The second part of our three-part series on the intersection of food and AI focuses on use cases in operations within the supply chain. If you missed it, check out Part I: Food Formulation and Part III: Consumer, along with our latest Future of Food Series: Food x AI video that ties the three parts together.
Bluestein’s Take
Breakdown Silos, Boost Efficiency: AI solutions streamline operations by breaking down silos, enhancing productivity, and increasing profitability. For example, solutions for inventory management also streamline ordering and reduce waste.
Regulatory Catalyst: The FDA's traceability regulations, effective by January 2026, are a catalyst driving the adoption of AI solutions for traceability and provenance within the supply chain. This could power competition among the current startup and incumbents to offer superior solutions.
Market Need Trumps Technological Superiority: Startups must address urgent market needs, such as reducing food waste and managing ingredient disruptions, to succeed. In this landscape, responsiveness to market demands often outweighs technological superiority.
Seamless Onboarding and Integration: For widespread adoption, AI driven solutions need to integrate with existing technologies and workflows, and minimize the capital investment required to start seeing results.
Full Autonomy Through AI Agents: Many startups are working on fully autonomous decision-making systems, but no one has proven this on a scale beyond demos. Much like self-driving vehicles, ‘edge cases’ and messy data make the challenge much harder than it first appears. Low-hanging fruit in the supply chain, like inventory planning, will likely be fully automated first before more complicated areas like global supply chain and logistics.
From Code to Cart: AI in Food Operations
Operations within the food industry encompass the vast supportive infrastructure that enables companies to meet consumer needs, and solutions cover a wide array of activities from manufacturing through retail, including production, processing, distribution, transportation, and waste management. AI has the potential to transform operations to enable a more comprehensive understanding and greater control of the supply chain. The market opportunity within operations is ripe due to a combination of supply chain pressures, regulation, and technology advancements.
With respect to supply chain pressures, companies have been hit with increasing costs from all sides, be it labor, transportation, and food waste. The food manufacturing industry has seen rising labor costs, with ArrowStream citing an average 3% increase in wages annually over the last five years;1 between 2020 and 2022, the average hourly rate for U.S. restaurant workers increased by 20%, rising from $16.65 to $18.71.2 This is further accelerated by increases in mandated minimum wages across states, with 21 states raising their minimum in 2020.3 Food waste also continues to be a source of lost revenue, with an estimated $482 billion lost at grocery retail according to a Coresight study in 2021; to better tackle food waste, 84% of grocery retail respondents in their survey cited increased investment in technology.4 In addition, the COVID-19 pandemic laid bare the fragility of our supply chain, as disruptions to freight movement caused industry-wide challenges across delays, reduced capacity, and higher costs. For example, during the pandemic, the average delivery time for freight shipments increased by 30%, with many shipments experiencing delays of up to 10 days; the trucking industry faced a reported 20% reduction in capacity due to labor shortages and lockdowns; and there was a 15% increase in average travel time for long-haul freight. All of which has stressed profit margins, driving the need for automation and resiliency.
Regulatory pressure is further driving AI adoption in the food industry. The Food Safety Modernization Act (FSMA) is one of the most comprehensive reforms of federal food safety laws in decades and has a compliance deadline approaching in January 2026. The FSMA requires that food companies adhere to stricter safety standards, enhanced traceability, and increased testing and documentation. Internal corporate sustainability goals have also created urgency. In a Coresight survey, 63% of grocery retailers view food waste reduction as very important in meeting their targets.5 Companies therefore need to quickly find solutions to meet these new standards.
As discussed in Part I, AI technology has been accelerating rapidly, and we’re only at the earliest innings. ModorIntelligence puts the market size of AI in the food industry at $9.7 billion globally in 2024, and projects it will increase at a 38% CAGR to $49 billion by 2029.6 AI is enabling technological leverage across the supply chain, providing a significant opportunity to achieve enhanced efficiency, reduced waste, and better responsiveness to market demands – ultimately enhancing profitability and productivity.
Byte-Sized Opportunities Companies looking for operational improvement and increased efficiency can use AI throughout the supply chain, from superior forecasting to more efficient traceability and production. Companies have existing datasets from operations that are high quality, readily available, and normalized, thereby reducing hurdles to rapid algorithm development and adoption.
