London's Convergence Secures $12M to Launch Revolutionary AI Agents with Human-Like Memory
London's Convergence Raises $12 Million to Develop AI Agents with Human-Like Memory
In an exciting development for the AI industry, London-based startup Convergence has announced its emergence from stealth mode with an ambitious goal: to create personal AI agents equipped with advanced long-term memory capabilities. This breakthrough aims to revolutionize how both consumers and businesses interact with AI. The company, founded by former Shopify and Cohere colleagues Marvin Purtorab (CEO) and Andy Toulis (CTO), has successfully secured $12 million in pre-seed funding. The funds will primarily be used to enhance the technology behind its flagship AI assistant, Proxy, which is designed to handle tasks as diverse as booking groceries and managing workflows.
What Happened?
Convergence, founded by AI experts Marvin Purtorab and Andy Toulis, has emerged from stealth mode with an exciting mission: to build personal AI agents that not only complete tasks but also evolve through continuous learning via long-term memory. The company has already raised $12 million in pre-seed funding from leading investors, including Balderton Capital, with additional backing from Salesforce Ventures and Shopify Ventures.
The company's flagship product, Proxy AI, is an intelligent assistant designed for both consumers and employees. For employees, the AI agent is tasked with automating administrative tasks and streamlining workflows, enhancing productivity. For consumers, it can take care of everyday activities like scheduling appointments or ordering groceries. What makes Proxy truly innovative is its ability to learn, adapt, and even manage its own memories—much like a human being.
Key Takeaways
-
Significant Funding: Convergence has secured $12 million in pre-seed funding to further develop its Proxy AI assistant, with the aim of improving its memory and learning capabilities.
-
AI Agents with Long-Term Memory: The Proxy AI agent can adapt over time by continuously learning and managing its memories. It is capable of adding or deleting information as necessary, a feature that mimics human cognitive processes.
-
Dual Use: Proxy is designed for both consumers and employees, making it a versatile tool. It can handle personal tasks like trip planning and groceries, while also streamlining administrative tasks in business environments.
-
Advanced Technology: The system is built on Large Meta Learning Models (LMLMs), a groundbreaking form of AI architecture focused on "learning to learn," allowing the AI to generalize across various applications rather than being confined to specific tasks.
Deep Analysis
Convergence’s innovative approach taps into a growing need for personal AI systems that not only automate routine tasks but also understand and adapt to the needs of users over time. While personal AI assistants like Alexa and Siri already help users accomplish simple tasks, Convergence is aiming much higher with its Proxy AI. By integrating long-term memory, Proxy can continuously evolve, becoming more useful as it learns from its past interactions. This means the more you use it, the smarter it gets.
The foundation of Convergence’s AI technology is Large Meta Learning Models (LMLMs), a relatively new class of machine learning models. Unlike traditional AI systems that are often limited to completing specific tasks, LMLMs have the capacity to "learn to learn." This allows the Proxy AI agent to adapt its knowledge to a variety of applications, making it far more versatile than existing AI assistants. By embedding memory management directly into the model's architecture, Proxy can decide when to remember or forget information during real-time interactions, optimizing its functionality without overwhelming the user with unnecessary data.
The team's combined expertise from working at top AI organizations like Shopify, Google DeepMind, Meta, and OpenAI underscores the depth of knowledge behind Convergence’s innovations. Their background in machine learning and recommender systems uniquely positions them to tackle the complexities of memory in AI systems.
The fact that the company has garnered interest from major investors like Salesforce Ventures and Shopify Ventures suggests that these industry leaders see the potential in Convergence's Proxy AI to transform not just consumer interactions, but also enterprise workflows. Their support will provide the startup with both the capital and the industry connections needed to bring its ambitious vision to life.
Did You Know?
-
Proxy’s Long-Term Memory: Unlike current AI assistants that only respond to simple, one-off tasks, Proxy AI has the ability to remember past interactions, making it capable of adjusting to your preferences over time. It can manage its memories dynamically, adding or deleting data as needed, a feature inspired by human memory functions.
-
Learning to Learn: The core of Convergence’s technology is its use of Large Meta Learning Models (LMLMs), which give the AI a remarkable ability to generalize across a range of tasks. This is a step beyond traditional AI models, which are typically designed for specific use cases.
-
Beta Launch: Proxy AI is currently being tested in a limited beta, giving select users a first-hand experience of its capabilities. As the system evolves, it will continue to improve its memory management and become more intuitive with use.
Convergence’s emergence from stealth mode signals an exciting shift in the AI industry. With the backing of leading venture capital firms and a team of seasoned AI professionals, the company is set to push the boundaries of what personal and enterprise AI assistants can do. By focusing on long-term memory and adaptive learning, Convergence may very well shape the future of how we interact with AI on a daily basis.