Utility Tools
Utility tools provide various helper capabilities that enhance your AI agents' functionality, from gathering additional information to searching the web to calling other agents.
Dynamic Forms
The Dynamic Forms tool lets the AI request additional information from users mid-conversation by creating custom forms. When the AI determines it needs more details to proceed, it creates a form with specific fields, the form appears in the chat interface, the user fills it out and submits, and the AI continues with that information.
Forms can include different input types: single-line text fields for names and short answers, multi-line text areas for descriptions and details, dropdowns for selecting from predefined options, and checkboxes for yes/no confirmations.
For example, the AI might request project details:
AI: To help you better, I need some additional information:
[Form: Project Name (text), Priority (dropdown: High/Medium/Low), Description (long text), Rush Order (checkbox)]
User: [Fills out and submits form]
AI: Thanks! Based on your input, here's what I recommend...
Set Chat Name
This tool automatically names conversations based on their content. The AI analyzes what was discussed and generates a descriptive name, making it easier to find conversations later and keeping your chat history organized.
Web Search
The Web Search tool lets agents search the internet for current information. When information isn't available in your knowledge base, when users ask about current events, or when general research and fact-checking are needed, the AI can search the web and incorporate what it finds.
Example:
User: What's the latest news about AI regulations?
AI: [Searches the web] Based on recent news: The EU passed new AI regulations last week... The US is considering similar legislation...
URL Content Extraction
This tool extracts and reads content from specific web pages. When a user provides a URL or when the agent needs information from a known source, it fetches the page content, extracts the readable text, and can answer questions about that content.
Use cases include summarizing articles, extracting specific information from pages, analyzing competitor content, or researching any online content.
Example:
User: Summarize this article: [URL]
AI: [Fetches URL content] Here's a summary of the article:
Main Points:
- Point 1...
- Point 2...
Sub-Agents
The Sub-Agents tool lets your main agent call other agents as specialized assistants. When the main agent receives a request, it can determine that a specialized sub-agent would help, call that agent with relevant context, receive the results, and continue with that information.
Sub-agents work well for specialized tasks like coding, analysis, or writing; complex workflows that benefit from multiple perspectives; and domain-specific expertise.
To configure sub-agents, first create the specialized agent, then add it as a sub-agent to your main agent. You can optionally give it a custom name that helps the main agent understand when to use it.
Example:
User: I need a technical analysis of this code and a user-friendly summary for stakeholders.
Main Agent: I'll get our technical expert to analyze the code... [Calls Technical Analysis sub-agent]
Now let me summarize for stakeholders... [Calls Communication sub-agent]
Here are both perspectives: ...
PDF Generation
This tool creates PDF documents from content, useful for reports, formal documents, printable content, and professional deliverables. The AI generates document content, formats it appropriately, and produces a downloadable PDF file.
Best Practices
When using dynamic forms, only ask for essential information, use appropriate input types, provide clear labels, and keep forms concise. Users are more likely to complete short, focused forms.
For web search, use it when your knowledge base doesn't have the answer. Verify important information from multiple sources when possible, and be aware that web content may be dated.
Sub-Agents
- Create focused, specialized agents
- Keep sub-agent prompts concise
- Consider response time with multiple agents
- Test sub-agent interactions
Tool Combinations
These utilities work well together:
- Research Flow: Web Search → URL Content → Artifact Creation
- Guided Workflow: Dynamic Forms → Processing → PDF Generation
- Expert System: Question → Sub-Agent Analysis → Summary