This is the first in a new series of posts about #StartupsOnAzure that will look at different companies within Microsoft for Startups Founders Hub and how they are using their credits to access a wide array of Azure services to help level up their startup.
Overview
With the ongoing reality of erratic weather patterns, the agricultural ecosystem and its entire supply chain have become unpredictable. Groundbreaking data-driven AI/ML can mitigate that unpredictability with greater accuracy and consistency, and decision support to the wine, food, and beverage supply chain.
Legacy systems hamper agricultural supply chain predictability
While data and AI are the key to a more resilient agri-food system, legacy data systems and database silos make data widely inaccessible. This makes it nearly impossible to use AI/ML predictive models to:
- Retrieve actionable data about weather effects on agriculture quality, yields, and harvest
- Model “what if… ?” scenario analyses for decision support
- Automate forecasts for higher yield and sustainable quality production using regenerative agriculture methods
Agricultural chain intelligence platform leader Trellis needed to gather data from legacy sources, requiring a secure cloud-based architecture to feed their proprietary engines and novel SaaS tools to serve clients around the world.
Trellis, a member of Microsoft for Startups Founders Hub, is at the cutting edge of providing much-needed predictive approaches to the agri-food supply chain. The challenge they faced, however, was building the required cloud architecture and data pipelines, which are crucial to gathering data from countless legacy platforms and silos. Accomplishing this goal requires a labor- and time-intensive deployment of a full-scale, secure, and private ML pipeline and infrastructure. But having this workflow in place could then drive their real-time predictive insights, powered by AI/ML, on top of each customer’s legacy enterprise and public data systems.
About Trellis
As an agricultural supply chain intelligence platform leader, Trellis takes a novel, data-driven approach to climate security to solve challenging issues along the food/consumer packaged goods value chain.
Trellis uses their proprietary AI/ML-driven engines and SaaS tooling to bring accurate, consistent predictions to the erratic agri-food supply chain to:
- Mitigate climate risk for global agricultural supply chain producers
- Predict and avoid supply chain risk
- Anticipate market demand shifts that impact food and beverage supply chain fulfillment
- Increase resource efficiency and scalability for food and beverage supply chain producers
- Help clients boost supply chain and food production by an average of 20% while increasing sustainability
About Azure Logic Apps and Azure ML
In a digital world, building data-gathering and ingress workflows along with the ML pipelines that deliver predictive intelligence is a challenging task for any business. Azure Logic Apps is a cloud-based platform where you can create and run automated workflows that integrate your apps, data, services, and systems. Microsoft’s solution enables the secure and private access and running of operations on various data sources via managed connectors in workflows.
Azure Machine Learning runs in the cloud to accelerate and manage your ML project lifecycle. Teams can then leverage MLOps to create ML models for data analysis that lead to accurate predictions to drive specific business outcomes. These solutions reduce the labor-intensive engineering needed for fast and actionable predictions in today’s food and beverage supply chain.
How Trellis Leverages Azure Logic Apps and ML to Support Legacy System Data Ingress/Analysis
Azure Logic Apps was the ideal solution to enable Trellis to securely connect to each customer’s legacy data systems such as ERP, supply chain management, WMS, etc. Logic Apps performs the heavy lifting of gathering all relevant data across all platforms via automated workflows and connector management. Trellis then applies different plugins to ingest and enrich the data via Logic Apps’ managed connectors workflow for process support, including:
- Normalization
- Outlier detection
- Error correction and data enrichment, including customer-specific business logic
“Azure Logic Apps and its connectors saved a massive amount of time it would take us to build and maintain connectors to legacy systems, while Azure Machine Learning provided the DevOps infrastructure. This enabled us to save engineering time and effort that we could devote to focusing on our core product offering — optimizing the global manufacturing of food & beverage to deliver incremental value to our business users,” said Trellis VP R&D Efrat Bar-Giora.
Trellis receives various datasets, such as field measurements, crop/weather pattern observations, factory/warehouse deliveries, production plans, and financial data from across the global agricultural ecosystem. This data triggers the proprietary Trellis AI/ML engines and system to create new predictions and insights, including:
- Outlier alerts
- Missing data
- Data imputation and inference based on machine learning and statistical modeling.
Logic Apps provides real-time monitoring of data ingress to deliver accurate alerts to the Trellis team via email. These inform the team if the system did not receive data or when processing errors occur requiring prompt correction. At the end of the process, the stored data is visualized in a proprietary knowledge graph that feeds the proprietary Trellis ML/AI engines.
Trellis can then ingest the data into their databases, allowing the team to run multiple transformations and ML solution models to create custom predictions and insights delivered to each customer.
Trellis uses Azure Cloud Services to create its cloud architecture environment comprising:
- VM instances
- Open-source PostgreSQL databases, as the primary data store for migrated client data
- An MLOps pipeline using Azure Machine Learning to manage their proprietary AI/ML engines for the creation of multiple predictive models to improve customers’ food and beverage supply chains
Conclusion
There are many reasons for a startup working in the supply chain ecosystem to use Azure Logic Apps and Azure Machine Learning. First, Azure Logic Apps can help manage the workflow between different systems. This is important in a supply chain where different parts of the process need to communicate with each other. Azure Logic Apps can also help automate tasks, such as sending notifications or reminders. This can save time and improve accuracy. Second, Azure Machine Learning can help with data analysis. This is particularly important in the agricultural ecosystem, where data is collected from a variety of sources. Azure Machine Learning can help make sense of this data and identify trends. This can help improve decision-making and help the startup to be more efficient.
To access the complete range of Azure products with up to $150,000 in credits, sign up today to Microsoft for Startups Founders Hub.
