Speaking at the SXSW London Festival, CEO Google Deepmind shared that his team is building the next generation of AI-powered email assistants. Not only does it filter and organize your messages, but it also responds to them on your behalf with your own tone of voice.
“What I really want and what we're working on is being able to receive the next generation of emails,” Hassavis said. “I want to get rid of my emails. I'll pay thousands of dollars a month to get rid of them.”
It's frustration that many people can relate to. Overflow inbox. I missed the message. Guilt of late reply. DeepMind's upcoming tools, still under development, aim to solve all of that. It identifies everyday emails, generates responses based on the user's writing style, and expresses only the most important ones.
This does not dig into the thread or issue an apology such as “I'm sorry for the late response.” And importantly, it can protect you from increasing pressure on other online algorithms that compete for your attention.
“It might basically give you more time and maybe protect your attention from other algorithms to get your attention,” he said. “I think we can actually use AI in our personal services.”
Behind the inbox: bigger image
For those who have been credited to major breakthroughs like Alphafold 2 (AI systems that crack protein folding mystery), working on email may seem like a strange step-down, but he suggests. Hassavis juggles two main roles. At Deep Mind, he leads high stakes scientific research. At the same time, he is providing innovations to Google's consumer products, including the Gemini chatbot and the video generation model VEO 3.
This balancing act reflects the broader tension that shapes the direction of AI.
“The capitalist engine did what it did its best,” he said, noting how commercial demand pushed AI into public use faster than originally imagined.
Still, he continues to focus on long-term artificial general information (AGI).
Agi on the horizon
AGI is not science fiction, and at least not in Hasabi. He predicts that such a system can perform intellectual tasks that humans can do – but could arrive within five to ten years.
“It's very short considering how important it will be,” he said. “It's nothing but a new industrial revolution.”
In this scenario, the AI can unlock what he calls “radical beings.” This includes everything from disease stiffening to an endless supply of clean energy. However, Hassavis warned that even the best future would require careful preparation.
“Even if you have good times when you get radical abundance and economic prosperity, can you make sure it's pretty shared and quite distributed?” he asked. “I hope economists and social scientists are thinking about it.”
He also urged countries, particularly the US and China, to come together to guide AGI development safely and collaboratively.
“After all, it's for the benefit of all humanity,” he said. “It will affect all of humanity.”
Advice for the next generation
Despite the increased footprint of AI, Hassavis has not lost faith in traditional skills. He believes that today's students still have to master mathematics, physics and computer science if they want to understand the systems that reconstruct the world.
You can understand “understanding the fundamentals of mathematics, physics, and computer science” and “how these systems are assembled.”
At the same time, he urges young people to embrace AI tools as part of their learning and work. Over the next decade, he hopes that a new type of work will emerge. It rewards people who work with intelligent systems and are comfortable with them.
He said AI will generate “new and invaluable jobs” over the next few years, particularly benefiting “technologically savvy people on the forefront of using these technologies.”
As Deepmind progresses in frontier research, it may be an email tool rather than a medical breakthrough that will land in your hands first. Irony is not lost in Hasabis. For now, he is starting small, amid AGI's bold promise. Inbox to take care of itself. A message that won't get lost. And there is one less reason to say, “I'm sorry for the late reply.”
