Omar Marks | Light Rocket | Getty Images
Cohere co-founder Nick Frosst is surrounded by talk of artificial general intelligence (AGI). He was more than happy to stay out of it.
Established in 2019 by formerGoogle An artificial intelligence researcher with an estimated multi-billion dollar stake, Cohere is one of the higher profile names in the generative artificial intelligence space, which has seen explosive growth since OpenAI launched ChatGPT in late 2022.
But it’s not a company that’s well-known among consumers, who use OpenAI, Google, and Perplexity for their chatbots and other tools. Instead, Cohere is all about business.
“I’ve been in meetings with companies in health care, banking and IT,” Frost told CNBC this week. “The questions I get are about securely automating tasks like HR, payroll, research, and fraud detection to increase productivity. No one has asked me about implementing AGI, let alone ASI.”
The latter is the abbreviation of artificial intelligence, that is, artificial intelligence that significantly surpasses human intelligence. Both OpenAI and Anthropic have made this a goal.
Cohere’s latest funding round in July bulge 500 million, with a valuation of $5.5 billion, more than double last year’s valuation. The company’s investors include NVIDIA, AMD, salesperson and Oracle.
While that’s historically a hefty price tag for a company less than five years old, it’s a fraction of what investors paid for OpenAI, which valued it at $157 billion CNBC confirmed this week that Anthropic is in talks to raise capital Valuation: $60 billion.
Some of Cohere’s main competitors in the artificial intelligence arms race offer products for consumers and businesses. For example, OpenAI launched ChatGPT Enterprise in 2023, and Anthropic Launch of Claude Enterprise September.
Cohere’s preference for the enterprise centers on the idea that large language models are best suited to automating tedious tasks and “becoming colleagues,” Frosst said.
“Really, this is an automated tool,” Frost said. “When I think about my personal life, there aren’t actually many things that I want to automate. I don’t want to write text messages to my friends faster. I don’t want to answer emails more efficiently in my life. But in I really, really want to do that in my work life.
“I want to be able to have the freedom to think creatively and not be bogged down,” Frost said.
Shortly after closing its financing in July, Cohere laid off about 20 people. A company representative said at the time that it was an “internal reorganization” and that Cohere had a “clear vision for the future.”
This vision includes an all-out effort to develop artificial intelligence agents.
Although the term artificial intelligence agent There is no clear definition and it is generally intended to describe artificial intelligence services that go beyond chatbots. Agents are typically designed for specific business functions rather than general purposes, and can be customized on large AI models.
They can perform complex multi-step tasks on behalf of the user and generate their own to-do lists so the user doesn’t have to step-by-step guide them through the entire process.
Maintain capital efficiency
On Thursday, Cohere launched an early access program for its AI agent platform called North, which is focused on allowing users with any level of technical background to “immediately customize and deploy AI agents” and “with just a few clicks.” ” can be achieved. the company said in a press release. Users can search for information across their entire organization in multiple languages and departments using previously unconnected programs.
This includes summarizing HR questions and answers, speeding up financial reporting and automating some core business functions in customer support and IT.
Frost said the platform could be used in any industry, but the company plans to target the financial and healthcare sectors, where data privacy and regulation are critical.
Martin Kon, Head of Operations at Cohere March told CNBC By focusing on enterprise AI, the company has been able to operate efficiently and control costs even amid chip shortages, rising Nvidia graphics processing unit (GPU) costs and changing AI model licensing fees.
Those dynamics are still at play, Frosst said, allowing Cohere to “increase capital efficiency,” which is increasingly “interesting among investors.” Competitors of popular consumer-focused AI products use a lot of computing on “consumer applications and scientific projects,” he said.
Frosst said that while the sales cycle for enterprise AI may be longer, “the recurring business we’re able to create is really resonating with investors right now.”
Competition is fierce and technology is evolving rapidly.
October, artificial explain Its artificial intelligence agents are able to use computers to complete complex tasks just like humans. The feature, called “Computer Use,” allows its technology to interpret content on a computer screen, select buttons, enter text, navigate websites and perform tasks through any software and real-time Internet browsing.
Open artificial intelligence Reportedly planned A similar feature will be rolled out soon. Last year, from Microsoft, Yuan Google often touts their goal of making their AI assistants more productive.
Even without a consumer business, Cohere would have to invest heavily in Nvidia’s expensive GPUs, which are in huge demand from companies that train models and run massive workloads. When Cohere was founded, the company gained reserves Google chip to help its pre-trained models. Over the past year, Cohere has turned more to Nvidia’s H100 GPU.
“We’ve increased our spending on them because they work so well,” Frost said.
