Let's make it happen
Ready to deploy? Review the sample code on GitHub for detailed deployment instructions to deploy as-is or customize to fit your needs.
Overview
This Guidance shows how to connect automated data sources to contact center systems. These sources could include Internet of Things (IoT) devices, interactive voice response (IVR) systems, customer relationship management (CRM) alerts, and automated quality monitoring tools. IoT integration enables three key benefits: real-time detection of errors and anomalies, automated incident resolution, and direct integration with omni-channel contact centers. Using artificial intelligence (AI), this workflow reduces average response time and helps to resolves common issues without human intervention.
How it works
Using Bedrock Agents
This AWS guidance demonstrates a comprehensive IoT-to-contact center integration solution that connects automated data sources like IoT devices, IVR systems, and CRM alerts to omni-channel contact centers using artificial intelligence.
Download the architecture diagram
Step 1
IoT devices publish to AWS IoT Core.
Step 2
Errors or alerts generated by devices are published to designated IoT topics, such as an "error topic" or "basic ingest topic". These inputs are then routed to an AWS Lambda function based on the defined IoT rule.
Step 3
Devices publish standard telemetry data to an IoT topic. This real-time data can be acted on, while a copy is also routed to a data store for the purposes of analytics and anomaly detection.
Step 4
The Lambda function transmits the error code, device ID, and any additional relevant information to Amazon Bedrock Agents.
Step 5
The data is streamed through Amazon Data Firehose and ingested into an Amazon Simple Storage Service (Amazon S3) data store.
Step 6
An Amazon EventBridge rule invokes a Lambda function at a schedule frequency specified by the user using a cron expression. If anomaly is detected in the stored data using Amazon SageMaker Canvas, it sends relevant information including device ID(s) to the Lambda function.
Step 7
Amazon Bedrock Agents reviews the received input and takes appropriate action by triggering the right action group.
Step 7a
The agent queries the QnABot on AWS knowledge base to retrieve any relevant supplementary information or policies, including remediation procedures. This is appended to the ticket if required in step 7b.
Step 7b
The agent creates a ticket within the contact center and enriches the message with any data from step 7a.
Step 8
If the automated remediation policy is enabled, the system will automatically remediate the IoT device by executing an AWS IoT Job.
Step 9
The site operator, owner, or customer is notified of the incident and any actions taken.
Step 10
The site operator engages with the QnABot on AWS solution along with the ability to request additional information or access historical data as needed.
Step 11
The company's manuals, standard operating procedures (SOPs), and engineering specifications are ingested into an Amazon S3 bucket, which the knowledge base then references.
Using Amazon Bedrock AgentCore
This advanced workflow leverages Amazon Bedrock AgentCore Runtime to orchestrate intelligent incident response when IoT devices publish telemetry and error messages to AWS IoT Core
Download the architecture diagram
Step 1
IoT devices publish telemetry and error messages to AWS IoT Core.
Step 2
Errors or alerts generated by devices are published to designated IoT topics, such as an "error topic" or a "basic ingest topic."
Step 3
These inputs are then routed to an AWS Lambda function based on the defined IoT rule.
Step 4
The Lambda function transmits the error code, device ID, and any additional relevant information to AI Agent.
Step 5
AI Agent queries Amazon Bedrock Knowledge Bases to retrieve any relevant supplementary information or policies, including remediation procedures. AI Agent reviews the received input and takes appropriate action by triggering the right tools. See detailed AI Agent architecture in slide 2.
Step 6
AI Agent creates a ticket in Amazon DynamoDB.
Step 7
AI Agent sends an email to the site contact operator using Amazon Simple Email Service (Amazon SES).
Step 8
If the remediation step includes calling the site contact operator, AI Agent initiates an outbound call using Amazon Connect.
Step 9
If the automated remediation policy is enabled, the system will automatically remediate the IoT device by executing an AWS IoT job.
Step 10
The user documents, such as company manuals, standard operating procedures (SOPs), and engineering specifications are ingested into an Amazon Simple Storage Service (Amazon S3) bucket. They are synced with Amazon Bedrock Knowledge Base.
Using Amazon Bedrock AgentCore (AI Agent Flow)
This architecture diagram illustrates how to effectively support connecting automated inputs to contact centers on AWS. It enables real-time error monitoring and resolution, and seamless contact center integration for IoT-connected systems, reducing downtime and optimizing issue resolution through AI-driven workflows. This slide shows Steps 1-5 for the AI Agent.
