Tools
Last updated
Last updated
Tool calling allows language models to use external systems, APIs, and services. Models can perform functions, get real-time data, do calculations, and integrate with your infrastructure.
Based on a users question, the model can identify if it needs anything from an external function. It can select the right function, call it, and use the results in its response. This means responses are no longer limited to text responses.
The tool calling process follows a structured workflow:
User Query Processing
When a user submits a request, the model analyses if it needs external tools for an accurate and complete response.
Tool Selection
The model compares the available tools and selects the most appropriate based on the users request and tool description.
Parameter Extraction
The model takes the relevant parameters of the user's request and formats them according to the tool's specified schema.
Function Execution
The tools are called using the relevant parameters and results are returned to the model.
The model then uses the results in a coherent, natural language response for the user.
Tool calling transforms a static chatbot into a capable assistant. They can access live data and perform real actions. You can get stock levels, check system statuses, search documents, and interact with databases without leaving the chat. It reduces context switching and ensures responses use accurate and up-to-date information.
Administrators are able to define tools through structured schemas. You can specify function names, descriptions, required parameters, and expected return formats.
Aiimi Insight Engine ships with a series of out of the box tool definitions. You can also create your own tools.