Testing Your First AI Agent

Step 1: Click on the Create Agent button in the left sidebar and give your Agent a name and a description. Once done, click on the Create Agent button.

Step 2: Now, upload the required documents so that the testing system can understand the particular requirements of your agentic chatbot. This helps the platform analyse your requirements and generate relevant scenarios for your chatbot.

Step 3: After the initial analysis, you can select which types of tests you want to generate. The platform offers a plethora of test categories to choose from. For instance, the Personality & Tone category helps you generate scenarios, ensuring that your agent is responding in a professional manner. You can also select multiple categories to generate a wide range of tests.

Step 4: Once done, click on the Generate Test Scenarios button. In the background, a team of specialised AI agents will start generating multiple test scenarios in parallel.

Step 5: As you can see, the test scenarios have been generated. You can filter them based on the test categories you had selected, and the system also assigns a complexity and risk level to each scenario. For instance, a scenario for tone adaptation in case of delivering bad news would suggest how your bot should respond.

Step 6: Once you are satisfied with the scenarios, enter your bot/agent's API URL and click on Run Evaluation.

Step 7: Under the Evaluation Results tab, you can see a breakdown of the execution results across different metrics, such as whether the conversation was relevant, if the agent stayed on topic, and more.

Note: If you have a voice bot, the system can test it by automatically generating audio for the test cases and evaluating the transcribed audio responses from your agent to ensure they are relevant.