The Universal Chatbot Connector integrates the SendSafely Dropzone into chatbot and AI agent workflows to automate the secure collection of sensitive customer information.
The connector is platform-agnostic and designed to work with any chatbot that has the capability to:
- Make API calls from within a chat session,
- Store a variable in the chat session,
- Add a button/prompt asking the user to answer a question, and
- Update the chat conversation using data from an API call
When to use the Universal Chatbot Connector
Customers should leverage the Universal Chatbot connector for any chatbot or AI agent for which SendSafely does not already have a native pre-built Dropzone integration.
The Universal Chatbot Connector is best suited to AI agent workflows, which typically provide the functionality described above. The connector may also support chat workflows involving human agents, depending on the chat platform's capabilities. To determine whether your target chat platform is compatible with our Universal Chatbot Connector, please contact your SendSafely account representative.
Requirements
In order to deploy the Universal Chatbot Connector you must:
- Verify, with your SendSafely account representative, that you target platform is capable of integrating with our Universal Chatbot Connector.
- Set up a SendSafely Service Account and enable its Dropzone. Refer to the “Initial Dropzone Setup” section of the Hosted Dropzone article for instructions.
-
Have permissions within AWS to set up and deploy new resources. Specifically you will need permission to:
- Create a new Lambda Function
- Create a new DynamoDB Table
- Create a new SNS Topic (Simple Notification Service)
- Create a new API (API Gateway Service)
- Create a new AWS Secret (AWS Secrets Manager )
- Define a custom IAM Role for the Lambda Function
- Acquire the CloudFormation template for the connector from your SendSafely account representative.
Deployment
Deploy the CloudFormation template in your AWS environment in the us-east-1 region.
- This deployment will generate a webhook URL, which you will then paste into the Notification settings for your Service Account's Dropzone.
- You will also need to populate the Dropzone Validation Key for that Dropzone in AWS Secrets Manager for the Chatbot Connector to use.
Chatbot Configuration
As the Universal Chatbot Connector is system-agnostic, configuration will vary from platform to platform and should be done live with your SendSafely account representative and a member of our Solutions Engineering team.
The high level workflow steps required to use the Chatbot Connector are outlined below:
-
Have the chatbot invoke the connector endpoint to obtain a unique upload id:
POST Request: https://your_connector_url/ ( {"action":"generateId"} )
Expected Response: {"UploadId": "aaabbbcccddd"}
-
Store the UploadId as a variable in the chat session
var upload_id = $Response.UploadId
-
Insert a link to Chatbot Dropzone Upload Page within the chat conversation, along with instructions for the end-user to upload their files using the link. Note that the link passes the upload id, dropzone id and your SendSafely hostname along with the email address of the end-user as shown below. If you chatbot has not already asked the user to provide their email address you should insert a step to do that earlier in this flow.
Link: https://files.sendsafely.com/html/cb_upload.htm?id=%upload_id%&e=%user_email%&dz=dropzone_id&h=yourcompany.sendsafely.com
-
Add a button/prompt asking the user to confirm when they are finished uploading their files. When the button is pressed, the following API call should be made to get the Secure Link:
POST Request: https://your_connector_url/ ( {"action":"getLink","uploadId":"%upload_id%","email":"%user_email%"}
Expected Response: {"Result": "Success",
"SecureLink": "https://yourcompany.sendsafely.com/receive/?thread=ABCD-1234&packageCode=aaabbbccc#keyCode=xxxyyyzzz"}
- Insert the SendSafely secure link from the previous step into the conversation and hand the user off to an agent.
If you do not want to hand the conversation off to a human agent for reviewing the submitted files/data, you can choose to have the AI agent retrieve the files/data for further processing. This will require an additional API integration and is not directly performed by the Universal Chatbot Connector at this time.
Comments
0 comments
Please sign in to leave a comment.