Matthew Kayser
Los Angeles restaurants operate in one of the most expensive and tightly regulated labor markets in the country. With minimum wages rising, break rules strictly enforced, and margins often measured in single digits, even small operational blind spots can quietly erode six figures a year.
That is where AI holds promise to solve problems that have plagued the restaurant industry for decades. Lavu, a fintech platform for restaurants, recently launched Marty, which is an AI-driven digital manager that closes those blind spots for restaurant operators.
Lavu’s CEO, Saleem Khatri, argues that most operators don’t lose money because they’re careless. They lose money because they can’t see what’s happening across their own data.
“Marty doesn’t run your restaurant during the rush,” he said. “It makes sure Monday morning decisions are surgical instead of gut-feel.”
The Saturday Night Problem
Khatri knows the drill. During a dinner rush, the general manager is rarely staring at dashboards.
“I’ve worked in restaurants,” he said. “During a Saturday rush, your manager is on the line, working expo, covering a no-show. They’re not checking whether one register is running 2.4 times the void rate of every other terminal. They’re not noticing that about 85% of their break-related alerts or potential scheduling issues may occur between 6 and 8 p.m.”
Marty is.
By Sunday morning, managers receive what Lavu calls a “Morning Deposit,” a plain-English summary of what happened, what it cost, and what to fix. In one restaurant group, Khatri said, evening meal-break risk indicators had increased by about 28% over three weeks, clustered squarely in the dinner window. Twelve employees were potentially affected by scheduling patterns that may not have aligned with break-timing guidelines. The fix was a staggered break rotation mapped to daily sales.
“One change,” he said. “$7,500 a year in potential compliance exposure, gone.”
The goal isn’t to replace managers, he added. It’s to give them clarity after the adrenaline fades.
Where the Money Hides
In California, where fast-food wages now start at $20 an hour in many jurisdictions, a two-point swing in labor percentage on a $10 million annual run rate equals $200,000. That math adds up quickly.
Khatri points to a recent analysis of 169 stores in which Lavu reviewed 4.17 million rows of POS, labor, and payroll data in a single quarter. The conservative annual recovery estimate: $1.86 million. The base case: $5.85 million.
The leakage wasn’t in one obvious place.
Revenue per labor hour ranged from $37 to $93 across the same portfolio, a 2.5x spread. Eighty-five stores were overstaffed by more than 36,000 hours annually. In one quarter alone, 9,558 unscheduled shifts were logged, costing $784,000. Kitchen overtime crept from 1.8 to 4.2 hours per week per employee in one group, generating $130,000 in premium pay in a single quarter.
“The money isn’t in one place,” Khatri said. “It’s spread across voids, comps, overtime, scheduling gaps, and discount leakage. No human cross-references POS, labor, and payroll at that speed. That’s why it gets missed.”
Compliance in a High-Risk Market
Los Angeles restaurants face some of the most complex labor rules in the country, from five-hour meal period requirements to daily overtime thresholds that trigger at eight hours, not just 40 per week.
“Most operators don’t know they’re potentially out of compliance until the lawsuit arrives,” Khatri said. “I’ve seen it happen. The compliance risks were present in the data for months, but nobody was looking.”
Marty cross-references clock data against California’s state-specific rules automatically, flagging potential break-timing issues and estimating possible financial exposure. In one group, about 85% of the system’s flagged break-timing alerts clustered between 6 and 8 p.m., a pattern Khatri called “a scheduling design flaw that could create ongoing compliance risk if not addressed.”
The system also reconciles inconsistencies between labor and payroll platforms. In one audit, 97.3% of employees were matched across systems, leaving 82 without a corresponding record in one or the other.
“That’s either a data gap or a possible compliance discrepancy,” Khatri said. “Both need answers.”
Winning Over Skeptics
Artificial intelligence has become a buzzword in nearly every industry, and independent restaurateurs can be wary of promises that sound too good to be true.
Khatri’s answer is straightforward: show them the money.
“Connect your POS. Ninety minutes. Within 48 hours, you get a Morning Deposit showing exactly where cash is hiding,” he said.
In one three-location group, Marty identified an estimated $127,000 in annual recoverable cash opportunities. The operator recovered about $41,000 in the first quarter after acting on several findings, according to Khatri—the monthly subscription: $499.
“Every finding shows the evidence,” Khatri said. “Which register. Which shift. Which pattern. Which dollar amount. We’re not saying ‘trust the algorithm.’ We’re saying: here’s Register 3, here’s the void pattern between 8:30 and 10:15 p.m., here’s the estimated $42,000 a year impact.”
Accuracy, he added, starts around 89% in week one and improves to 97% by week four as the system learns the operator’s patterns.
“Marty doesn’t replace judgment,” he said. “It arms it.”
The Five-Year View
Khatri envisions a near future in which restaurants rely less on periodic P&L reviews and more on continuous operational intelligence.
Today, data often lives in silos: POS tracks sales, labor platforms track hours, and payroll systems track costs. Few operators have the time or tools to connect all three.
“Marty already does,” he said.
The company recently introduced predictive notifications designed to warn managers before a shift begins that they are likely overstaffed or drifting into overtime based on historical patterns. The next phase, he said, is network benchmarking, allowing a 12-unit operator to compare performance against hundreds of anonymized peers in similar markets and revenue brackets.
“The end state is continuous operational auditing,” Khatri said. “Not quarterly P&L reviews. Not annual consultants. Daily intelligence that catches a potential $7,500 compliance exposure before it compounds for months.”
Khatri, who began his career scooping ice cream, frames the technology less as a futuristic leap and more as overdue infrastructure.
“What we’re building now is what I wish every operator had then,” he said. “Every data source a restaurant generates, turned into plain-English, dollar-denominated actions by 6 a.m. every morning. The digital GM that never misses a shift and gets smarter every week.”