ARCHITECTURE CONFIGURING SERVICE USING THE CLOUD FORMATION TEMPLATE The AWS CloudFormation template for Pindrop service simplifies provisioning and management on AWS. The Pindrop template can be used to easily configure Pindrop Service within Amazon Connect. Two values are required to configure Pindrop service using the CloudFormation template: AWS Connect Pindrop ® Integration Guide INTEGRATION GUIDE AMAZON CONNECT Users should be familiar with Amazon Connect and Amazon Web Services (AWS) prior to leveraging the Pindrop service. Please refer to the Amazon Connect User Guide and the AWS Lambda Developer Guide for detailed information on both services. PINDROP Pindrop has created an integration with AWS Connect that enables customers to detect risky behavior on every incoming call. Before integrating, please contact Pindrop for pricing information and access to the service via an API key. Figure 1. Dataflow diagram of Pindrop architecture within Amazon Connect
5
Embed
AWS Connect Pindrop Integration Guide · service and drag and drop the “Invoke AWS Lambda Function” box into the contact flow (Figure 2) and connect it to the Entry point (Figure
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ARCHITECTURE
CONFIGURING SERVICE USING THE CLOUD FORMATION TEMPLATEThe AWS CloudFormation template for Pindrop service simplifies provisioning and management on AWS. The Pindrop template can be used to easily configure Pindrop Service within Amazon Connect.
Two values are required to configure Pindrop service using the CloudFormation template:
AWS Connect Pindrop® Integration Guide
INTEGRATION GUIDE
AMAZON CONNECT Users should be familiar with Amazon Connect and Amazon Web Services (AWS) prior to leveraging the Pindrop service. Please refer to the Amazon Connect User Guide and the AWS Lambda Developer Guide for detailed information on both services.
PINDROPPindrop has created an integration with AWS Connect that enables customers to detect risky behavior on every incoming call. Before integrating, please contact Pindrop for pricing information and access to the service via an API key.
Figure 1. Dataflow diagram of Pindrop architecture within Amazon Connect
1. An API Key to include in the API calls to Pindrop
2. The Pindrop API endpoint
Both of these values can be furnished to customers after completing the new customer registration process with Pindrop at https://www.pindrop.com/pindrop-amazon-connect/.
CONFIGURING SERVICE MANUALLYSTEP 1 - NOTIFY PINDROP OF CALLThe Pindrop service requires a notification as soon as the call starts to initiate the fraud analysis process. A notification to Pindrop can be made by invoking an AWS Lambda function via the Contact Flow Editor.
Prior to modifying the Contact Flow - create a new AWS Lambda function with a Node.js runtime. The CreateCall.js Lambda function can be found in Appendix A.
Within the Lambda UI, the Lambda function role defines the permissions for the CreateCall.js Lambda function. The role should be set according to your own security policies - however, note that the script requires outbound HTTP access.
Once the function is created, return to the AWS Connect service and drag and drop the “Invoke AWS Lambda Function” box into the contact flow (Figure 2) and connect it to the Entry point (Figure 3)
Figure 2. Invoke AWS Lambda function within the AWS Connect Contact Flow Editor
Figure 3. Invoke AWS Lambda function connected to the call entry point
Edit the new function with the information found in Figure 4 below. To invoke the proper Lambda function, enter the Amazon Resource Name of the Create Call Lambda function under the Function ARN section.
The Call Start function requires two input parameters, an API key (ApiToken) and the destination URL (BaseUrl) for the API requests. Pindrop will provide both of these inputs at the time of integration.
STEP 2 - RETRIEVE RISK SCOREDuring a caller’s interaction with the contact center, Pindrop can assess the risk of the caller via an API request to the Pindrop cloud. The API call to identify risky behavior is typically executed at key transitions, for example, after answering a KBA, or before transferring to speak to an agent.
API requests should be made further along in the IVR workflow - allowing Pindrop to analyze as much of the call as possible to provide better risk detection results. During the integration process, Pindrop Solutions Architects can work with your staff to identify the best points within the call flow to initiate the API requests.
API requests to the Pindrop cloud are made directly from Lambda function calls within the Contact Flow Editor.
The FetchRisk.js Lambda function can be found in Appendix B. Similar to the Create Call function in Step 1 - users are required to create the Lambda function, and keep note of the Function ARN to reference within the Contact Flow Editor. The same fields in Figure 4 will be required when invoking the Fetch Risk Lambda function.
Connect must be granted the permission to invoke each lambda function.
One must use the Amazon CLI to grant this permission. For more information, see http://docs.aws.amazon.com/connect/latest/adminguide/connect-lambda-functions.html
STEP 3 - OPERATIONALIZING THE RISK SCOREOnce the Pindrop service returns the risk score for a caller, the contact flow can be modified to handle the caller based on the risk score, including showing the risk score to an agent via the Amazon Connect Contact Control Panel (CCP).
To action off of the risk results, the output from the Lambda function needs to be set as a usable variable within the Contact Flow Editor. Drag and drop the “Set contact attributes” box into the contact flow and connect it to the Fetch Risk Lambda function (Figure 5).
Figure 5. Processing the risk score
By default, the Lambda function returns the risk score of a caller into a RiskScore attribute and a flag to indicate a high risk caller (IsHighRisk). Modify the new “Set contact attributes” to store the results of the RiskScore into a new Destination Key. In our example (Figures 6 & 7), we will store the risk score into a RiskScore variable and the high risk flag into the IsHighRisk variable.
Figure 6. Setting the RiskScore attribute
Drag and drop a second “Set contact attributes” box into the contact flow and connect it to the previous “Set contact attributes” (Figure 5) to take action on if a caller is flagged as high risk. Similar to setting the RiskScore attribute, repeat
To specifically condition off of the results of the IsHighRisk attribute, drag and drop a “Check contact attributes” function into the Contact Flow Editor (Figure 5).
Modify the “Check contact attributes” function to check the condition of the IsHighRisk attribute (Figure 8).
Note: A high risk caller, by default, is any caller with a risk
score greater than 0.50. To configure the risk weights, please
reach out to Pindrop to access the “API Reference Guide”
whitepaper. The RiskScore attribute may also be conditioned
off of rather than the isHighRisk attribute.
Figure 8. Checking if a caller was flagged as High Risk
The output of the “Check contact attributes” function can then be leveraged to route the caller appropriately. In Figure 5 above, our example plays a prompt to the caller if the IsHighRisk result equaled true, and transfers the caller to a queue if the IsHighRisk flag was false. As mentioned above, the IsHighRisk flag as well as the raw RiskScore can be shown to an agent via the Amazon Connect CCP.