Austin-based Rebel Cheese maintains a lean team. But after a busy holiday season, CEO Kirsten Maitland realized the firm’s shipping partners were overcharging $250,000, by her estimate. 

Baffled by the sheer cost of these charges, she took to AI, devising an agent to audit invoices and compare them line by line against the firm’s contract with its shipping partners to flag discrepancies. The agent also processes photos of boxes, looking for bulging by as little as an eighth of an inch, which can trigger massive shipping overcharges. 

All in all, the agent has saved Maitland $400,000, a massive amount for a young firm—in which Shark Tank star Mark Cuban invested $750,000—that’s estimated to reach a $20 million valuation by year’s end.

“As a small-business owner, I do feel like AI gives us advantages similar to what large companies have,” Maitland told Fortune. “It levels the playing field and gives us similar opportunities.”

AI models today are advancing at a breakneck pace. And while a growing number of Americans shudder at the idea of an AI job apocalypse, others are diving headfirst into the technology, leveraging it to improve efficiencies and to answer complex questions.

How the AI saves $40,000 a month

Rebel Cheese’s AI agent works like this: Maitland has her executive assistant upload weekly invoices into the AI tool alongside a version of the company’s contract that specifies rates based on shipping specifications, such as zones, weights, and dimensions. The agent then flags every discrepancy, providing proof and data which Maitland then uses to dispute charges and secure credits from shipping partners.

The agent can also inspect boxes for physical abnormalities that might trigger price hikes, while auditing package dimensions and weight. Once it flags discrepancies, Rebel Cheese can make changes before shipping orders to avoid costly surcharges.

Maitland said the savings have dipped below $40,000 in recent months, but she believes that’s because her shipping partners are now more aware that her AI tool is catching overcharges.

AI agents differ from the typical large language model that people turn to for drafting an email or devising a recipe. Agents are autonomous AIs that can perceive their environment and complete tasks without a human. For Rebel Cheese, there’s still a human in the loop, with both Maitland and her assistant interacting with the agent to produce an outcome.

Why other firms are (or aren’t) deploying AI agents

Some companies today are deploying agents at scale, working alongside human workers to automate rote tasks, analyze data, and boost productivity. McKinsey’s 2025 State of AI report found that 62% of companies are actively implementing, or planning to implement, AI agents into their workflows. Some business leaders, like Nvidia CEO Jensen Huang, foresee 100 AI agents for every human employee in 10 years. 

But even as more executives grow bullish on the technology, a recent study from the National Bureau of Economic Research found that among 6,000 executives from firms across the U.S., U.K., Germany, and Australia, the vast majority see little impact from AI on their operations.

And even as AI agents grow more sophisticated—building out entire workflows in some cases—only 6% of companies fully trust AI agents to handle business processes, according to a Harvard Business Review report.

Still, for those small businesses implementing the technology, like Maitland’s, the payoff is high, and the expertise the agent provides is irreplaceable.

“I don’t think even if I did hire two full-time people, I don’t think they would even find all of these discrepancies because it’s just so much data,” she said. “It’s unmanageable honestly, and that’s where AI is really strong.”