Download the architecture diagram
Step 1
AI Agent is built using Strands Agents pattern, which enables complex reasoning and decision-making workflows for automated customer service applications. The AI Agent is deployed on Amazon Bedrock AgentCore Runtime, which provides purpose-built infrastructure with the scalability, reliability, and security, critical to real-world deployment.
Step 2
AWS Lambda functions, such as log errors, send emails, initiate outbound calls, and clear faults, are exposed as MCP Servers in Amazon Bedrock AgentCore Gateway. This allowed AWS Lambda functions to be accessible to AI Agent as MCP-compatible tools.
Step 3
Amazon Cognito handles authentication and authorization for secure connection between Amazon Bedrock AgentCore Runtime and Amazon Bedrock AgentCore Gateway.
Step 4
AI Agent processes inputs using large language models (LLMs) hosted on Amazon Bedrock. The agent enhances its responses by retrieving relevant information from Amazon Bedrock Knowledge Bases through retrieval augmented generation, allowing it to incorporate domain-specific knowledge into its decision-making process.
Step 5
Amazon Bedrock AgentCore Observability tracks and monitors all AI Agent's activities.
Deploy with confidence
Everything you need to launch this Guidance in your account is right here.
Well-Architected Pillars
The architecture diagram above is an example of a Solution created with Well-Architected best practices in mind. To be fully Well-Architected, you should follow as many Well-Architected best practices as possible.
Operational Excellence
This architecture implements comprehensive operational monitoring and automation through multiple layers. Real-time IoT device monitoring is handled through AWS IoT Core, automated processing of alerts and anomalies uses Lambda functions, and intelligent ticket creation and routing is handled by Amazon Bedrock. This approach maintains automated feedback loops to detect, process, and resolve device issues with minimal human intervention, while Amazon CloudWatch provides detailed metrics and logging to enable proactive monitoring and quick troubleshooting. These services create a fully automated operational workflow that significantly reduces mean time to resolution for device issues, freeing up human operators to focus on more complex matters. The managed nature of these services also reduces operational overhead, allowing the team to concentrate on optimizing application logic and business workflows rather than managing underlying infrastructure.
Security
This architectural framework incorporates multiple layers of robust security controls. AWS IoT Core provides device authentication and encryption of data in transit. Each individual component operates with the principle of least-privilege AWS Identity and Access Management (IAM) roles, helping to ensure secure service-to-service communication. All data stored within Amazon S3 is encrypted at rest, and access to the knowledge base is strictly regulated. Furthermore, this approach implements secure communication channels between the contact center and automated systems, safeguarding sensitive customer and device data.
Reliability
AWS IoT Core provides reliable device communication, featuring automatic retry mechanisms and message persistence. And, this architectural framework incorporates multiple data processing paths, handling real-time alerts as well as batch telemetry processing. Data is durably stored in Amazon S3, and the knowledge base maintains redundant information to facilitate incident resolution. Furthermore, the automated workflow includes error handling and retry logic at each step, helping ensure this workflow can withstand service disruptions and maintain robust operation.
Performance Efficiency
By separating real-time processing from batch operations, resource utilization is optimized to support efficient workloads. In addition, real-time alerts are processed immediately through Lambda functions, while telemetry data is efficiently handled through Data Firehose. And through the intelligent routing of issues based on severity and type, critical problems receive immediate attention. Lastly, Amazon Bedrock provides rapid AI-driven decision making for ticket classification and routing.
Cost Optimization
This architecture optimizes costs through efficient resource utilization and automated scaling. The serverless components only incur charges during actual usage, while the pricing model of AWS IoT Core effectively handles large numbers of connected devices. Data storage costs are further optimized through the appropriate use of Amazon S3 storage tiers and lifecycle policies. Additionally, the automated incident resolution reduces operational expenses by minimizing the need for human intervention in addressing routine issues. This approach minimizes operational costs while maintaining high service levels. The pay-per-use model for serverless components helps ensure that costs scale directly with usage, and the automated workflow reduces human operational expenses. The efficient data handling and storage strategies also optimize the ongoing infrastructure costs.
Sustainability
The serverless components of this architecture automatically scale to match actual demand, eliminating idle resource waste. This framework also optimizes data storage through the efficient use of Amazon S3 storage tiers and lifecycle policies. Furthermore, the automated incident resolution minimizes the energy footprint by reducing the need for human intervention and associated workplace infrastructure. This design minimizes the environmental impact through efficient resource utilization and automated scaling. The serverless architecture helps ensure that computing resources are only used when required, while the automated workflow reduces the energy footprint of operational activities. Lastly, the efficient data handling and storage strategies also minimize the infrastructure necessary to maintain the architecture.