When Tony Leopold, the chief technology and strategy officer at United Rentals, held a hackathon to test out a new artificial intelligence agent, he said he wanted to clear one final hurdle before debuting the tool at the equipment rental company’s annual management meeting in January.
“I am personally going to interrogate it and ask it every question I can throw at it,” Leopold recalls of his mindset before going into the four-hour hackathon. “I’m going to try to break it.”
But with three months of pilot testing preceding his stress testing efforts in December, Leopold felt confident enough to publicly announce on Wednesday the launch of the company’s new “Business Intelligence Agent,” which United Rentals built with cloud-software company Snowflake. The tool is now available to thousands of frontline employees across more than 1,600 branches that rent equipment serving the construction industry, industrials, utilities, and residential customers.
The tool allows employees to analyze financials, customer data, and operational questions using natural-language prompts, such as “which of my sales representatives are discounting the most?” or “which customers should I focus on today to collect revenue?” Previously, this information was shared through corporate reports or dashboards that an employee had to toggle through manually. With AI, United Rentals hopes to speed up the information-gathering process.
Leopold, who joined United Rentals in 2010 and held a variety of leadership roles before ascending to his current title late in 2024, says his framing is key. “We’re calling it the beta phase, because the one thing that’s very important—in particular with AI agents—is they’re not always perfect,” he adds.
United Rentals and Snowflake developed a “thumbs up, thumbs down” feature to encourage consistent user feedback, leveraging Snowflake’s Cortex Code, an AI coding agent designed to speed up the development, testing, and iteration of AI agents. When feedback is positive, the Business Intelligence Agent will surface similar responses in the future, at a faster pace. But for responses that fail to retrieve the right information, Leopold’s IT team works behind the scenes to tweak the prompt engineering and ensure the right data is available to be retrieved to address the user prompts.
This reflects an approach that’s evolved as more businesses deploy AI, especially as more complex agentic applications gain some traction. It isn’t enough to launch a new AI tool and make it available to thousands of employees with a dash of training. On the back end, technologists like Leopold are consistently mining human feedback and pushing through regular updates. Four out of every five users of the Business Intelligence Agent give it a thumbs up. “But we want 100%,” says Leopold. “That’s our aspiration.”
For Snowflake, which has recently inked separate $200 million deals with AI startups Anthropic and OpenAI to integrate their large language models directly into its platform, working with clients like United Rentals could help turn the tide on Wall Street’s fears that the company’s AI-based tools are not flourishing enough to meet investor expectations.
And for United Rentals, which ranks No. 285 on the Fortune 500, the new AI agent builds on a slow-yet-steady embrace of AI. The company largely stayed on the sidelines in 2023, in the wake of the late 2022 debut of ChatGPT, followed by some pilots in 2024 and expanded into broader deployments last year. AI tools that Leopold has rolled out include an internal chatbot built on Anthropic’s Claude, which had 4,000 users in the last month (United Rentals has close to 28,000 employees in total).
Last year, United Rentals also debuted “Manual Assist AI,” an application built with Amazon Web Services that allows service teams to access thousands of pages of manuals and address equipment issues far more quickly than if they had to call a hotline or thumb through the documents by hand. Leopold launched the tool in July and within just two months it had reached 4,000 monthly users.
He says that traditionally, it would have taken up to 18 months to reach this level of penetration. AI’s user interface, Leopold says, is “so much more intuitive that I think the adoption potential is so much higher than what we had through traditional web applications.”
United Rentals’ AI investments focus on three core priorities. There is the handling of administrative tasks, data entry, and other business work at the company’s vast rental network. Then, there is an effort to make the online digital experience more seamless for customers, which includes the development of an AI search strategy to optimize content for ChatGPT, Gemini, and other LLMs. AI commerce is in the earliest stages of development and brands like United Rentals are only just beginning to wrap their heads around the concept.
Finally, the company is deploying AI tools for corporate employees. One application currently in the mix is the use of AI to help draft employee performance reviews. United Rentals invested in some training to encourage this use case, coaching 1,500 employees in a single week.
“2025 was really learning how to scale, how to do it securely, how to do it with enterprise data and connecting into legacy systems—not just chatting on an enterprise agent,” says Leopold. “2026 and 2027 is where we’re really starting to harvest some of those wins.”
John Kell
Send thoughts or suggestions to CIO Intelligence here.
This story was originally featured on Fortune.com