aioneers Builds AI-Driven Supply Chain Intelligence Platform with Plotly Dash and Azure Databricks

Introduction

aioneers provides an AI-driven supply chain intelligence platform, AIO, that helps its clients achieve supply chain intelligence to make smarter, faster, and data-driven decisions. Its clients include billion-dollar enterprises, spanning machinery and equipment, consumer goods, and automotive. The development and data science team at aioneers handles the productization of supply chain solutions like inventory management, Sales & Operations Planning, and material availability along with new feature development like the supply chain copilot.

Challenge

The company initially faced challenges managing customer data through multiple third-party tools. While the AIO platform displays insights from customer data, the company does not interact with customer backend directly. Instead, data connectors import data into instances of AIO installed on customer-dedicated infrastructure, an architecture critical for customer security and compliance requirements.

Security is administered using Azure cloud services. As a Microsoft reseller, aioneers offers flexible options: customers can opt for a new Azure subscription set up by aioneers, or they can establish their own and provide aioneers with the principal access. Prior to establishing analytics, the aioneers team engages in thorough discussions about data scope and security to reassure customers that their data remains under their control and is fully secure.

With the evolution of AI technology, aioneers' product strategy evolved towards intelligent decision-making supported by smart dashboarding tools. aioneers chose Plotly and Dash after scoping the market for the best UI and graphing interfaces.

They have since started developing their own Dash-based components and using them in their own application to provide to their customers.

Solution

aioneers embeds a Dash application into their software solution. This application offers dynamic operations via filters using customized Dash components based on client needs.

The solution enhances existing Dash components, integrating them into customized interactive workflows. aioneers sets up these web apps for clients, which offer editing and workflow capabilities beyond those achievable in PowerBI and Tableau, and integrate into the full-stack application. aioneers' product implementation experts then benefit from a custom-built drag-and-drop interface for ease of building interactive analytics solutions for clients. 

In AIO, data is stored using a Hive Metastore and queried using SQL through the Databricks Python SQL connector. aioneers is optimizing its system by switching to a serverless SQL warehouse to serve heavy data more cost-effectively. It also leverages Databricks Unity Catalog for improved data management.

Results

By using Dash with Databricks, aioneers has achieved significant improvements in its analytics capabilities, providing enhanced business value to its clients. 

• The platform allows for on-brand analytics solutions that seamlessly integrate with existing systems used for supply chain management at industry-leading companies.
• Dash fills gaps left by traditional BI tools, offering greater flexibility and customization options without requiring a full-stack approach. 
• The use of Dash AG Grid for writeback capabilities enables users to edit the numbers and send back those to customer ERP systems, offering a more interactive and integrated experience. 
• Users can navigate from a single cell in a table to any other part of the application within the platform, enhancing workflow efficiency. 
• Both clients and aioneers personnel appreciate the all-in-one — but flexible — analytics platform, leading to increased satisfaction and adoption rates. 

Future enhancements in the AIO platform include a supply chain copilot — an AI persona that works autonomously and learns from the outcomes of all actions. It allows users to request specific charts, such as inventory charts, and have them generated based on available data. This interaction is facilitated without passing any data to the language model, ensuring data security.