However, this amount includes large sums of money from corporate backers, such as Microsoft's injection into OpenAI and Amazon's funding for Anthropic. When narrowed down to traditional VC investments, the amount of funding for AI startups in 2023 was much lower, at a pace comparable to total funding in 2021.
Pitchbook senior analyst Brendan Burke said in a report that venture capital funding will go into “underlying core AI technologies and their ultimate vertical applications, rather than general-purpose middleware across audio, languages, images, and video.” He pointed out that he was becoming more concentrated.
In other words, a GenAI app that helps businesses increase e-commerce sales, parse legal documents, and maintain SOC2 compliance is probably just an app that makes the occasional nifty video or photo. is a more reliable option.
Clay Bavor, co-founder of Sierra, said AI startups are moving toward a B2B model not necessarily because of the cost of compute or cloud APIs, but because they want to target specific customers and iterate their products based on that feedback. He stated that he believes there is a high possibility of benefits from doing so. “I think everyone, myself included, is pretty optimistic that these AI models will get better and the costs will go down,” Baber says.
“There is real power in having a clear problem to solve for a specific customer,” he says. “And you can get feedback saying, 'Did this work?' Does this solve your problem? And if you can build a business around that, that becomes very powerful.”
ChatGPT has sparked an AI boom in part because it can agilely generate code one second and compose a sonnet the next, but Arvind Jain, CEO of AI startup Glean, says the nature of the technology remains He says he prefers narrow tools. According to him, on average, large companies use more than 1,000 different technology systems to store corporate data and information, so there is an opportunity for many smaller companies to sell their technology to these companies. is being born.
“In this world we live in, we have a bunch of fundamentally functional tools, each solving a very specific need. That's the way of the future,” says a Google searcher who has been searching for more than a decade. Mr. Jain, who has been involved in the project, says: Glean powers your workplace search engine by connecting to a variety of corporate apps. Founded in 2019, the company has raised more than $200 million in venture capital funding from Kleiner Perkins, Sequoia Capital, Coatue, and others.
Error checking
Tailoring generative AI products to serve enterprise customers presents challenges. Errors or “hallucinations” in systems like ChatGPT can have more serious implications in corporate, legal, or medical settings. Selling Gen AI tools to other companies also means meeting that company's privacy and security standards and, in some cases, legal and regulatory requirements for the field.
“It's another thing for ChatGPT and Midjourney to be creative for the end user,” Bavor said. “It’s quite another for AI to be creative in the context of business applications.”
Bavor said Sierra has “spent a lot of effort” establishing safety measures and parameters to ensure it meets security and compliance standards. This includes using AI to adjust Sierra's AI. He explains that while it uses an AI model that generates correct responses 90% of the time, much higher levels of accuracy can be achieved when layered with additional technology that can find and correct some errors.
“AI systems need to be well-established for enterprise use cases,” says Jain, CEO of Glean. “Imagine a nurse in a hospital system using AI to make some decisions about patient care. You can't go wrong.”
Unpredictable threats to small AI companies selling their products to enterprise customers: A giant generation of AI unicorns like OpenAI, with fast-growing sales teams, have been built by idiosyncratic startups. What if you decide to deploy the exact same tool?
Many of the AI startups WIRED spoke to are moving away from full reliance on OpenAI's technology by leveraging alternatives like Anthropic's Claude or open source large-scale language models like Meta's Llama 3. trying to break away from. Some startups intend to eventually build their own AI technology. However, many AI entrepreneurs are being forced to pay for access to OpenAI's technology while potentially competing with it in the future.
Peiris, a Tome resident, considered this question and said he is currently focused specifically on sales and marketing use cases and is “amazed by the quality of production we produce for these people.”
