The Importance of AI Memory

Key Concepts

AI strength pillars (compute, power, data, algorithms, memory), algorithmic advancement, AI memory, genetic AI, tool use, Model Context Protocol (MCP), AI collaboration.

Reaction to Infrastructure Deals and AI's National Imperative

The speaker views the recent infrastructure deals, particularly US technology investments in Gulf data centers and development, as a positive sign. These deals highlight that AI is not just about corporate success but a "national imperative," underscoring the industry's importance. A decade ago, there were concerns about the US maintaining its lead in AI.

Four Pillars of AI Strength and Algorithmic Advancement

AI strength rests on four key pillars: compute, power, data, and algorithms. The US has historically led due to its strength in algorithms. As the US develops technological alliances for compute, power, and data, maintaining its lead in algorithmic advancement is crucial. The US possesses the best researchers and engineers globally.

The Importance of AI Memory

Beyond the four pillars, memory is an increasingly important component of AI. AI agents need to remember past interactions and their own states. For example, in healthcare, an AI agent assisting a physician should recall previous patient interactions, including data, vitals, and the nature of the communication. Companies like Open Evidence are developing tools to facilitate this.

China's AI Industry and US Leadership

The Bloomberg Big Take suggests that China's AI industry is thriving despite US political policy, with over 50% of AI researchers based in China. While China has impressive researchers, the US also has creative and brilliant minds.

Tool Use and Model Context Protocol (MCP)

At Sequoia A.S.'s annual conference, "tool use" was the top technical topic. Tool use involves AI agents working together, essentially teaching computers how to use other computers. A significant advancement is the Model Context Protocol (MCP), which acts as a universal translation mechanism. Each agent or software (e.g., a CRM) is considered an expert in its domain, but MCP enables them to communicate effectively, despite using different languages. This collaboration is vital for the US to maintain its AI leadership.

Real-World Application: Rocks and MCP

Rocks, a portfolio company, exemplifies the power of MCP. Rocks helps salespeople conduct research before meeting potential buyers. Using MCP, Rocks can connect to other tools to create personalized pitch decks and even generate demos using cognition or cloud code.

Conclusion

The US can maintain its AI leadership through collaboration, algorithmic advancement, and the adoption of protocols like MCP. By enabling AI agents and software to communicate and work together, the US can leverage its research and engineering talent to stay ahead of the curve. The development of AI memory and the focus on "tool use" are key areas for future growth and innovation.