Tool chaining is a sequence in which one tool invocation produces information that guides the next. It extends an agent beyond choosing one tool and answering immediately, allowing repeated interaction with results. This supports multistep tasks but creates a growing need to track intermediate results within the model’s context. section 8.12
Restaurant sales and reviews
To answer which Google rating belongs to the highest-grossing restaurant location, an agent reads the spreadsheet, analyzes sales figures, identifies the location, and then searches the web for its current rating. section 8.12
Figure 8.7 The general agent architecture allowing for multiple actions per response
The architecture diagram helps visualize how an agent can return to the language model after each tool call and continue acting. section 8.12
