Case study
Using the Bedrock Agent to respond to customer queries
Summary
The CACI team has developed a method using Amazon Bedrock Agent to respond to customer queries.
Industry
Technology
Products used
Challenge
CACI has multiple software products and most of these products have their help pages. These help pages are hosted in different environments using different formats. When customers have any questions related to the products to related to how to do things, they go to these sites/documents, browse through the pages and finally find the answer. This process is not user friendly and often wastes a lot of valuable time. Some products have help pages saved in different files, which makes finding answers more difficult.
The challenge was to find a way to easily find answers to questions from the help manual(s).
Solution
CACI resolved this challenge using Amazon bedrock agent and demonstrated this in Pin Routes. Users can click on the icon and ask questions. The agent returns satisfactory answers. If the agent cannot find answers to the questions in the manual, it returns a message saying it could not find the answer to the questions.
Amazon bedrock agent has some advantages over other technologies that were investigated:
- Native AWS services. Low setup effort and maintained by AWS.
- Built in Guardrails, blocking harmful queries and inappropriate responses
- Well documented API and SDK support allowing for relatively easy integration
- Functionality can be extended as required
AI backend can be standardised across multiple products and services, the integration will vary.
Results
Amazon bedrock agent has revolutionised how customers access the help or retrieve answers to their questions.
This granular view is a game-changer for any organisation or product wanting to improve user experience and increase the usage of the products.
