The article describes the first steps that will help to prepare for creating an Agent and include identifying the tasks solved by the Agent and the required functionality.
Before creating a new Agent, define:
- The goal of creating an Agent (for example, to help a client book a flight or onboard a new employee);
- The target audience (for example, the company employees or university students)
- The problems of the Bot User that the Agent is designed to solve (for example, selecting the right flight or finding the right office)
- The advantage of creating an Agent over other services used by your company (for example, what are the tasks that can be solved by the Agent but can’t be solved with the help of a website or app).
This should be kept in mind while creating a Dialogue Script.
The Agent’s goal and tasks are closely connected with its target audience. For example, an Agent designed to onboard new employees is supposed to cover a broader range of topics than the one designed to help those employees who have been working for the company for a while. You need to come up with a list of the conversational topics the Agent is supposed to support, the questions it should answer, and the tasks it should solve.
For example, topics can be as follows: general information about the company, issuing certificates, business travel documents checklist, etc.
Examples of questions: “How do I receive a certificate?”, “What documents need to be drawn up for a business trip?”
Examples of tasks: send a certificate to ERP, provide information about the company, change the password, find the necessary product in the product catalog, and extract some information about an employee from the database.
Having decided on the goal, target audience, and tasks of the Agent, you need to think over the functionality of the future Agent.
Prepare a draft of the possible Dialogue scripts (the interaction between the Agent and the Bot User can be displayed as a diagram, for clarity) and decide what functionality you need. This could be:
- Recording and storing data received from the Bot User;
- Extracting information from knowledge bases;
- Extracting information from external systems via integrations with such systems
- Sending information to external systems via integrations with such systems.
Once these steps are completed, you can move on to preparing data for the Agent.