analysis and long document processing to code auditing and harness optimization, hinting at a future where models are natively RLM-aware.