As a contact heart agent, would you somewhat deal with having productive buyer conversations or get distracted by having to lookup buyer info and information articles that might exist in varied methods? We’ve all been there. Having a productive dialog whereas multitasking is difficult. A single detrimental expertise might put a dent on a buyer’s notion of your model.
The Live Call Analytics with Agent Assist (LCA) open-source resolution addresses these challenges by offering options comparable to AI-powered agent help, name transcription, name summarization, and rather more. As a part of our effort to satisfy the wants of your brokers, we try so as to add options based mostly in your suggestions and our personal expertise serving to contact heart operators.
One of many options we added is the power to write down your individual AWS Lambda hooks for the start of call and post-call to customized course of calls as they happen. This makes it simpler to customized combine with LCA structure with out advanced modification to the unique supply code. It additionally permits you to replace LCA stack deployments extra simply and rapidly than if you happen to had been modifying the code immediately.
As we speak, we’re excited to announce a function that permits you to combine LCA along with your Buyer Relationship Administration (CRM) system, constructed on prime of the pre- and post-call Lambda hooks.
On this submit, we stroll you thru organising the LCA/CRM integration with Salesforce.
LCA now has two extra Lambda hooks:
- Begin of name Lambda hook – The LCA Name Occasion/Transcript Processor invokes this hook initially of every name. This operate can implement customized logic that applies to the start of name processing, comparable to retrieving name abstract particulars logged right into a case in a CRM.
- Put up-call abstract Lambda hook – The LCA Name Occasion/Transcript Processor invokes this hook after the decision abstract is processed. This operate can implement customized logic that’s related to postprocessing, for instance, updating the decision abstract to a CRM system.
The next diagram illustrates the beginning of name and post-call (abstract) Lambda hooks that combine with Salesforce to lookup and replace case information, respectively.
Listed below are the steps we stroll you thru:
- Arrange Salesforce to permit the customized Lambda hooks to lookup or replace the case information.
- Deploy the LCA and Salesforce integration stacks.
- Replace the LCA stack with the Salesforce integration Lambda hooks and carry out validations.
You want the next conditions:
Create a Salesforce linked app
To arrange your Salesforce app, full the next steps:
- Log in to your Salesforce org and go to Setup.
- Seek for App Supervisor and select App Supervisor.
- Select New Linked App.
- For Linked App Title, enter a reputation.
- For Contact E-mail, enter a sound electronic mail.
- Choose Allow OAuth Settings and enter a worth for Callback URL.
- Below Accessible OAuth Scopes, select Handle consumer knowledge by way of APIs (api).
- Choose Require Secret for Webserver Move and Require Secret for Refresh Token Move.
- Select Save.
- Below API (Allow OAuth Settings), select Handle Shopper Particulars.
- Confirm your identification if prompted.
- Copy the buyer key and client secret.
You want these when deploying the AWS Serverless Application Model (AWS SAM) utility.
Get your Salesforce entry token
For those who don’t have already got an entry token, you must receive one. Earlier than doing this, just be sure you’re ready to replace any purposes which are utilizing an entry token as a result of this step creates a brand new one and will invalidate the prior tokens.
- Discover your private info by selecting Settings from View profile on the highest proper.
- Select Reset My Safety Token adopted by Reset Safety Token.
- Make word of the brand new entry token that you simply obtain by way of electronic mail.
Create a Salesforce buyer contact document for every caller
The Lambda operate that performs case look-up and replace matches the caller’s cellphone quantity with a contact document in Salesforce. To create a brand new contact, full the next steps:
- Log in to your Salesforce org.
- Below App Launcher, seek for and select Service Console.
- On the Service Console web page, select Contacts from the drop-down checklist, then select New.
- Enter a sound cellphone quantity beneath the Telephone subject of the New Contact web page.
- Enter different contact particulars and select Save.
- Repeat Steps 1–5 for any caller that makes a cellphone name and take a look at the mixing.
Deploy the LCA stack
Full the next steps to deploy the LCA stack:
- Comply with the directions beneath the Deploy the CloudFormation stack part of Live call analytics and agent assist for your contact center with Amazon language AI services.
- Just remember to select ANTHROPIC, SAGEMAKER, or LAMBDA for the Finish of Name Transcript Abstract parameter. See Transcript Summarization for extra particulars.
The stacks take about 45 minutes to deploy.
- After the primary stack reveals
CREATE_COMPLETE, on the Outputs tab, make an observation of the Kinesis knowledge stream ARN (
Deploy the Salesforce integration stack
To deploy the Salesforce integration stack, full the next steps:
- Open a command-line terminal and run the next instructions:
Use the next desk as a reference for parameter selections.
|The Area the place you will have deployed the LCA resolution
|The consumer identify of your Salesforce group that has permissions to learn and create circumstances
|The password related to your Salesforce consumer identify
|The entry token you obtained earlier
|The patron key you copied earlier
|The patron secret you obtained earlier
|The login URL of your Salesforce group
|The Salesforce API model (select default or v56.0)
|The Kinesis knowledge stream ARN (CallDataStreamArn) obtained earlier
- After the stack efficiently deploys, make an observation of
PostCallSummaryLambdaHookFunctionArnfrom the outputs displayed in your terminal.
Replace LCA Stack
Full the next steps to replace the LCA stack:
- On the AWS CloudFormation console, replace the primary LCA stack.
- Select Use present template.
- For Lambda Hook Perform ARN for Customized Begin of Name Processing (current), present the
StartOfCallLambdaHookFunctionArnthat you simply obtained earlier.
- For Lambda Hook Perform ARN for Customized Put up Processing, after the Name Transcript Abstract is processed (current), present the
PostCallSummaryLambdaHookFunctionArnthat you simply obtained earlier.
- Make it possible for Finish of Name Transcript Abstract isn’t DISABLED.
Validate the mixing
Make a take a look at name and be sure to can see the start of name AGENT ASSIST and post-call AGENT ASSIST transcripts. Seek advice from the Discover stay name evaluation and agent help options part of the Live call analytics and agent assist for your contact center with Amazon language AI services submit for steerage.
To keep away from incurring fees, clear up your assets by following these directions when you’re completed experimenting with this resolution:
- On the AWS CloudFormation console, and delete the LCA stacks that you simply deployed. This deletes assets that had been created by deploying the answer. The recording S3 buckets, DynamoDB desk, and CloudWatch log teams are retained after the stack is deleted to keep away from deleting your knowledge.
- In your terminal, run
sam deleteto delete the Salesforce integration Lambda capabilities.
- Comply with the directions in Deactivate a Developer Edition Org to deactivate your Salesforce Developer org.
On this submit, we demonstrated how the Stay-Name Analytics pattern mission can speed up your adoption of real-time contact heart analytics and integration. Somewhat than constructing from scratch, we present tips on how to use the present code base with the pre-built integration factors with the beginning of name and post-call Lambda hooks. This enhances agent productiveness by integrating with Salesforce to lookup and replace case information. Discover our open-source project and improve the CRM pre- and post-call Lambda hooks to accommodate your use case.
In regards to the Authors
Christopher Lott is a Senior Options Architect within the AWS AI Language Providers group. He has 20 years of enterprise software program growth expertise. Chris lives in Sacramento, California and enjoys gardening, aerospace, and touring the world.
Babu Srinivasan is a Sr. Specialist SA – Language AI companies within the World Extensive Specialist group at AWS, with over 24 years of expertise in IT and the final 6 years targeted on the AWS Cloud. He’s captivated with AI/ML. Exterior of labor, he enjoys woodworking and entertains family and friends (typically strangers) with sleight of hand card magic.