The Steak Flavor Experienced at the Hotel... AI Is the Chef [AI Revolution](95)

Foodtech Startup 'Beyond Honeycomb'
Combining AI for Optimal Recipes with Automated Robots
Versatile Grill Chef from Samgyeopsal to Tteokgalbi
Gaining Attention with AI-Driven Jjapaguri... North America Expansion Next Year

A thin brown strip formed by the Maillard reaction (a chemical reaction that occurs when the surface of meat browns at high temperatures, producing a special flavor) wraps around the meat. When the steak is sliced, it shows a uniformly light pink color from the outside to the inside. This color can be seen only when heat is evenly distributed, like in the sous vide method (a cooking technique where food is sealed in a bag and slowly heated in water at a precisely controlled temperature) used in high-end restaurants. Taking a bite of the meat, a rich, savory oil flavor spreads throughout. Despite its thick cut, it chews tenderly. The chef who grilled this five-star-worthy steak is none other than artificial intelligence (AI). Beyond Honeycomb, a food tech startup, has developed an AI solution specialized in grilling.

Jung Hyun-ki, CEO of Beyond Honeycomb, is slicing a steak grilled with 'Grill X'.
[Photo by Choi Yu-ri]

"Cooking is science"

Jung Hyun-ki, CEO of Beyond Honeycomb, started the business with the philosophy that "cooking is science." He believes that by quantifying taste, the optimal recipe can be found. Since the flavor of food traditionally depends on the chef’s hands and intuition, he saw an opportunity to innovate through technology.

The main menu was chosen to be grilled dishes. Although difficult to quantify, grilling is a popular cooking method. To master grilling, they developed an AI model that quantifies the taste of food in real time during cooking. First, molecular sensors quantified aspects such as the Maillard reaction, juice loss, and fat condition of delicious dishes. The AI was then trained with cooking data to reproduce the same taste.

However, acquiring cooking data was not easy since data can only be generated by actual cooking. To overcome this, CEO Jung designed a lightweight model that learns from hundreds of data points. After finding the optimal recipe with AI, another challenge arose: even with the recipe in mind, actual implementation was a different matter. This led the company to develop 'Grill X,' a robot that automatically grills meat. Jung explained, "Our company was formed by former Samsung Research members, so we handled everything from AI model design to robot assembly ourselves," adding, "With Grill X, we automated chef-level cooking and cut cooking time in half."

Jung Hyun-ki, CEO of Beyond Honeycom. Photo by Yongjun Cho jun21@

Goal to supply AI chefs to every home

Beyond Honeycomb supplies Grill X on a subscription basis. They manage dining services ranging from corporate cafeterias at companies like Naver and Neowiz, to apartment resident meals, and even dining services at Andaz Hotel. As word spread, their client base grew. They are developing AI chef solutions with a pork belly grilling franchise, Hanam Dwaeji Jip, as well as a third-generation Tteokgalbi (grilled short rib patties) restaurant. The AI learns the unique flavors of each restaurant by digitizing their distinctive tastes, enabling it to replicate traditional flavors.

The next step is to enter the U.S. market, which is not only diverse in grilling dishes but also the world’s largest market?13 times bigger than Korea. They already glimpsed success potential at CES 2022, the world’s largest electronics and IT exhibition. At the proof of concept (PoC) stage, their AI recreated 'Jjapaguri' (ramyeon dish) featured in the movie Parasite, attracting significant attention. CEO Jung said, "There is a much more severe labor shortage for grill cooks, known overseas as Line Cookers," and added, "We plan to expand starting with North America, then Australia, Singapore, and Europe."

The mid-to-long-term goal is to develop a home-use AI chef. The AI would analyze ingredients available at home and grill them using the optimal recipe. To achieve this, more cooking data must be collected and the AI model further refined. Jung expressed confidence, saying, "The optimal menu or cooking method for meat stored in a home refrigerator for one day versus four days will be completely different," and added, "We aim to automate home dining life and become the global number one in cooking AI."

Jung Hyun-ki, CEO of Beyond Honeycom. Photo by Yongjun Cho jun21@