AI in Action: Expenses
94% of Businesses Reported AI Has Helped Reduce Operating Costs.
The stories in this chapter are fictitious and meant to illustrate concepts. Any resemblance to actual companies, projects, or individuals is an unintentional coincidence.
The Pub Cut Its Food Waste in Half
James owned an Irish pub called The Clover. The food was great and the regulars were loyal. But every week, he watched hundreds of dollars worth of perfectly good ingredients go straight into the trash. Not leftovers from plates, but product that expired before it ever made it onto one.
He estimated that roughly 10% of his food purchases were ending up as waste. In a business where profit margins are notoriously thin, often between 3% and 9%, that kind of waste could mean the difference between a profitable month and a losing one. After fifteen years in the business, James had solid instincts. He stocked up on corned beef before St. Patrick's Day and doubled his dessert order around Valentine's Day. The obvious patterns were already accounted for. But the waste kept adding up, and he couldn't figure out where the remaining blind spots were.
So James turned to AI. He uploaded several months of purchasing records, sales data by dish, and his weekly waste logs and asked it to find the patterns he was missing.
The results surprised him. AI was remarkably attuned to differentiating a random demand spike from the start of a real trend. It flagged what James now fondly calls "The Great Guinness Beef Stew Week" - a one-week surge in orders that he had chased with extra inventory, only to watch it all spoil. AI recognized the characteristics of a random blip and would have told him to hold steady. On the flip side, AI knew the characteristcs signaling the start of a new trend. Back in 2024, there was gradual decline in dairy-heavy dishes that he had dismissed as normal variation. It turned out a wellness influencer had posted a viral TikTok about cutting dairy, and the message had resonated with a meaningful portion of his customer base. James kept ordering the same amounts. But week after week he would see pounds of cream and cheese expire. AI identified the early signs of a persistent shift and would have flagged it before the waste piled up. There was still a lot that the AI didn't know about running The Clover, but he trusted that this kind of intelligence could be a valuable input to ordering decisions.
James was ready to put the AI to work. He asked it to take everything it had found and help him figure out the right amount to order each week going forward. AI ran thousands of simulations factoring in traffic patterns, seasonal shifts, and random variability to recommend ordering ranges for each ingredient.
The simulations didn't produce a single "correct" number. They gave James a range for each ingredient along with the trade-offs: order this amount and there's a 5% chance of running out; order a bit more and waste goes up slightly but stockouts nearly disappear. He could see the risk on both sides and make informed decisions.
James also used AI to brainstorm creative ways to move ingredients approaching expiration. The AI suggested rotating "chef's specials" built around whatever needed to be used first, not as a last resort but as a genuine menu feature. Customers loved the variety, and the specials gave James a pressure valve for the weeks when demand didn't perfectly match the forecast.
Crucially, AI didn't make the final calls. James still overrode recommendations when his experience told him something different. He knew when a catering order or local event would drive unexpected traffic. The system worked best as a combination: AI processing thousands of data points to surface the patterns a human couldn't spot, and James applying the judgment that only comes from fifteen years behind the bar.
Over several months, The Clover's food waste dropped by roughly half. For a pub operating on tight margins, those savings flowed almost directly to the bottom line - thousands of dollars a year that had previously been thrown in the trash.
The Baker Negotiated an Unthinkable Lease
Mary owned Golden Flour Bakery, and it was the kind of place that people built their weekends around. Customers would drive many miles just to pick up her sourdough and almond croissants. She had been in the same location for over a decade and was a fixture in her local community.
So when her lease came up for renewal and the landlord proposed a significant rent increase, Mary felt stuck. The landlord knew how much the location meant to her. His justification for the increase was straightforward: the property generated strong foot traffic, and Mary's business benefited from that. He positioned it as a fair price for a premium spot.
Mary didn't fully agree with his logic, but she wasn't sure it was worth fighting over. She had limited knowledge on how to even conduct such a negotiation and almost signed it just to avoid the confrontation. What was a couple thousand dollars in the grand scheme of things?
But then she wondered ... before she made a decision, could AI help her do some research and think through her options?
Mary started by simply describing her situation to AI and asking how she might approach the negotiation. Rather than jumping straight to a script, AI walked her through the kind of information that would make her case more compelling: comparable lease rates in the area, her business's contribution to local foot traffic, and the concept of "anchor tenants" versus "in-line tenants" in commercial real estate.
That last distinction turned out to be the key. In commercial real estate, an anchor tenant is a business that draws people to a property, the reason other tenants benefit from being nearby. An in-line tenant is one that benefits from the traffic the anchor creates. The landlord had framed Mary as an in-line tenant who owed her success to his property. The data suggested the opposite.
Mary gathered foot traffic data from her point-of-sale system showing peak hours and customer volume, a sample of customer reviews mentioning that people traveled specifically to visit her bakery, and comparable lease rates from nearby commercial listings she found online. She fed it into the AI and asked it to help her build a case.
After a few rounds of back-and-forth refining the tone, adjusting the emphasis, and adding details, AI produced a document that exceeded anything Mary could have created on her own. It wasn't just a list of bullet points. It was a polished, professional narrative with her bakery's branding that laid out a clear argument: Golden Flour Bakery was an anchor tenant that drove foot traffic to the property, not a beneficiary of it. The document cited her customer data, the comparable lease rates, and the economic reality that replacing a destination business with a generic tenant would likely reduce traffic for every other business on the block.
When Mary walked into the negotiation, she wasn't bluffing or posturing. She had data, a coherent argument, and a professional presentation backing it up.
She and the landlord ultimately agreed to terms that kept her in the same location at a rate that was fair to both sides, saving her thousands of dollars a year compared to the original proposal. That was money she could reinvest into the customer experience and her growing business.
AI didn't replace professional advice. For a more complex or contentious negotiation, Mary may well have got an attorney involved. But for researching her position, organizing her data, and presenting a compelling argument, it helped her walk into the room prepared. For the first time, the small business owner sitting across the table from a more experienced negotiator had the tools to prepare with confidence.