TORONTO — Businesses that sell to other businesses are primarily using artificial intelligence to juice their own operations, and most are building their own technology to pair with the tools they’re buying, according to a new survey. Here’s what you need to know.

The source: Toronto-based investor Georgian commissioned the survey of 601 executives at global B2B firms with up to $200 million in annual recurring revenue. 

Function: Four-fifths of respondents said their firms are using AI to boost internal productivity, and 26 per cent said getting more out of their teams was the main reason they adopted the technology. Almost half said they’re most likely to be deploying tools that help generate insights from their data. 

Firms are also adopting AI products with which their business clients interact. Over a third are employing chatbots and other customer service tools, while 29 per cent use the technology to offer recommendations. But they’re less sold on the returns—just 17 per cent are driven by the expectation that AI could help them generate more revenue. Companies that make and sell generative tools have said businesses are adopting them in pursuit of efficiencies, rather than new ways to grow. 

Form: Most B2B firms are buying AI products, but also building their own. 

Some 50 per cent of the 300 R&D executives surveyed said their firms were using these commercially-available large language models (LLMs); ChatGPT maker OpenAI was by far the most popular seller, followed by Google and Anthropic. Another 41 per cent were tweaking or training open-source LLMs for their own purposes, drawing mostly on Meta’s Llama offerings.

Despite OpenAI’s prominence, many firms are hedging their bets, trying out models from multiple providers and developing their products in ways that allow them to switch as the technology continues to improve.

But AI isn’t free. R&D executives said they’re spending more on data storage and processing and the hardware to run the technology, as well as training their teams to use it. To cap those costs, most are only applying AI to high-priority projects. 

Firms are also worried about the security and privacy of the data they’re feeding into AI tools, with 60 per cent of R&D executives citing that as a concern, while 49 per cent said they’re worried about whether they’re getting a return from the technology.