In a mid-sized restaurant kitchen, the difference between a profitable month and a forgettable one usually hides inside a stack of supplier invoices that nobody has time to read. Tomato prices drift up by 8 percent. A distributor quietly swaps a SKU. A line cook over-orders proteins for a slow Tuesday. By the time the accountant catches it, the margin is gone. Ruka.ai, a Santiago-based startup founded in 2023, is betting that an AI agent reading those invoices in near real time can give Latin American restaurant operators something they have rarely had: a live view of food cost while there is still time to do something about it [Ruka.ai].
The product itself is narrow and, on paper, useful. Ruka describes it as software that automates invoice processing, monitors input prices, and optimizes food cost in real time, with customizable alerts when something moves [Ruka.ai]. A second description from the Latin American startup database Descubre.vc adds that the platform digitizes purchase processes and connects restaurants with their suppliers, generating reports on the back end [Descubre.vc]. In practice that means an operator photographs or forwards an invoice, the agent extracts the line items, and the system flags variances against historical prices and recipe costs. The wedge is unglamorous accounting work that owners hate doing and rarely do well.
The bet
Founder and CEO Camilo Silva Caviedes has framed Ruka's mission, on his LinkedIn, as catalyzing economic growth in Latin America by giving small and medium-sized businesses access to AI tools the enterprise tier has had for years [LinkedIn]. That is the strategic bet underneath the restaurant wedge. Restaurants are the beachhead because their pain is acute, their margins are thin, and their data is messy enough that automation pays for itself quickly. If the agent architecture works for a 12-seat bistro in Providencia, the same plumbing should extend to bakeries, hotels, and eventually any SMB that buys physical inputs from a rotating cast of suppliers.
The Latin American context matters here. Restaurant tech in the United States is crowded with incumbents like Toast and MarginEdge, but in Chile, Mexico, and Colombia the typical mid-market restaurant still runs procurement on WhatsApp threads and Excel. A locally-built tool that speaks Spanish, ingests the formats Chilean distributors actually send, and integrates with regional electronic invoicing regimes (Chile's SII e-invoice mandate is one of the strictest in the hemisphere) starts with a structural advantage that an imported product would take years to replicate.
Why it could matter
The back of the envelope is what makes this interesting. A typical full-service restaurant runs food cost at roughly 30 percent of revenue. On a restaurant doing US$600,000 a year, that is US$180,000 of inputs. Industry rule of thumb is that disciplined cost control, catching price creep, reducing waste, fixing portioning, can recover 2 to 4 percent of food spend. Call it US$5,400 a year on the low end (estimated). If Ruka prices at something like US$80 to US$150 per location per month, the payback for an operator is measured in weeks, not quarters. Multiply across the roughly 90,000 formal restaurants in Chile, Colombia, and Peru combined (estimated, based on national chamber figures), and the regional SAM for a single vertical is in the low hundreds of millions of dollars annually. That is not a venture-scale outcome on its own, but it is a real business, and it is the kind of wedge that funds expansion into adjacent SMB verticals.
The team
Ruka lists four co-founders: Camilo Silva Caviedes as CEO, Benjamin Vega Ruiz as CFO, Lorenzo Verdugo Varoli as Chief Growth Officer, and Enzo Cristóbal Zerega Cruzat. Silva Caviedes has prior startup experience at 500 Startups, Etiner Labs, and Etiner [RocketReach], and has written publicly about pivoting an earlier company before landing on the current thesis [LinkedIn]. Vega Ruiz is an industrial civil engineer studying data science [LinkedIn], a relevant combination for a company whose moat will live or die on extraction accuracy from messy invoice PDFs. Verdugo Varoli runs growth [LinkedIn]. The team is technical, regional, and on its second swing, which in a market that rewards distribution stamina more than novelty is a reasonable shape.
The honest counterfactual
The risk worth naming is competitive compression. Restaurant cost-control software is not an empty category, and the global incumbents have deeper pockets and the option to localize whenever Latin America looks attractive enough. Bears would also note that invoice-extraction AI is becoming commoditized as foundation models improve, which compresses the technical moat toward zero. The bull answer, supported by what Ruka has actually built, is that the durable advantage is not the OCR layer but the integration graph: the supplier connections, the SII compliance work, the alert logic tuned to Chilean buying patterns, and the trust earned one restaurant at a time [Descubre.vc]. That is slow, local work, and it is exactly the kind of work that imported software is bad at.
What to watch
The next twelve months are about three things. First, a disclosed seed round, pre-seed Latin American SaaS companies that show clean retention typically raise within 12 to 18 months of product launch, and Ruka is in that window. Second, a customer count. The website lists multiple tenant subdomains (casaenelaire, satira, alaire) which suggests live deployments [Ruka.ai], but a public number would let outsiders judge the slope. Third, vertical extension. If Ruka stays restaurant-only, it is a tidy regional SaaS. If the agent framework genuinely ports to bakeries, hotels, and clinics, the story gets considerably larger.
The incumbent Ruka has to beat is not a software company. It is the spreadsheet, specifically the Excel file that the owner's cousin updates on Sundays. That sounds like a soft target. It is not. Excel is free, infinitely flexible, and already trusted. Ruka's pitch has to be that an AI agent doing the same job at 3 a.m., without complaint and with a real-time alert when the cost of beef moves, is worth paying for. On the unit economics, that case is gettable. On the execution, the next funding announcement will tell us a lot.
Watts Lindqvist covers climate, energy, and the unit economics of decarbonization for Startuply.