Tools
Tools are used to connect the AI Agent’s conversational intelligence to the outside world. They let the agent perform actions, access external data or execute logic beyond what the language model can do on its own.
When an AI Agent decides it needs to use a tool, it formulates a structured request (based on the tool’s defined parameters), executes it, and then continues the conversation with the result. This enables dynamic, goal-driven interactions where the agent doesn’t just talk, but can act.
LiveHub AI Agents platform supports two types of tools:
-
Customer-defined tools – typically used to interact with other systems via programmatic APIs, for example, get the weather forecast or book an appointment. These tools must be configured by customer as described below.
-
Pre-defined tools – Live Hub AI Agents platofrm provides a few pre-defined tools that address common tools usage scenarios, for example, call transfer to a human agent. These tools are accessible to any agent that wants to use them.
Tool structure
Each tool consists of the following elements:
-
Description – describes the tool’s purpose.
-
Schema – definition of parameters accepted by the tool; for each parameter name, description and type are defined.
-
Business logic – implementation of tool’s business logic; for example for custom-defined tools you define URL and HTTP method, and the tool’s business logic sends RESTful API request to the specified URL with parameters defined by schema.
You only need to specify schema for tool’s 'input' parameters. LLM automatically parses the tool’s response (result), and automatically identifies its schema.