Enterprise Startups Race to Cash In on ChatGPT Mania5 min read

Two of the largest venture-capital deals across all sectors in the fourth quarter were for generative AI startups. Jasper, an Austin, Tex.-based marketing and promotions content-generation platform, in October announced a $125 million Series A funding round at a $1.5 billion valuation, led by Insight Partners with participation by Coatue and Bessemer Venture Partners. That same month, Stability. AI, a generative AI startup with offices in London and San Francisco, announced a $101 million seed round led by Coatue, Lightspeed Venture Partners and O’Shaughnessy Ventures LLC.  

Generative AI software is designed to produce wholly new text, images and computer code from a few simple prompts by processing vast stores of data through an AI-enabled language model or generative pre-trained transformer—the GPT in ChatGPT.

ChatGPT, a free generative AI chatbot launched in November by San Francisco software startup OpenAI, attracted more than 100 million monthly users in January, according to estimates by UBS Group AG . Since then, analysts say, more startups have sought to carry that momentum into the commercial market with enterprise apps built on top of ChatGPT’s language models or leveraging similar generative AI capabilities. 

“Most of the discussion has been around how generative AI works ‘out of the box’ since anyone can access a system set up for public use, directly on the internet,” said Bret Greenstein, a partner at consulting firm PricewaterhouseCoopers LLP, who covers data, analytics and AI. 

But beyond the current online craze, he said, countless enterprise software apps can be built by fine-tuning the tools’ underlying language-recognition models with a company’s own data. “Generative AI has crossed an important threshold in the last year,” Mr. Greenstein said. “It has gotten very good at many of the tasks that knowledge workers perform in every industry.”

All told, the global generative AI market is expected to reach $109.37 billion by 2030, growing at a compound annual rate of 34.6% over the next seven years, according to market research firm Grand View Research Inc.

While many companies are taking steps on their own to integrate ChatGPT into existing tech stacks, startups are hoping to also sell them on a range of customized tools.  

“We are working with customers across the board, from ride-hailing companies to sports franchises, from appraisers to technology companies, and indeed, the banking, finance, and insurance brands,” said PD Singh, vice president of software products at SambaNova Systems Inc., a generative AI startup based in Palo Alto, Calif. 

SambaNova, an early developer of the pre-trained foundation models that underpin language-recognition technology, recently launched a suite of “enterprise-ready” generative AI systems that includes ChatGPT chatbots trained specifically for banks, law firms, healthcare providers and other sectors, Mr. Singh said. 

The six-year-old startup’s last fundraising round was a $676 million Series D round in April 2021, which lifted its private-market valuation above $5 billion. New and returning investors included Temasek, GIC, Intel Capital, GV, and funds and accounts managed by BlackRock.

Given complaints that ChatGPT can produce false or unintelligible output, Mr. Singh said he recommends companies test new generative AI tools in a staged manner, introducing them initially to a small set of users, rather than a full rollout across the enterprise.

Typeface, a San Francisco generative AI startup that makes software designed to generate customized marketing and other promotional materials, recently unveiled a $65 million oversubscribed seed round, including funding by Lightspeed Venture Partners, GV, Microsoft Corp.’s venture fund M12, and Menlo Ventures.

Abhay Parasnis, Typeface’s chief executive, said the startup uses multiple generative AI platforms, including GPT and Stable Diffusion, to build custom AI models for each corporate customer. “We built an enterprise-grade application from day one,” he said, adding that current users include marketing, advertising, sales, human resources and customer support departments at businesses across industries.

Beyond customer-facing services, startups are also developing generative AI tools for internal business administration. RFPIO, a Beaverton, Ore.-based startup with $26.5 million in investor funding, recently launched a GPT-integrated software app designed to quickly check grammar, spelling and punctuation in written responses to requests for proposals, security and due diligence questionnaires and other documents.  

More than a simple spell check feature, RFPIO’s software can also convert answers from passive to active voice, smooth out awkward phrasing and improve readability, elaborate on key points, summarize large blocks of text and more, said AJ Sunder, the startup’s co-founder and chief information and product officer. He said the startup has nearly 2,000 commercial customers worldwide. 

“We are fine-tuning the language model underlying ChatGPT to create and edit responses based on trusted and verified content within our customers’ systems,” Mr. Sundar said. 

For better or worse, he said, ChatGPT was trained on public material that was scraped off the internet and carries some inherent risk of producing inaccurate results. One reason companies are being cautious around generative AI is that “the underlying technology is still nascent and evolving rapidly,” Mr. Sundar said.

Write to Angus Loten at Angus.Loten@wsj.com