Convergence, a London startup creating personal AI agents for everyone, has snapped $12 million in pre-seed funding led by Balderton Capital, which recently backed Tilt and Checkly. Salesforce Ventures and Shopify Ventures also participated in the round.
Funds utilisation
The funding will be used to develop novel models that power Proxy assistants, going beyond the current generation of AI, with a focus on enabling continual learning and skill acquisition through memory.
James Wise, Partner at Balderton Capital, said: “Few people have the experience and skill that Marvin and Andy have, which makes them well-placed to take on the complex technical challenge of a product like Proxy. In only a few short months, they have built a stellar team and are ready to ship a remarkable product that could transform how we live and work. I look forward to being a part of their growth journey.”
Team behind Convergence
The company was founded in April 2024 by machine learning engineers Marvin Purtorab and Andy Toulis. They met at Shopify while working on recommender systems and AI assistants, before joining Cohere, an AI platform for enterprises. The team comprises ML experts and engineers from PolyAI, Google, Meta, and DeepMind.
What does Proxy do?
Reportedly, 62% of an average employee’s workday is lost to repetitive, mundane tasks. Convergence has launched Proxy in beta, with a limited number of spots available, to resolve this challenge.
Pairing users with personalised AI, Proxy can learn tasks and workflows, freeing workers from their administrative burden and unleashing productivity. It is also applicable beyond the workplace and can be tailored to consumers, offering everyone the opportunity to have their own personalised AI agent to support them with tasks from booking holidays to ordering groceries.
Unlike most agents that are designed for specific workflows, Proxy will be able to work across a range of tasks and domains, acquiring skills like a human would through the use of long-term memory and continual learning. This is made possible by a new class of models called Large Meta Learning Models (LMLM), which are trained to acquire the skill of learning itself.
Marvin Purtorab, co-founder and CEO of Convergence, said: “Imagine an agent that keeps growing and learning whenever you teach them something new. Over time, the agent will take on mundane tasks allowing you to focus on higher-level tasks that are more impactful and also more enjoyable. The key to our approach is the ability for Proxy to learn and memorise tasks on the job, remembering them for the long term and being able to use its memories to extrapolate to new tasks – just as a person would. Our mission at Convergence is to build new, foundationally different technologies that are a clear pathway to useful intelligence. Balderton shares our vision deeply and we are looking forward to turning it into a reality with them.”
Andy Toulis, co-founder and CTO of Convergence, said: “We’re building a future where people and Proxies are collaborating every day on the job. There’s unlimited potential in human-machine collaboration – together, we’ll solve important problems for humanity. We’re excited to be building this in London which has some of the best AI talent in the world.”
