Reflection AI, a startup building AI agents, has raised new funding at a $100 million valuation, Business Insider has learned.
Sequoia Capital invested in the funding round, according to three sources familiar, valuing it at $100 million, two sources familiar said.
AI agents promise to execute difficult tasks, like booking an appointment or updating Salesforce. Reflection aims to take this promise even further by building “superhuman general agents that automate knowledge work done on a computer,” according to the company’s website.
Reflection AI and Sequoia declined to comment.
The company’s cofounders, Misha Laskin and Ioannis Antonoglou, left Google’s DeepMind to launch their new startup, reported The Information earlier this year.
In a recent Sequoia podcast, Laskin said “a universal agent needs to be a broad, a very general agent that can do many things, can handle many inputs, but it also needs to have depth in the kind of task complexity it can achieve.”
In the podcast, Laskin highlights the different types of AI agents in the market. There are examples like AlphaGo, an AI program that beat professional players of the strategy board game Go, which is an agent with deep expertise on a specific task. “AlphaGo is probably the deepest agent that has ever been built. It can do one task. So not that useful. It can play Go, but not tic tac toe,” he said.
Large language models like Google’s Gemini, Anthropic’s Claude, and OpenAI’s ChatGPT and GPT models that are broad and haven’t been “trained for the agency,” Laskin said in the podcast.
Laskin conducted AI research at the Berkeley Artificial Intelligence Research Lab and most recently worked at Google DeepMind, the company’s AI research lab. He partnered with his colleague Ioannis Antonoglou, one of the creators of AlphaGo. Antonoglou recently led reinforcement learning from human feedback (RHLF) for Google’s large language model Gemini.
Reflection isn’t the only startup building AI agents. Imbue, building reasoning-focused agents, gained a $1 billion valuation after raising a $200 million Series B last September, according to its website. Other companies focus on specific verticals. Decagon, for instance, focuses on customer support, while Sybill targets sales reps.
Startups like Emergence, AgentOps, Crew AI, and Phidata provide infrastructure for enterprises seeking to build their own agents. And multi-agents systems are becoming the talk of the town among VCs.
Agent startups are also already being acquired. In June, Amazon hired away the cofounders of AI agent startup Adept, which raised more than $400 million in funding, and licensed its technology, reported GeekWire.
“Ioannis and I could have stayed and tried pushing agents, you know, at DeepMind,” Laskin said in the podcast. “The reason we decided to do it in our own way is because we think we can move quickly, much faster against this goal.”
“Some of this urgency is driven by a real belief that we are three or so years away from something that resembles a digital AGI…that’s what I’d been referring to as a universal agent,” he added.
Source link
lol
