AI21 Labs Ltd., a generative artificial intelligence startup that rivals OpenAI LP, said today it has secured additional investment to close on a hefty $208 million funding round that raises its valuation to $1.4 billion.

The new investors in the Series C round include Intel Capital and Comcast Ventures, who join previously announced investors, bringing the company’s total amount raised to $336 million.

AI21 Labs is both a research lab and a commercial operation that’s focused on natural language processing, a subset of AI. It has some notable founders, including its co-Chief Executive Yoav Shoham (pictured, right), a Stanford professor emeritus and serial entrepreneur who has already sold two companies to Google LLC. There’s also Amnon Shashua (center), a professor at Hebrew University and founder of Mobileye NV, which was acquired by Intel Corp. for $15.3 billion in 2017, and co-CEO Ori Goshen (left), founder of the crowdfunding platform CrowdX Ltd.

The startup’s mission is to shift and shape the way people read and write by making AI a “thought partner” to humans, and to achieve that it has created a number of NLP-based applications. It explains that many large language models available today are a “jack of all trades” focused on versatility and applicable to a wide range of use cases.

On the other hand, AI21 Labs says its LLMs are much more focused. It offers application programming interface-based access to what it calls “Task-Specific Models” based on its Jurassic-2 LLM that have been engineered to excel in distinct NLP capabilities.

By focusing on more specific applications, the startup reckons, its models perform with superior accuracy while reducing so-called “hallucinations,” where models fabricate their responses when they don’t know how to respond to a prompt. It says its models can cater to a diverse range of enterprise needs, providing contextual answers, summarization and more.

For instance, it offers a Contextual Answers model that has been especially designed for grounded question answering, based on enterprises’ proprietary datasets. Other apps include Wordtune, a multilingual reading and writing AI assistant, and Wordtune Read, which is used to analyze long documents in seconds and create a simplified summary.

Shoham stated his belief that the generative AI industry will shift to a more comprehensive “systems approach” where LLMs are enriched with knowledge, reasoning and statistical inference. “This enables us to define a flexible architecture with multiple LLMs, complemented by discrete knowledge and reasoning modules,” he said.

In addition to building proprietary LLMs that can be fine-tuned by customers to perform specific tasks, the startup also provides tools for application development through the AI21 Studio. It offers a neuro-symbolic architecture that combines its LLMs with external knowledge sources and discrete reasoning.

According to Constellation Research Inc. Vice President and Principal Analyst Andy Thurai, AI21 Labs has an advantage over some of its rivals, because Jurassic-2 is generally more accurate than other custom LLMs. “It’s comparatively cheaper to consume too, and its text generation and summarization capabilities, plus the applications built on top of it, such as Wordtune, are very good,” he said.

The company has also built its own chatbots, similar to ChatGPT, in order to showcase the capabilities of its LLMs. One of its quirkiest is the Ask Ruth Bader Ginsburg AI bot, which was trained on more than 27 year’s worth of published interviews, speeches and written texts from the late U.S. Supreme Court Judge of the same name.

Anthony Lin, corporate vice president and head of Intel Capital, said the multidisciplinary approach being pursued by AI21 Labs is vital to deliver AI to end users. “The AI21 full-stack offering combines foundation models with successful applications and operation tools that will help enterprises accelerate generative AI adoption to increase productivity and affect their bottom and top line,” he said.

AI21 Labs’ other co-founder and co-CEO Goshen said the funding from today’s round will allow it to increase mindshare that “one size doesn’t fit all” when it comes to LLMs. “Mass deployment of AI requires deep understanding of high-performance language models that can deliver better value and impact,” he explained. “Our approach is about designing AI with purpose, making it significantly more efficient than building from scratch, and much more cost-effective.”

Photo: AI21 Labs