Skip to content
FlexiPedia logo

AI Agents with Web-Search, MCP Servers, and Knowledge Base

Unlock the full potential of AI Agents with seamless integrations and enhanced functionalities. Whether you're managing complex data with the AI Knowledge Base or leveraging web search capabilities, these tools simplify processes and reduce errors. Perfect for businesses looking

Flexito OmniAI
Beginnerv4
Read in:

Listen to an audio overview

A short, AI-generated podcast-style summary of this article.

Unlock the full potential of AI Agents with seamless integrations and enhanced functionalities. Whether you're managing complex data with the AI Knowledge Base or leveraging web search capabilities, these tools simplify processes and reduce errors. Perfect for businesses looking to streamline operations and improve efficiency.

## Overview

AI Agents now offer significant enhancements that simplify the creation of more robust and dependable agents. With built-in support for AI Knowledge Base, MCP servers, and Web Search, you can avoid complex workarounds and manual functions.

## 1. Setting Up the AI Knowledge Base

### Enable OpenAI Provider Ensure the OpenAI provider is connected in your workspace. This connection is essential for creating and linking vector stores and files. Navigate to Integrations > OpenAI.

### Preparing Files for Vector Stores Files need valid extensions such as .pdf, .xls, .doc, etc. Without these, the system cannot read or index them, leading to errors or rejection. You can upload files from your computer or provide a file URL.

### Creating a Vector Store Head to the Vector Store section to create a new store. Assign it a name and, if needed, an expiry date (set to 0 for no expiry). Select the files you've uploaded in the File section.

### Building the Knowledge Base Once you've set up a vector store, proceed to create a Knowledge Base. Provide a name and an optional description, then select the associated Vector Stores. This Knowledge Base is now accessible to AI Agents.

## 2. Setting Up OpenAI Responses Model Inside an AI Agent ### Setting It Up Within an AI Agent, navigate to Settings → Models. Choose OpenAI - Responses and select your preferred model. The default is gpt-4.1.

### Using Tools with OpenAI Responses

You can now directly link MCP Servers and the Knowledge Base to the AI Agent, eliminating the need to trigger separate AI functions or actions. This simplifies processes and reduces errors.

### Web Search Integration

Activating Web Search lets you search websites directly from within the AI Agent. This feature is handy if you lack MCP servers or a knowledge base, allowing AI agents to retrieve answers straight from the web.

  • ### Restricting Domains

  • Format: domain.com, sub.domain.com (comma-separated).

  • Do not include https:// or .

  • Example: google.com, app.google.com

### Limitations by Model For Web Search

  • GPT-4.1-mini and GPT-4.1-nano: Web search is supported, but domain restriction is not supported.

  • GPT-4.1 (default model): Supports web search with domain restriction.

### Notes on GPT-5 Usage (Not Recommended)

  1. GPT-5 offers deep reasoning but results in slower response times and more timeout errors.

  2. Requires more tokens due to extended reasoning. A minimum of 2,000 tokens per reply is recommended. Note: Higher token usage increases costs.

  3. Use GPT-5 only for complex reasoning tasks. For simpler tasks, opt for lighter models to maintain efficiency.