Skip to main content
Adaline’s Playground offers a safe sandbox environment to test how your LLM interacts with tools. If your tools are configured with request object, the Playground will automatically invoke the tools and continue the conversation with the tool’s response. Learn more about how to configure tools in the adding tools section.

​
Playground Tool Call Workflow

When the LLM determines it needs external data, it generates a tool call request that appears as a distinct content block in the conversation flow. The role that makes the actual tool call is Assistant. Learn more about roles in Adaline in the roles and content blocks section. Below is a typical Playground flow that requires the use of tools. The process is the same as creating role-based prompts, but, in this case, this happens inside the Playground. Suppose you have added a tool that answers “yes” or “no” to a prompt. The image below shows the user’s query and the LLM’s response: A tool responding yes or no In such a case, you may want to force the LLM to a more precise answer. For example, it may answer yes but also suggest to take a look at the weather. To do so, you can modify the answer field and run the prompt again: A tool responding yes or no Proceeding with this approach, you could directly ask the LLM to respond based on the actual weather forecast: A tool responding yes or no As the image shows, the LLM has autonomously switched to "answer": "no". This happened because the LLM has no access to the actual weather forecast. Also, it is forced to only answer “yes” or “no”, so it forced itself to such an answer.