Revolutionize Your Customer Support with AI: Amazon Bedrock & Mistral Models

Transforming Customer Support with AI

AI agents are bridging the gap between large language models (LLMs) and real-world applications, revolutionizing customer service across industries. These intelligent systems enhance decision-making, automate workflows, and enable more natural human-AI interactions.

Key Areas of Impact

• Enhancing decision-making: Providing deeper, context-aware insights to improve customer support outcomes.
• Automating workflows: Streamlining customer service processes from initial contact to resolution.
• Human-AI interactions: Enabling more natural and intuitive interactions between customers and AI systems.
• Innovation and knowledge integration: Combining diverse data sources to address customer queries more effectively.
• Ethical AI practices: Ensuring transparent and explainable AI systems to build trust.

Building a Personalized Customer Support Experience

This post demonstrates how to use Amazon Bedrock and LangGraph to build a personalized customer support experience for an ecommerce retailer. By integrating the Mistral Large 2 and Pixtral Large models, we guide you through automating key customer support workflows such as:

• Ticket categorization
• Order details extraction
• Damage assessment
• Generating contextual responses

Solution Overview

The solution involves customers initiating support requests through email, which are automatically converted into support tickets in Atlassian Jira Service Management. The automation solution then:

1. Identifies the intent behind each query.
2. Categorizes the tickets.
3. Assigns them to a bot user for processing.

LangGraph orchestrates the workflow, extracting key identifiers like transaction IDs and order numbers from the support ticket. It analyzes the query and uses these identifiers to call relevant tools, generating a comprehensive and context-aware response.

Responsible AI

It's crucial to validate inputs and ensure LLM outputs are secure and responsible. Amazon Bedrock Guardrails provide configurable safeguards to monitor and filter both user inputs and AI-generated responses, ensuring interactions remain safe and aligned with organizational policies.

Observability

Observability is vital for customer support applications to provide real-time visibility into model performance, usage patterns, and operational health. This helps identify and mitigate risks like hallucinations, prompt injections, and toxic language.

Prerequisites

To implement this solution, you'll need:

• A Jira service management project with API token access.
• Custom fields for Category and Response.
• A bot user to assign tickets.
• An AWS account with access to Mistral Large 2 and Pixtral Large models.
• A LangGraph application running locally.

Implementation with LangGraph

The core of customer support automation involves specialized tools and functions designed to collect, analyze, and integrate data from service management systems and databases. Key components include:

• BedrockClient class: Wrapper for interacting with Amazon Bedrock services.
• Database class: Facilitates interactions with a SQLite database.
• JiraSM class: Interface for interacting with Jira Service Management.
• CustomerSupport class: Encapsulates customer support processing logic using LangGraph and Amazon Bedrock.
• Utility class: Provides essential functions for logging, file handling, and image processing.

Run the Agentic Workflow

The workflow can be executed for various ticket types, fetching corresponding Jira tickets and downloading attachments. The State object is initialized with ticket details and used throughout the workflow execution.

Example Workflow

The following code snippet invokes the workflow for a Jira ticket:

# Example code to invoke the workflow

Clean Up

After implementation, ensure to:

• Delete any IAM roles and policies created for this post.
• Remove access to Amazon Bedrock FM if no longer needed.
• Delete temporary files and guardrails used.

Explore Further

Check out the GitHub repo for step-by-step instructions and code examples.