TSUNAN SAKE BREWERY Presents AI-Assisted Development of Table Rice Sake at the 78th Annual Meeting of the Society for Biotechnology, Japan
TSUNAN SAKE BREWERY President Kengo Suzuki spoke at the 78th Annual Meeting of the Society for Biotechnology, Japan, on September 16, 2026, at Hokkaido University’s Sapporo Campus. In a symposium organized by the society on combining microbial interactions with generative AI in fermentation and brewing, he presented “Development of premium sake using table rice through generative AI technology.”
The talk explained how Smart Brewing, the generative AI system the brewery has used since December 2023, was applied to GO GRANDCLASS Uonuma Koshihikari Edition, a junmai daiginjo made entirely from Uonuma Koshihikari, a rice variety grown for eating rather than specifically for sake brewing.
Talk overview
– Meeting: 78th Annual Meeting of the Society for Biotechnology, Japan
– Organizer: The Society for Biotechnology, Japan
– Dates: September 15¨C17, 2026
– Venue: Sapporo Campus, Hokkaido University
– Talk: Morning of September 16, 2026
– Title: Development of premium sake using table rice through generative AI technology
– Speaker: Kengo Suzuki, President and Representative Director, TSUNAN SAKE BREWERY
Why Koshihikari is difficult to brew with
The brewery has been brewing with Uonuma Koshihikari since 2021. Suzuki described three difficulties. Steamed Koshihikari is sticky, so the grains clump during koji-making and the koji mold has trouble spreading, which tends to produce koji with weak enzymes. The rice contains more protein and fat than sake rice, which can lead to off-flavors. It also dissolves poorly in the fermenting mash, which can upset the balance between saccharification and fermentation.
How Smart Brewing works
Brewery staff enter questions, and the system answers using three layers of knowledge: published brewing science, the brewery’s own production records and analytical data, and site-specific know-how. It combines a large language model with retrieval-augmented generation (RAG). Final decisions are made by people: the toji (master brewer) checks each AI proposal against brewing science before deciding what to do.
The talk gave three examples:
– Fermentation planning: A conventional approach to table rice (weak koji, maximum 15¡ãC) left sugar unfermented, the Baum¨¦ reading never reached zero, and the sake meter value ended at -8.9. The AI-proposed approach (high-enzyme koji, maximum 11¡ãC) reached Baum¨¦ zero 16.8 days after the final addition and finished at a sake meter value of ¡À0.
– Temperature control: The proposed profile, capped at 11¡ãC, was designed to limit off-flavors from Koshihikari’s protein and fat. In actual brewing, the rice dissolved fully and the amino acid level stayed within the predicted range.
– Comparing finished sake: Given final mash data from two junmai sakes, the AI suggested food pairings, serving temperatures, and options such as blending, aging, and pasteurized versus unpasteurized versions.
The brewery also measures four aroma and four taste attributes using a sensory evaluation system from ProfilePrint of Singapore. Among the GO brand’s products, GO GRANDCLASS Uonuma Koshihikari Edition scores lowest for bitterness and highest for freshness.
GO GRANDCLASS Uonuma Koshihikari Edition is polished to 50% of its original weight and has 16% alcohol. It won a Gold award in the Junmai Daiginjo category at KURA MASTER 2024 and a Platinum award at the Milano Sake Challenge 2025.
GO GRANDCLASS product page:
Other projects
Suzuki closed with projects that combine local products, fermentation and generative AI:
– GO LUNAR: a trial junmai daiginjo made with Uonuma Koshihikari and fermentation technology using Euglena, aged in wine barrels. The brewery has recreated a lunar production site, imagined for 2040, in a metaverse space.
– Green sake: a trial brew made with the brewery’s yeast and Euglena extract. Compared with a control, its organic acid profile was almost unchanged, while ethyl caproate, a ginjo aroma compound, rose from 0.2 mg/L to 0.9 mg/L.
– SAKESOME: nanoscale particles found in sake, named and studied by group company Regenesome Inc. using a microfluidic separation technique.
– Materials from sake lees: work on graphite and other materials derived from sake lees, using a spark plasma sintering (SPS) system under development at Okayama University of Science.
Comment
Kengo Suzuki, President and Representative Director, TSUNAN SAKE BREWERY
“Sake brewing is a very complex biological process in which koji mold and yeast work at the same time in a single tank. Toji have long read that complexity through experience. Our AI is not meant to replace that experience. It brings together published knowledge, the brewery’s records, and practical know-how to support the toji’s decisions. Brewing a junmai daiginjo from Koshihikari, a rice loved for eating, shows that limits in raw materials can be overcome through process design. We believe this approach can extend beyond sake to other fermented foods and even to biopharmaceutical production.”
Next steps
TSUNAN SAKE BREWERY will keep adding to the Smart Brewing knowledge base and its production data, and widen the range of work shared between AI and the toji, from planning a brew to deciding when to press. It will also continue trial brews such as green sake and GO LUNAR, and research on SAKESOME and materials from sake lees.
About TSUNAN SAKE BREWERY
TSUNAN SAKE BREWERY is a sake brewery in Tsunan, Nakauonuma District, Niigata Prefecture, one of the snowiest areas in Japan. It brews with natural water from the surrounding mountains, which rise to about 2,000 meters. Under its brand concept “Brew for Future,” it makes traditional sake from locally grown Gohyakumangoku sake rice and Premium Table Rice Sake from locally grown Uonuma Koshihikari.
Website:
