Areas resistant to automation – like legal services and healthcare – are attracting novel applications that could even displace human workers, according to a Bank of America report.

It’s old news by now that business processes are being transformed into artificial intelligence (AI) operations. Companies such as Salesforce, Hubspot, and Microsoft unveiled a slew of AI “agent” capabilities this year for business functions such as customer service and sales.

Now, a wave of privately backed software firms are using AI to build brand-new applications from the ground up, to re-invent areas traditionally resistant to technology such as legal services and healthcare, according to Bank of America.

“We expect AI-native startups to proliferate over the next several years and increasingly cannibalize the significantly larger $12.3 trillion US Services industry,” writes the firm’s software and services analyst, Alkesh Shah, in a December 13 report based on a virtual conference held last week to discuss trends in AI.

Also: Microsoft introduces 10 AI agents for sales, finance, supply chain in Dynamics 365

Shah likens the latest crop of venture-backed startups to the Internet’s early days, stating that they are “emerging like it’s 1996.”

The startups featured at the conference include San Francisco-based Hippocratic AI, founded in 2022, which uses large language models to automate non-diagnostic healthcare tasks such as assessment of individuals to determine the need for an emergency room visit.

The analysts heard from Hippocratic’s co-founder and CEO Munjal Shah that “adoption is ramping and agents receive high satisfaction scores, while costing significantly less at $9-10/hour vs $50-90/hour for human nurses.”

The company claims AI-based software can perform some tasks better than human nurses can, such as identifying which over-the-counter drugs may be unsafe for a given patient, or what dosages of such medications could prove toxic.

Another startup, vLex of Barcelona, Spain, uses large language models to, among other things, generate hypothetical arguments that opposing counsel in a lawsuit might use, to help lawyers and paralegals strategize.

The company’s software, Vincent AI, can also speed searching through numerous documents. “The time required to analyze privacy law regulations across seven different countries could be reduced to minutes from weeks,” write the Bank of America analysts. The Vincent AI platform has two million users among eight of the world’s top ten law firms.

Both companies are examples of automation that may start to eat into human jobs, writes Shah. “It may become increasingly difficult to compete with AI agents. According to the US Bureau of Labor Statistics, there are approximately 3.3 million registered nurses ($41/hour average pay), 55,000 medical scribes ($18/hour average pay), 859,000 lawyers ($70/hour average pay), and 366,000 paralegals and legal assistants ($29/hour average pay).”

Also: Enterprises are struggling with what to do with Gen AI, say venture capitalists

The Bank of America report suggests that, broadly speaking, uses of generative AI are starting to move beyond the sales and customer “experience” domain to incorporate more “vertical” functions specific to industries.

The rise of commercial software packages focused on AI may help bridge the divide for the large portion of enterprises that struggle on their own to know how best to use the technology.

https://www.zdnet.com/article/ai-software-startups-set-to-take-over-12-trillion-us-services-industry/