Better Forecasting: AI models can provide better forecasting across the supply chain. This includes predicting ingredient supply/demand and pricing fluctuations, as well as anticipating delays in delivery due to adverse weather patterns or other forces. Such predictive analytics enable companies to make more informed decisions and maintain a smooth flow of goods through the supply chain. Recently Danone and Microsoft announced a partnership to build an AI-enabled supply chain to manage production and logistics through predictive forecasting, real-time adjustments, and streamlined operations.7
Enhanced Efficiency in Production: AI can dramatically improve production efficiency, shifting companies from reactive to preventive production strategies to streamline and optimize manufacturing, as well as reduce downtime and errors. For example, AI solutions can enable companies to monitor machine performance instantaneously and reduce costly interruptions; for instance, predictive maintenance systems could alert operators before machinery fails, allowing for timely interventions that avoid unexpected breakdowns and extend the lifespan of critical equipment. Nestle has implemented an AI-powered predictive maintenance system using machine learning and sensors to proactively detect potential equipment failures – thereby increasing productivity.8
Improved Traceability and Provenance: An infrastructure of sensors and analytics can provide unprecedented level of detail in tracking products from source to destination. This facilitates better transparency and traceability, in addition to improved food safety and quality, reduced food waste, adherence to regulatory requirements, and increased consumer trust. One example of this is Walmart’s partnership with IBM Food Trust’s solution built on the IBM Blockchain Platform to support food safety, traceability, and transparency.9 The FSMA regulation discussed above is likely to act as a further catalyst for the adoption of AI solutions for traceability and establishing provenance.
Labor Automation: The promise of fully autonomous ‘agentic’ AI systems can help businesses control increasing labor costs and regulations by requiring fewer employees in the loop. While most AI systems are used as tools now, they are likely to improve over time, thereby eliminating the need for a human ‘in the loop’. Some early attempts at AI agents use integration services like Zapier to program ‘trigger’ events that execute a certain task. For example, when a purchase order email is received, the PDF could be uploaded to cloud services like Dropbox automatically. More advanced AI agents have the agent interact with the actual user interface of a website or app as a human would, without requiring API keys to integrate. True AI agents likely will not be deployed until they can execute multi-step tasks and problem solve to handle unusual situations. As people and companies become more familiar with using technologies like generative AI, there will be more acceptance and trust in AI systems operating and making decisions independent of human oversight.
Integration Advantages for Deployment: AI within operations also holds the promise of easier integration. The data requirements for implementing AI in operations are relatively easier to design versus other segments like formulation and consumer behavior because operational data such as inventory levels and power usage by machinery can be easily measured and captured. This makes the integration of AI systems more straightforward and less resource intensive.
Key Challenges
While there are promising opportunities for AI within operations, several challenges must be addressed for AI to realize its full potential.
Building Software Systems with Necessary Interfaces and Controls: While gathering data from operational processes is often straightforward, the larger challenge is in seamless data integration to drive value for customers. This integration involves creating different interfaces for various team members along the supply chain - some might need technical tools, while others require user-friendly dashboards or apps.
There's also complexity in incorporating real-time sensor data, especially in manufacturing settings. The task involves processing large volumes of data at high speeds, dealing with potential sensor inaccuracies or failures, and making split-second decisions based on this information. The system needs to filter out noise, detect anomalies, and respond to changes almost instantly, all while avoiding false alarms that could disrupt production. Getting a detailed view on manufacturing and supply chain processes is helpful, but companies need to set up effective alert systems and troubleshooting processes to address problems as they occur.
Evidence-based Applications for Traceability: Supply chain digitization efforts are still early. Many companies tout bold claims, but the feasibility of the solutions remains questionable given the lack of data from real world use cases. AI systems often fail in the ‘edge cases’, and real-world data is messy. The data is noisy because supply chain disruptions can be due to many factors like weather, geopolitics, and/or infrastructure failures. While the system may outperform humans in certain use cases, evidence for fully autonomous capabilities in decision-making remains low. An AI system that only sometimes works may not justify implementation.
Data Quantity and Quality for Reliable Predictions: The food value chain suffers from inconsistent and incomplete data collection, creating hurdles for digitization efforts, and thus the ability for AI to truly reach its potential. For example, purchase orders for distributors often arrive and are sent via email through a largely manual process; large distributers use Electronic Data Interchange (EDI) systems that automate some aspects of purchasing, but adoption remains lower for medium and small businesses. Even when high quality data exists, like automatic temperature readings of food during manufacturing, there is a lack of synthesis with other data sources to generate valuable real-time insights.
Technical Challenges for Full Autonomy: Fully autonomous systems need to handle real world tasks requiring multiple steps and technical proficiency. For example, an autonomous inventory tracking system would need to discern the degree of ripeness and spoilage of perishable food products before determining what ingredients need to be discarded and ordered. Some restaurants change their menus often, and an AI system could fail if not trained on how to identify and process a new ingredient. Autonomous food handling and production itself requires dexterity and multiple sensory inputs that are still in early development and will likely be prohibitively expensive for foodservice adoption upon release.
Market Segments
To better understand the AI landscape, the chart below maps out key players within operations across the supply chain, from ingredient sourcing through logistics.
