New artificial intelligence (AI) tools are changing the way restaurants analyze customer data, test food safety, and even recreate scents in the lab. The technology includes a natural language assistant for querying loyalty programs, an electronic tongue that achieved 95% accuracy in detecting food problems, and a system that can capture and reproduce specific scents in different locations. Included.
Restaurants use new AI tools to decipher customer data
Paytronix this week announced an AI assistant that helps restaurants and convenience stores understand loyalty program data through simple questions and answers.
New tools allow staff to query customer information in plain language, like asking about top-selling products or tracking redemptions, instead of digging through complex reports and analytics.
“PX Assistant brings an overview of important data and recommendations for improving your loyalty program to your program in seconds,” said Christine Cocce, director of marketing at Legal Seafoods, who tested the system in beta, in a news release. I am.
The AI assistant analyzes spending patterns and suggests marketing campaigns based on actual customer behavior. Track daily, weekly, and monthly loyalty spend across stores, allowing businesses to identify trends in how customers use their rewards.
By simplifying access to customer insights, this tool helps level the playing field for smaller operations that don’t require a dedicated analytics team. The Assistant provides relevant suggestions based on transaction data for each brand.
While AI tools are becoming more commonplace in restaurant tech stacks, Paytronix aims to make them especially accessible by focusing on natural language queries that don’t require technical expertise. Masu.
AI proves it has better taste than humans in food safety technology
In a surprising development that could revolutionize food safety testing, AI has proven it has better “taste” than its human creators. Researchers at Penn State University found that when they let an AI-powered electronic tongue decide on its own how to analyze food and beverages, it achieved an astonishing 95% accuracy. This is better than the 80% accuracy achieved using human-specified parameters.
The breakthrough device, detailed in Nature, combines a graphene-based sensor with a neural network to detect everything from watered down milk to spoiled juice. But the real story happened when the researchers let the AI define the evaluation criteria, rather than using human-specified parameters.
Using game theory, the team gained unprecedented insight into the AI’s decision-making process, revealing that the AI analyzes data holistically rather than checking individual parameters like humans do. I did. This more nuanced approach has the potential to transform food safety testing and medical diagnostics, providing a faster and more accurate alternative to traditional testing methods.
Scientists use AI to remotely capture and recreate scents
Osmo scientists have combined AI and molecular analysis techniques to develop a system that can analyze and reproduce scents in various locations.
The process uses gas chromatography-mass spectrometry (GCMS) to identify the molecular components of odors, which are then mapped and reconstructed using an AI-guided formulation robot. In initial tests, we were able to successfully recreate the scent of coconut throughout a laboratory space.
“Our ongoing data collection process will reduce the number of mysterious molecules and provide new ways to reproduce them,” Osmo founder and CEO Alex Wilczko said in a news release. This is an effort to find out.” The team has built the largest AI-compatible scent database to date.
This system faces significant technical hurdles, especially regarding subtle compounds that are difficult to detect but important to the overall aroma profile. Sulfur compounds found in tropical fruits have proven particularly difficult to capture and reproduce accurately.
This technology combines several AI approaches. The basic algorithm processes existing scent data, and the machine learning model predicts the formula from the GCMS readings. The “Key Smell Map” plots scents in multidimensional space to guide recreation. Although this process is largely automated, it still requires human oversight.
For molecular analysis, liquid samples are injected directly into the GCMS system, whereas for solid objects such as fruit, headspace analysis is required to capture airborne molecules around the sample. The data is uploaded to cloud storage, where an AI system interprets the molecular patterns and instructs a robotic system to mix recreationally suitable compounds.
The research team plans to begin limited public testing, but acknowledges that further development is needed to perfect the technology. This research could have implications for a variety of industries, from perfumery to food science, where accurate scent analysis and reproduction is a valuable tool.
