Artificial intelligence start-up Anthropic is preparing to release a range of new features focused on business users, pitching its AI models at enterprise over seeking mass market adoption.
Mike Krieger, the Silicon Valley group’s chief product officer, told the Financial Times that it was focused on how computer systems could “help with knowledge work” for “people who spend all day in meetings or in Excel or Google Docs”.
“It feels like we should be able to make that easier,” added Krieger, who was previously a co-founder of Instagram before joining Anthropic last year.
The group is banking on businesses driving new revenues for products based on its popular Claude AI models rather than chasing rivals such as OpenAI and Google which are also building chatbots and agents designed for more widespread use.
“I hope Claude reaches as many people as possible, but the critical path is not through mass-market consumer adoption right now,” said Krieger on stage during the recent HumanX conference in Las Vegas.
The comments come after the company closed a $3.5bn fundraising round this month, tripling its valuation to more than $60bn.
While the fundraise was a sign that investor interest in AI start-ups was showing no sign of waning, there are still concerns over when such bets can start to deliver meaningful returns through creating successful business models for generative AI.
Anthropic’s rivals have also invested heavily in building AI products for business users. Microsoft has rolled out “Co-pilot” across its Office productivity software, while OpenAI is marketing “ChatGPT Enterprise” for businesses seeking more powerful uses of its AI models customised for specific organisations.
Anthropic said its revenue from developers and enterprises via its application programming interface — or API — was growing twice as fast as its consumer subscriptions.
Internally, the company has been testing a prototype which it plans to “commercialise” and build into Claude in the coming months.
The feature examines workers’ calendars and prepares a report ahead of their meetings, analysing internal and external information about the client, such as who else they have met internally and previous notes about the company.
“Ideally, you’re a salesperson who walks into every one of those meetings as prepared as possible, but they don’t have time to do that report for every single company,” Krieger said.
“That’s a way in which the models do what they’re really good at: go off and do asynchronous work, think about the problem, and then create something consumable. Hopefully, that will be an accelerant for our sales folks.”
Anthropic also plans to launch voice-driven AI systems, where people can prompt and respond to AI models through a spoken conversation. It is an area where competitors OpenAI and Elon Musk’s xAI have also recently released features and Meta is also focused on.
“We will do voice internally, too; it is a useful modality to have. We have prototypes,” he said.
“We are doing some work around how Claude for desktop evolves . . . if it is going to be operating your computer, a more natural user interface might be to [speak to it].”
Krieger said Anthropic was discussing the launch with “a bunch of partners” already with AI voice products to potentially speed up the release.
One person familiar with the move said Amazon and London-based AI start-up ElevenLabs were part of preliminary discussions but no agreements had been signed.
Anthropic is backed by both Google and Amazon and has recently partnered with the latter to deliver its AI-powered Alexa product.
