Concept visual: Wild Bite Club. The cited IP filings do not confirm a product launch.
A new filing defines a food-development platform that models ingredient interactions and carries scientific, sensory, consumer and business constraints across the R&D workflow.
| Filed mark | NOTCO AI |
|---|---|
| Applicant | NotCo Chile AI SpA |
| Filing office | USPTO |
| Application | 50062775 |
| Filing date | 2026-08-20 |
| Nice classes | 9, 42 |
| Current status | LIVE — New application awaiting assignment to an examining attorney |
| Evidence level | FILING + MULTIPLE SIGNALS |
Commercial clue, not launch proof: The record protects a proposed market territory; execution, distribution and availability remain unconfirmed.
The Commercial Territory
NotCo Chile AI SpA has applied to protect NOTCO AI across software and product-development services that are unusually specific about the work of food R&D. The Class 9 language covers software for identifying, evaluating, selecting and optimizing ingredients, formulations and product compositions; modelling interactions among ingredients and product attributes; combining scientific, technical and consumer data; and generating formulation recommendations. Class 42 extends the scope into food-and-beverage research, new-product design and testing, ingredient discovery, technical consulting and online AI assistants for product development.
The US application, filed on August 20, 2026, claims priority from a Chilean application dated August 5. It is live and awaiting examination. The record is an early IP indicator, not proof of a new service launch or of the platform's performance. Its importance lies in how deliberately it defines the commercial object: not another plant-based food, but the software and workflow used to create, reformulate and scale many kinds of products.
The Operating Model
If the model proves useful across customer programs, food-development software could move from a specialist formulation tool to a shared workflow for R&D, procurement and operations. A chocolate team might compare cocoa-reduction routes before commissioning trials. An ice-cream team could screen plant-based structures against texture, calorie and cost targets. Procurement could see which substitutions create downstream process risk, while factory teams could capture what changed during scale-up.
That future is conditional. A platform that recommends attractive bench formulas but cannot absorb pilot-plant data, supplier variability or regulatory constraints will remain a concept engine. The stronger model would follow a formulation through handoffs: from brief and ingredient selection to sensory work, pilot production, claims review and manufacturing. Commercial value would come from fewer uninformative trials, clearer decisions and retained organizational learning—not from removing food scientists or chefs. It could also change consulting engagements, because the durable deliverable would be both a product and a structured record of why it works.
The Mechanism Behind It
The mechanism starts by translating a product brief into constraints the system can compare. These may include a nutritional target, excluded ingredients, cost ceiling, sensory profile, regulatory requirement, available process or supply limitation. Ingredient records then need more than names: they require composition, functionality, provenance and evidence about how each material behaves in particular matrices. Models can use those relationships to propose candidate formulations and predict which combinations are least likely to fail before a bench trial.
Physical testing remains essential. A food scientist makes prototypes, measures texture, stability or flavour, records what happened and feeds the result back into the project. The useful loop is therefore not “AI writes a recipe.” It is brief, prediction, experiment, measurement and revision. The operating advantage appears when the result from one project becomes structured learning for the next. A failed emulsifier combination, an unexpected aftertaste or a process-sensitive ingredient can become searchable evidence rather than a note buried in a spreadsheet.
The filing describes software that also connects product attributes with scientific, technical and consumer information. In practice, that could help teams test trade-offs rather than optimize one variable in isolation. Reducing cocoa, sugar or saturated fat may affect cost, viscosity, colour, flavour release and factory settings at the same time. A credible platform must expose those conflicts, not hide them behind a single recommendation. The hard work is data quality: normalizing old experiments, preserving supplier and batch context, distinguishing measured results from assumptions, and knowing when a model has moved beyond its reliable domain.
Why Operators Should Care
Food companies already use modelling tools, formulation databases and statistical experiment design. The more consequential shift is packaging those capabilities as a connected operating layer that other R&D teams can buy. NotCo's current site presents Giuseppe AI as a system spanning proprietary product-development data, formulation work, procurement, marketing and operations. That description aligns with a broader business change reported on September 9, 2026: Green Queen says the company has been divesting parts of its own-brand manufacturing footprint while moving toward AI-led product development for other manufacturers.
Named customer work makes the proposition more than a software category claim. Reuters reported that Barry Callebaut would use the platform to explore chocolate recipes, cocoa alternatives and process efficiency. A separate Reuters report said the Magnum Ice Cream Company planned to use it for calorie reduction, plant-based development and commodity-cost pressure. For consultants, the commercial question is whether food knowledge that normally disappears inside individual projects can become reusable infrastructure across briefs, categories and factories.
Proof Points to Watch
The next evidence should show how customer teams use the platform after the first demonstration. Watch for named products or reformulations, disclosed development timelines, the number and type of physical trials avoided, and whether predictions remain useful at pilot and factory scale. Independent case studies should separate software contribution from the work of experienced scientists and suppliers.
Also look for integrations with laboratory information systems, product-lifecycle tools and supplier specifications; methods for recording sensory and processing data; controls on confidential formulas; and clear provenance for every recommendation. Repeated contracts from companies in different categories would be stronger than another partnership announcement. For the recent corporate pivot, headcount and investment in the AI operation, renewal activity and evidence that divested food businesses still use the system would clarify whether the platform is becoming a durable business rather than a strategic narrative.
WBC Assessment
This is a strong food-innovation signal because the legal scope, operating platform and named industrial partnerships point in the same direction. The filing does not merely say “AI for food.” It identifies ingredient selection, interaction modelling, formulation optimization, product testing, scientific consulting and virtual assistance as parts of one commercial system. That specificity makes the strategic shift legible.
The proposition is plausible because large manufacturers already face reformulation pressure from commodity volatility, nutrition targets and changing regulation, while much of their experimental knowledge remains fragmented. NotCo also has an unusual source of training data: years of developing physical foods rather than observing product development from outside.
The uncertainty is execution. Predictions must survive sensory panels, processing equipment, shelf life, supplier changes and regulation. Customers also need evidence that proprietary formulas remain protected and that recommendations can be audited. WBC would treat the platform as an emerging R&D operating model with credible industrial demand, while withholding judgment on productivity and product quality until independent customer outcomes show what changed from brief to factory.