Companies went all in on AI this year, and are now reckoning with heavy bills. One startup wants to get under the hood and figure out which ones are worth it.
Larridin, a San Francisco-based startup, just raised $30 million in Series A funding to give companies a clear picture of how the money they spent on AI is — or isn’t — working for them.
Jeff Kearl, a general partner at Pelion Ventures, which led the funding round, told Business Insider that AI has gone from being a small, even “nonexistent,” expense for businesses to one of the top five to 10 most expensive items on their ledger.
“All of a sudden they’re like, ‘Wow, that’s a lot of money going to AI. We run a lot of apps. We have a lot of models. Where does it all go, who uses it, and how do I make sense of that?” Kearl said.
Larridin’s main product is a measurement platform that tracks the extent to which AI improves employee productivity and identifies opportunities for better adoption. The software integrates with companies’ internal systems to analyze employees’ AI usage, including their prompts to different AI tools, and determine who is using AI effectively, who needs more training, and how much time they’re saving.
Cofounder and CTO Ameya Kanitkar said the platform protects employee privacy by analyzing their AI interactions without storing their contents, instead generating scores and high-level insights.
Always a need for a third-party measurement platform
Russell Fradin, CEO and cofounder of Larridin, previously worked at Comscore, a media and advertising measurement platform. He said that, in his experience, whenever a lot of money moves into a category, someone builds a third-party measurement platform for that category, evidenced by Comscore, Nielsen, or Datadog.
“We said, look, there’s going to be a ton of AI spend,” Fradin said. “There’s going to be a third-party audit measurement company, and let’s try and be that company.”
The company has a third cofounder, Jim Larrison, who worked with Fradin at Comscore.
Larridin’s $30 million Series A was led by Pelion Ventures, Andreessen Horowitz, Gradient Ventures, GFT Ventures, W23, Thayer Investment Partners, DG Daiwa Ventures, Alumni Ventures, and Cougar Capital. A16z and Gradient Ventures were part of its $17 million seed round.
Alex Rampell, a general partner at A16z, said one thing that convinced him about Larridin’s business was that it addressed a need in enterprise AI for a sort of “AI boss,” who could tell staff what they should or shouldn’t do with AI.
“If you’re in sales, somebody should nudge you and say, ‘Hey, call this account. Don’t call that account.’ If you’re in coding, ‘fix this bug, don’t fix that bug,'” Rampell said.
“But underlying that was the same idea, which was like, people are using AI tools to do all sorts of things. Are they doing the right thing? Is the outcome actually appreciably different? Are you getting any kind of ROI?” he added.
Fradin said that business picked up around the second quarter of this year, when more companies became concerned about AI costs. The company went from having one enterprise customer in December last year to 50 now.
Post-Series A, he said he wants to double Larridin’s 20-person head count, adding more sales, engineering, and marketing staff.
AI spend behavior has evolved
Larridin’s Series A comes after months of twists and turns in the field of AI spending.
In the first few months of the year, tokenmaxxing was in vogue, which led companies to urge their employees to get on the hype train and experiment with AI — essentially a mandate to accelerate adoption or be left behind.
They urged their employees to use AI tools as much as possible and burn as many tokens as they could, with some, like Duolingo, even adding AI use as a performance metric for their employees. Uber CTO Praveen Neppalli Naga went viral in April when he said Uber had already burned through its entire Claude Code budget for the year.
AI budgets ballooned. At least 17 companies, from Salesforce to Oracle, announced AI-related layoffs, with Oracle specifically saying in June that AI deployment and adoption have reduced its workforce.
But the sentiment shifted sometime in the middle of the year, when ugly AI bills started rearing their heads.
Companies started paying more attention to getting the best return on investment from their AI spending. Modelmaxxing, which, as Business Insider wrote about in July, involves routing prompts to different AI models based to maximize token use, emerged as a tactic to save costs. More businesses started using cheaper open-source models developed by Chinese AI labs, such as Moonshot AI and Z.ai.
