Quickstart
Introducing Caddy
Caddy is an AI-powered assistant acts as a copilot for customer service agents, empowering them to provide high-quality, actionable advice quickly and securely. When agents Caddy a question, Caddy will provide an answer based on all the content sources it’s been provided and give the agent references to the information.
Caddy is designed to be used in settings where customer service advisers are supervised by experienced staff who can verify Caddy’s responses. Caddy’s responses are sent to a supervisor to review before they are made available to the customer-facing adviser. This “human in the loop” validation system is employed to mitigate risk, ensuring advice accuracy and reliability.
Environments
Caddy is built using AWS Serverless architecture and currently integrates with Google Workspace. We plan on building a Teams version later in the year.
Installation
Adding users
There are two levels of user: supervisors and advisors.
Advisors can ask questions to Caddy, and will only see responses that have been approved by a supervisor (they will see a supervisor’s comments if the supervisor does not approve Caddy’s response).
Supervisors have elevated privileges. They can: approve Caddy’s responses, add users (including other supervisors) and remove users.
The mechanism for adding both types of user is the same. To add a user, a user with supervisor permissions types ‘/addUser’ in the chat bar of a supervisor space.
Removing users
Supervisors can remove other users.
Reviewing a Caddy response
Approve the response
Reject the response