
Win wire

MICROSOFT CONFIDENTIAL
SOLUTION

Azure Data & AI and Apps
INDUSTRY

Discrete Manufacturing
SEGMENT

SMC West
At a Glance
CHALLENGE
Employees unable to find information about company policies and processes
Reliance on HR and IT
SOLUTION
AI Design win for IT/HR policy chatbot
IMPACT
Employees can find answers quickly and efficiently
Less strain on HR and IT resources
Potential to expand chatbot to customers and partners
Their challenge
Campbell Scientific, Inc. (CSI) designs and manufactures instruments that help countries better prepare for and mitigate extreme climate events. In 2023, MCA Connect began working with Campbell to globally deploy Dynamics 365 Finance and Supply Chain Management. As the manufacturer looked to continue their transformation with generative AI, they turned to MCA Connect for help bringing their first use case to life.
Like many companies, Campbell struggles to make up-to-date policy documents easily accessible to employees. When employees have questions about HR, cybersecurity, or ISO, finding the right documentation can be tedious and time-consuming.
As a result, HR and IT resources spend a lot of time answering questions and sharing information. Whenever a policy changes, they must also complete the time-consuming task of updating documentation and communicating changes.
To solve this problem, Campbell developed an AI chatbot that ensures employees can quickly find the up-to-date information they need. However, Campbell needed a partner to help improve and refine the chatbot design and ensure it follows best practices.
After reviewing several potential partners, Campbell chose MCA Connect due to our deep bench of data scientists and Microsoft AI expertise. Our team provided thought leadership during the sales process on model evaluation for LLMs and different chunking methods within Azure AI Search. We also brought expertise that gave the Campbell team confidence the chatbot solution would be “enterprise-grade.”
Our solution
MCA Connect and Campbell Scientific worked together to develop a IT/HR policy chatbot. Used internally, this chatbot helps employees find answers to specific policy questions on topics like HR, cybersecurity, and ISO. Policy updates are automatically reflected in the chatbot’s database, ensuring employees always have the most up-to-date information.
The policy chatbot utilizes Azure OpenAI for AI and Azure App Services for the front-end interface. Azure Cosmos DB is used to store chat history for accuracy tracking and refinement. The Azure Well-Architected Framework and Azure security baseline for Azure OpenAI were inputs to ensuring the security of the data and APIs.
MCA Connect provided technical services, including technical architecture, education, security hardening, and assistance through deployment to production.
Expected results
As the first of many future AI capabilities, Campbell Scientific’s IT/HR chatbot will serve as a virtual assistant. This chatbot will provide all employees with timely and accurate information on company policies, procedures, and relevant regulations. By reducing reliance on HR and IT resources, the chatbot will help Campbell employees work more efficiently.
Azure impact (USD/month)
- $3,000 AI
- $3,000 Cosmos DB
- $3,000 Apps
Campbell Scientific hopes to extend the chatbot functionality to support other key business roles such as R&D, engineering, and customer service.
MCA Connect will also guide Campbell Scientific and other manufacturing customers to take advantage of similar high-value “chat with your data” use cases. These use cases may include:
- Service: Chat that uses product and equipment manuals to accelerate troubleshooting
- Contracts: Chat that analyzes customer entitlements and SLAs to identify legal risk and optimize accountability
- Products & Process Design: Chat that uses product design documents to accelerate manufacturing design
- Patents: Chat that summarizes existing patents and advises on defensive patents to pursue
MICROSOFT RECOGNITION
Thanks to Tim Walton, Tim Larson, Tarik Lakabi, Justin Rister, Joe Sornsin, Ahalia Ramlogan, and Clint Will for your help and support with this AI design project.