AI agents (2025)

1. An interview with Convergence’s cofounder and CEO, Marvin Purtorab

   London-based Convergence AI raised Europe’s largest pre-seed round of 2024: a $12m deal led by Balderton Capital. It has since built its first prototype AI agent — named Proxy — which the company claims is the best performing in the market on a series of metrics. “In the long run, the aim is to build anything that can get arbitrary work done”, says CEO/cofounder Pourtorab. What exactly Proxy will be used for, Convergence is still figuring out.

   On B2C vs B2B

   The company is still experimenting with its target audience. “The AI agent space is just over a year old,” says Purtorab. “The most valuable thing for us at the current stage is to see what people use [agents] for. The reason why we went mainly [to the] consumer first is because it helps us gather a lot of information on what users want to do with it”.

   But he sees Proxy, like other agents, becoming an enterprise-focused product in the long-run, mostly because “people probably prefer to automate things at their job versus automating things in their personal life.” And those use cases are already emerging through behavioural analysis.

   On monetisation

   As with any new tech, pricing strategies will partly depend on client focus. For now, the most common AI agent revenue stream is pay-per-seat and/or subscription models, but that’s still up for debate.

   “Some people are trying pay-per-solved-tasks, which is attractive because AI agents aren’t that powerful yet. But agents are getting more powerful every single week, if not day, and we’re going to have very general agents in a year or two”. At that point, he says that pay-per-task will become unattractive: “That's the reason why we switched, even like humans, to salary pay. We’re basically a subscription to human labour”.

   On competition & infrastructure providers

   In a space where multi-billion companies ship products at the infrastructure level, it may be disheartening for startups to make a break for it. “We want to give people the ability to build their own agents, but I would prefer everyone to have that ability, and not just specialised AI firms that use our API to then sell another product. This should be so simple that everyone can do it”.

   The AI agent world is already getting crowded. “There's tons of AI companies and agent companies out there now, but most of them are vertical agents focusing on one use case at a time. I don't think that humans will have these tens of thousands of different vertical agents. You need to work at a higher level, which is a general interface”.

   On public resistance & regulation

   The big worry about AI agents is that they will eventually replace humans. “In general, there's initial resistance to any technology. But a product gets adopted if it makes people's lives better, not worse”. On those lines, Purtorab believes the productivity bumps from AI agents won’t necessarily translate to aggressive cost-cutting in companies.

   “When automation keeps growing, we probably need to tax it quite strongly — it’s natural. I don’t think there’s anything bad about that,” he says. Though in the long-term, things will get more blurry. “If we get to a point where AI takes whole jobs, I’m not sure what the social contract will be. [e.g. reskilling, or the birth of publicly subsidised incomes through “automation” taxes]. In that scenario, you’re essentially taxing future profits of AI companies.”

Who early stage startups are up against