Working with Tools in Gen AI Builder

Suggest edits

What are Tools

Tools in Gen AI Builder are Griptape-powered components that allow AI Assistants and Structures to perform actions beyond basic text generation. They provide the ability to interact with external systems, perform calculations, retrieve live data, and execute custom logic.

Tools make AI applications active — capable of not just answering questions, but taking meaningful action.

In short: Tools give your AI Assistants and Structures superpowers by extending them with actionable capabilities.

For a deep dive, see: Tools explained.

Why use Tools

  • Extend LLM capabilities: Perform actions that LLMs alone cannot do.
  • Integrate with external systems: APIs, databases, services.
  • Enhance user experience: Provide dynamic responses based on real-time data.
  • Enable business processes: Automate workflows triggered by user interactions.
  • Modularize logic: Package reusable functions that can be invoked across Assistants and Structures.

Without Tools, your Assistants and Structures are limited to retrieving knowledge and generating text. With Tools, they can act, compute, query, and integrate.

When to use Tools

Use Tools when you want Assistants or Structures to:

  • Query external APIs (e.g., stock prices, weather, exchange rates).
  • Perform calculations or business logic.
  • Access internal databases or services.
  • Enrich conversations with live or dynamic data.
  • Automate multi-step processes.
  • Provide custom retrieval or transformation logic.

How Tools fit into Gen AI Builder

Runtime flow with Tools:

User Input → Assistant → Retriever → Knowledge Bases → Retrieved Content
→ Rulesets → Behavioral Guidance
→ Tools → Action Execution
→ Memory → Context Maintenance
→ LLM → Response → User Output

Tools can also be invoked:

  • Directly by Assistants via Tool calls.
  • Indirectly as part of a Structure execution.
  • In Data Source pipelines for data transformation.

Key features of Tools

  • Implemented as Griptape Tools (Python classes).
  • Packaged as Zip files or deployed from GitHub.
  • Configurable with:
  • Tool Config file (YAML / Python)
  • Environment variables (API keys, configuration)
  • Executable by:
  • Assistants (via LLM-driven tool invocation)
  • Structures (as components of workflows)

Typical Tool patterns

  • API Connectors: Query external services.
  • Calculators: Perform complex calculations or conversions.
  • Data Fetchers: Retrieve data from internal systems.
  • Data Transformers: Preprocess or post-process data.
  • Process Automators: Orchestrate multi-step workflows.

Getting started

See Create a Tool for a full step-by-step guide.

Typical workflow:

  1. Develop Griptape Tool locally.
  2. Package Tool as Zip.
  3. Upload Tool Zip to Data Lake.
  4. Create Tool in Gen AI Builder.
  5. Assign Tool to Assistants or Structures.

Related topics

← Prev

Working with Threads in Gen AI Builder

↑ Up

Agent Studio in Gen AI Builder

Next →

Gen AI Builder Hub

Could this page be better? Report a problem or suggest an addition!