AI That Watches What You Don’t Eat: Kikleo Uses Cafeteria Leftover Data to Fight Food Waste
In a bustling corporate headquarters canteen, lunch trays glide along a conveyor belt toward the dishwashing station. Before the plates disappear, a camera captures everything left uneaten: half a chicken breast here, untouched vegetables there.
Within seconds, artificial intelligence identifies each item and calculates not just what was wasted, but how much it cost and what environmental toll it took.
This is the vision that Vincent Garcia and Martin d'Agay have been pursuing since founding Kikleo in 2019. What started as a technology to speed up cafeteria checkout lines has evolved into a sophisticated weapon against one of the food industry's most persistent problems: waste. Now, with a fresh €3.5 million fundraising round led by Newfund, the Lyon-based startup is preparing to take its solution across Europe and the Atlantic.
"We want to transform data in the kitchen into action," Garcia said.
From Engineering School to AI Innovation
The stakes are enormous.
Food waste generates nearly 10% of global greenhouse gas emissions and costs the restaurant industry an estimated €132 billion annually. In France alone, the average European citizen throws away 132 kilograms of food per year—even as 33 million Europeans can't afford a complete meal each day.
Meanwhile, regulatory pressure is mounting. The EU has mandated that restaurants reduce food waste by 30% by 2030. Kikleo finds itself positioned at the intersection of compliance, cost savings, and environmental responsibility.
The technology behind Kikleo's solution draws on deep academic roots, combining computer vision and artificial intelligence. Both Garcia and d'Agay graduated from INSA Lyon, one of France's leading engineering schools, and they turned to researchers at the institution's LIRIS laboratory to help develop their most critical innovation.
"Our algorithm is trained on a million images of food, which allows it to automatically recognize any type of food that is present on the tray, on the plate, or in the production bins," Garcia said.
Food recognition alone wasn't enough. The real breakthrough came in the second part of their algorithm. "The first part, food recognition, is fairly traditional image recognition," Garcia noted. "The second part, recognizing and reconstructing the volume and weight of foods, is the whole thing that will be patented."
This volume estimation capability—developed in partnership with INSA Lyon researchers—is what allows Kikleo to move beyond simple identification to precise measurement of waste, translating uneaten food into exact financial and environmental costs.
A Market Ripe for Disruption
Kikleo operates primarily in collective catering: corporate cafeterias, school lunchrooms, hospitals, and government facilities. It's a sector that has lagged in digitalization, which presents both opportunities and challenges.
"These are sectors that have seen little digitization in recent years," Garcia said. "There is still a certain reluctance to embrace data and technology. The biggest challenge for us is to come up with a simple but effective solution to get them to use this data effectively."
For that reason, Kikleo's technology integrates seamlessly into existing workflows.
"In terms of installing the tools, we're not changing the way the restaurant operates," Garcia emphasizes. "It can be deployed very quickly and effectively within a week of being rolled out."
The system's dual-scanning approach captures waste at two critical points. In kitchens, staff scan large aluminum containers before disposal. In dining areas, "we will often have conveyors, which are moving walkways that bring back the trays," Garcia explained. "And there, they automatically pass under the camera."
The core products include:
- Dining room/tray scan (Guest Terminal/Conveyor Scan)
This smart camera-based system is installed in the dining room (e.g., on a tray conveyor or sorting table) and automatically detects, identifies, and quantifies leftover food items on guests’ plates or trays using computer vision and AI. It works autonomously once installed.
(see photo at the top of the article)
- Kitchen / back-of-house waste weighing & tracking
This connected-scale system digitalizes the weighing of kitchen-side waste, enabling more precise measurement of what’s discarded before serving. There are two modes: fully automated (camera + scale) or semi-automated (e.g., scale + tablet), depending on the kitchen setup.
- Analytics & Dashboard Platform
All data (from the dining room scans and kitchen waste) is consolidated into a web platform/dashboard for real-time monitoring of waste patterns, economic cost of losses, environmental metrics, modular data views, and action plans and advisory content tailored to the restaurant’s data.
- Support & Services
The company provides one-time or ongoing analyses, support in interpreting data, and help in designing reduction strategies. They also supply awareness materials and guidance for staff based on results.
"We have a whole team on our side dedicated to customer support, helping them understand how to use the dashboards, how to exploit the data, and how to transform all this data into tangible action plans, such as reworking portions, reworking quantities, reworking recipes," Garcia said.
The startup now has more than 250 restaurants equipped with its technology and has analyzed 5 million meal trays since launch. The company maintains a customer renewal rate above 95%, according to Garcia.
Reducing Waste To Fuel Growth
Kikleo reports achieving an average 20% reduction in food waste during clients' first year, with some locations reaching up to 65% reduction over time. That translates to savings of up to €50,000 per site annually.
Between 2022 and 2024, the company saved 240 tons of food from landfills—equivalent to 400,000 meals. For restaurant operators, that meant €2 million in savings. For the environment, it represented 1,000 tons of CO2 and 800 million liters of water not wasted on food production, according to Garcia.
The startup projects €12 million in revenue this year, with 90% coming from collective catering.
But Garcia and d'Agay aren't content with the French market alone. Kikleo is already present in Portugal, Belgium, and Luxembourg, with Spain and Italy next on the expansion roadmap. The United States presents both the biggest opportunity and the biggest challenge.
"We have requests from major universities in the United States, where we are already present," Garcia said. "We also want to be able to strengthen our position and send our solution directly across the Atlantic."
The Challenge of Going Global
The new round of funding was led by Newfund and also included money from Bpifrance and other historical shareholders. Part of the funding will be used to expand into new territories. The ambition is to reach more than 1,000 restaurants by 2027, quadrupling the current client base in just two years while opening numerous international markets.
But international expansion brings several hurdles for a company such as Klikeo.
The first starts with its core AI for food recognition. Different cuisines mean different foods, different presentations, and different challenges for the algorithm.
"Every time we export to a new country, we obviously have dishes and foods that may be a little different," Garcia said. "We have to adapt our algorithms, which will be done gradually in a semi-automated way."
The American market has proven particularly tricky. "Many foods have a tendency to be breaded; they are more difficult to recognize," Garcia noted, a detail that highlights the granular challenges of deploying AI in real-world food service settings.
The second major concern is maintaining service quality while scaling internationally.
To help with that, in 2024, the founders brought on Jean Dussaix, a seasoned veteran with 20 years of experience in collective catering. Dussaix has more than 20 years of experience at Sodexo, a global leader in food services, facilities management, and workplace experience solutions.
His roles included managing procurement strategies across 80 countries, giving him an insider’s understanding of how large food service organizations operate, buy, and innovate.
AI Evolution Continues
Just as adapting its AI poses a challenge in some ways, Garcia is hoping it will also facilitate expansion in other ways.
"Our goal, when exporting abroad, is to continue to maintain this link with all the new establishments that we will be onboarding," he said. "It's about being able to expand on a large scale while maintaining the quality of our services that we have today in France."
The startup's solution is to make the technology so intuitive that it can be deployed remotely.
"One of our big goals and challenges for the future is to have solutions where we have a screen on our solution that is intuitive enough that you can take it anywhere, send it anywhere in the world, and have it installed directly by the customer," Garcia said. "We don't need to go on site, so we can deploy on a large scale very quickly."
For Garcia, this represents a natural evolution.
"We have always been involved in AI. We have been working in AI for years," he said. "It was more traditional AI, whereas now we are implementing generative AI, which will enable us to accelerate our processes and deployment."