Skip to main content

AI Tool Selector

Use the Tools menu for manual selection and connection controls.

Manual tool selection with Google Search selected and a one-of-three counter.

Tool Selector Overview

Use the Tool Selector in Chat and Workshop to choose available tools and integrations.

The Tool Selector lets you choose tools and integrations for the task you want to do. In manual Chat selection, choose up to three entries. Automatic tool selection discovers eligible tools separately while work proceeds.

In Chat, turn off Auto-Tool first. New chats open with Automatic Tool Selection switched on, and while it is on tools are chosen for you — picking them by hand has no effect. Open the Tools menu in the message box, switch Auto-Tool off, and then choose up to three tools.


Integrations

Access Integrations in Secure Chat and Workflows from the Tool Selector. You can connect integrations directly from the Tool Selector, or on the Connections page in Workshop.

Looking for beta integration access?


AI-Powered Search Tools

One of the most common uses of AI tool-calling is for web search. The Hatz AI platform offers an enhanced toolset with AI-powered search capabilities, providing users the flexibility to select specific tools based on their needs. Available search tools can include Google Search, Perplexity, Firecrawl, Tavily, Exa, and web search subagents depending on what is enabled in your workspace.

By understanding the unique functionalities and suitable use cases for each available search option, users can more effectively tailor their search activities to meet specific objectives, thereby improving both efficiency and productivity.

1. Google Search

  • Purpose: Utilizes Google for rapid and efficient information retrieval.

  • Use Case: Ideal for quick fact-checking, retrieving concise data, or immediate browsing.

  • Speed: Optimized for fast responses leveraging Google's capabilities.

  • Depth: Provides surface-level information for quick access.

2. Firecrawl

  • Purpose: Designed for comprehensive data collection from specific websites.

  • Use Case: Suitable for gathering extensive datasets and performing detailed analyses.

  • Speed: Slower than Quick Search due to the volume of data being processed.

  • Firecrawl helps retrieve content from supported web pages. Ask for the specific pages or material you need and check the returned result.

3. Perplexity

  • Purpose: Focuses on delivering the most current information, especially on news and events.

  • Use Case: Best for staying updated on the latest news, trends, and current event analyses.

  • Speed: Provides a balance of speed and depth, ensuring up-to-date and contextual information.

  • Depth: Offers detailed insights into current events, providing context and updates.

4. AI Search Engines

  • Exa and Tavily are AI search engines that can provide the same kind of results that Google, Perplexity, and Firecrawl offer. Try different tools to see what your preferences are!

Tool Usage Table

The following table provides a quick reference for when to use each available AI-powered search tool:

Use Case/Need

Quick Search

Firecrawl

Perplexity

Tavily

Exa

Quick fact-checking

✔️

✔️

✔️

Retrieving concise information

✔️

✔️

✔️

Gathering information from a website

✔️

✔️

✔️

Extensive data analysis

✔️

✔️

✔️

Staying updated on current events

✔️

✔️

✔️

Understanding news trends

✔️

✔️

✔️

Immediate answers

✔️

In-depth exploration of a topic

✔️

✔️

✔️

✔️

Check the result

After selecting a tool, describe your task and inspect the tool result. Use Connect or Reconnect when shown; an organization-managed connection needs its administrator. Model choice, tool selection and connection permissions are separate controls. For a classic workflow, configure its step tools in the builder; a saved automation uses its saved tools.

Watch: start a chat and use tools

Choose a tool for a specific chat task

  1. Start a chat and keep Auto selected for the model.

  2. Open Skills and tools → Tools. Turn off Auto-Tool to choose manually, then enable the permitted tool your task needs.

  3. Describe the task and the output you want. The walkthrough selects Python Code Execution and asks it to add fictional project hours.

  4. Open the completed activity and check what actually ran. Compare the answer with the original inputs.

Watch: Select tools for one task

Turn off Auto-Tool to choose a tool manually, run a small Python calculation and inspect the completed activity. This example keeps Auto selected for the model.

Did this answer your question?