Ingredient Sourcing: AI can enhance agility in sourcing raw materials, helping companies innovate faster and respond to disruptions. Companies like TraceGains are creating an ingredient procurement marketplace to help companies diversify their suppliers to mitigate ingredient and supplier risks.
Manufacturing: AI-driven controls and automation can provide value across production pipeline and in food safety protocols. Relevant solutions in production include machine controls for optimizing power consumption or reducing downtime, and/or alerting the need for machine maintenance (Montblanc AI). Solutions in food safety range from management of HACCP protocols (FoodReady) to the use of sensors and analytics (Robovison, SmartNanotubes) to monitor for food safety and defects.
Distribution: These AI solutions optimize all critical aspects of transporting a product from its manufacturing origin to its final point of sale. This involves technologies like real-time inventory tracking in warehouses and retail outlets (Aeriu, Pensa, GatherAI, Dronescan), alert systems for managing risks such as supply disruptions (Helios), food spoilage (Strella), and anticipating spikes in consumer traffic. Some solutions connect all enterprise processes related to distribution and analyze the impact on profitability (Silo).
Retail/Foodservice: Retailers and restaurants face high labor costs, inflation, and changing consumer behavior, and AI has the potential to help companies become more data driven in their decision making to increase efficiency and profit, as well as reduce food waste. Solutions in retail and foodservice include demand forecasting tools (Guac, Metafoodx, Shelf Engine), ordering tools (foodbit), pricing optimization (Luca), and restaurant operations management (Clear COGS, 5-out, PreciTaste).
Traceability: Traceable supply chains increase food safety, reduce spoilage, and decrease the chance of adulteration. The data generated also helps to understand and quantify the environmental impact of transportation to enhance sustainability. Pandemic related supply chain disruptions highlighted the fragility of our current supply chain, and companies therefore increasingly sought out solutions to provide transparency and visibility as products move through the supply chain (Dori AI, Roambee, Aanika Bio) using an infrastructure of sensors (Moeco, Bluicity) to closely track timing, location, and temperature – and adulteration (Scentian Bio).
Logistics: The logistics industry faces challenges related to labor shortages, rising fuel costs, and supply chain disruptions. AI-driven logistics solutions aim to optimize the actual movement of products (FourKites, Route Freight), including improving brand to shipper matching, tracking, routing, warehouse automation (Hopstack.io), and last mile delivery (Shipsi, Vecino). Solutions in logistics need superior technology and the ability to overcome the hurdle of broker networks and relationships to gain market share over incumbents.
Evaluative Criteria
AI has incredible potential to optimize operations. The best solutions in the space demonstrate the following:
Solve A Critical Need: The overarching factor driving the adoption and success is the ability to address urgent systemic needs, rather than merely having superior technology. For example, the supply chain for ingredients has faced significant disruptions due to events like COVID-19 and ongoing political and climate stress.10 AI-driven ingredient sourcing solutions can offer vital information on alternative ingredients or more cost-effective options, helping businesses navigate these crises. Food waste is also a large and growing issue, and solutions that address waste need to enhance profitability and productivity with a high return on investment to be compelling.
Integrated System: The long-term success of a solution hinges on its ability to seamlessly integrate into the necessary operational areas to reduce onboarding time and increase efficiency. The food industry is complex with multiple stakeholders across the supply chain, and point solutions are not sufficient to drive adoption. For example, traceability solutions need buy-in & integration from many players across the value chain. For instance, an AI-based ERP system would effectively integrate operations related to inventory, logistics, and accounting, offering a holistic approach that point solutions might struggle to match. Startups offering specific solutions need to keep a broader vision in mind to be able to offer a comprehensive platform that can avoid being disintermediated by larger players.
Data Infrastructure Solutions: Current data generation methods like barcodes are pretty basic. To fully realize the value of AI, startups in the space need to think about creatively leveraging existing infrastructure, like cellular networks for smart labeling, or creating new infrastructure to generate more nuanced data. By focusing on ‘data infrastructure', companies can fully take advantage of the capabilities of their algorithms, own a unique and valuable dataset, and create a differentiation among other players in the space. Data is the biggest opportunity for a moat within the AI startup landscape
What’s Next
AI in operations is an incredibly dynamic space and we’re excited by real-time supply chain tracking, optimization, and prediction, as that’s where we see the largest white space. If you’re an entrepreneur in building in operations – or know of one – we’d love to hear from you.
Check out Part I: Food Formulation and Part III: Consumer, along with our latest Future of Food Series: Food x AI video that ties the three parts together.
A special thank you to our Analyst, Aaditi Tamhankar, and our PhD fellow, Tarini Naravane, for leading the effort on this series!