Convergence, a London-based startup, has raised $12 million in pre-seed funding to advance its personal AI agents, known as Proxy. The funding round was led by Balderton Capital, alongside contributions from Salesforce Ventures and Shopify Ventures. This funding will be directed toward building a new generation of AI models, focusing on continual learning and skill acquisition, with long-term memory at its core.
Empowering AI with Memory
Convergence is pushing the boundaries of current AI technology by developing models that enable AI agents to continuously learn and adapt, much like humans. Their personal AI agent, Proxy, has the potential to transform how individuals manage their daily tasks—whether at work or in personal life—by learning and memorizing workflows, thereby freeing users from repetitive, mundane tasks.
James Wise, a Partner at Balderton Capital, praised the team’s expertise, saying, “Few people have the experience and skill that Marvin and Andy bring to the table. In just a few short months, they’ve assembled an exceptional team and are on the verge of delivering a groundbreaking product that could revolutionize the way we live and work. We’re thrilled to support their journey.”
A Team of Visionaries
Convergence was founded in April 2024 by machine learning engineers Marvin Purtorab and Andy Toulis, both of whom have a history of working with AI-driven systems. The two first crossed paths at Shopify, where they collaborated on recommender systems and AI assistants, later joining Cohere, an AI platform for enterprises. Their growing team includes machine learning experts and engineers from top-tier organizations like Google, Meta, PolyAI, and DeepMind.
Introducing Proxy: A Game Changer for Workplace Productivity
The vision behind Proxy is to tackle the inefficiencies that plague the modern workplace, where, according to reports, employees lose up to 62% of their workday to repetitive tasks. Convergence has launched Proxy in beta, with limited spots available, to help users automate routine workflows and boost productivity. Unlike many AI agents that are narrowly focused on specific tasks, Proxy is designed to handle a wide variety of tasks across different domains.
Whether it’s scheduling meetings, booking holidays, or ordering groceries, Proxy’s key differentiator lies in its ability to acquire new skills over time through long-term memory and continual learning. This is achieved by leveraging a new class of AI models called Large Meta Learning Models (LMLM), which are engineered to learn how to learn.
Building for the Future of Human-Machine Collaboration
Co-founder and CEO Marvin Purtorab describes the potential of Proxy: “Imagine an AI agent that continuously learns and evolves, allowing you to offload routine tasks and focus on more meaningful work. Proxy’s ability to memorize and apply what it learns will be a game-changer for both personal and professional productivity.”
CTO Andy Toulis envisions a future where humans and AI work hand-in-hand to solve some of the world’s biggest challenges: “We see limitless potential in human-machine collaboration. By empowering AI to continuously learn, we are not just automating tasks—we are building a future where people and Proxies collaborate daily to unlock new possibilities.”
Located in London, Convergence aims to capitalize on the city’s reputation as a hub for top AI talent as they continue to refine and expand Proxy’s capabilities. With their innovative approach to AI and the backing of leading investors, Convergence is poised to redefine how we interact with AI in our everyday lives.