Tools

Tools are external functionalities or APIs that can be used by a large language model (LLM) to perform specific actions or retrieve information beyond its inherent capabilities. AIP Agents equipped with tools use a chain-of-thought, which has a slower time to first token and less predictable. Limiting the number of tools assigned to the LLM generally results in a more guided and controlled interaction.

Types of tools

There are five types of tools available:

• Actions: Gives your agent the ability to execute an Ontology edit.
• Object query: This tool specifies the object types that the LLM can access. You can add multiple object types and specify accessible properties to make queries more token-efficient. The object query tool supports filtering, aggregation, inspection, and traversal of links for configured objects.
• Function: This allows the LLM to call any Foundry function, including published AIP Logic functions. The latest version of the function is automatically used, but you can also specify a published version for more granular control.
• Ontology semantic search: This tool can use a TypeScript function or a vector property to retrieve relevant information. Ontology semantic search is different than Ontology context because semantic search is only called if the chain-of-thought reasoning chooses to do so.
• Request clarification: Found under Advanced tools, this tool allows the agent to pause its execution, request clarification, then resume the execution with that clarified request. If the LLM is in the midst of chain-of-thought and needs more clarity from the user, the LLM can call the Request clarification tool.

Inspect reasoning

When deployed in edit mode, view mode, Workshop, or AIP Threads, users can select the "eye" icon to investigate the chain-of-thought reasoning process used to generate a response.