How to Perform Real-Time Web Search via Pabbly

[tokyo2025]

I’m trying to perform real-time web searches via Pabbly in the simplest way possible.
The goal is to define search intent for SEO articles, and this use case has massive demand among content creators and marketers.

Right now, I’m struggling to find any method to integrate real-time search models like gpt-4o-search-preview using Pabbly. It seems only the standard chat endpoint is supported, and no clear way exists to connect to OpenAI's or Claude's search-capable APIs.

This feature is critical, because:

• Search intent discovery depends on fresh, live web results
• GPT-4o and Claude 3.7 Sonnet support this already — just not through Pabbly
• There is huge potential for automating SEO research using AI

I’m stuck and would really appreciate any guidance:

• Is there any method to enable real-time search with Pabbly?
• Can we configure a custom endpoint or use a workaround?

Please help — I believe this feature will benefit many users, and I’d love to know how to make it work.

 

[Preeti Paryani]

Hello @tokyo2025,

You can try enabling the Map option in the OpenAI step and manually enter gpt-4o-search-preview as the model. Then test if it returns the search results you need. This may allow access to real-time search.

 

[tokyo2025]

Thank you for the suggestion.

I’ve tried manually entering gpt-4o-search-preview as the model and enabling the Map option as you described. However, based on OpenAI’s documentation and my own tests, this approach does not activate the real-time search functionality.

The issue is not just that the search tool isn’t working — it’s that when using the /v1/chat/completions endpoint, the model behaves exactly like a regular chat model, ignoring the search-preview capabilities entirely.

To properly activate the real-time search feature, OpenAI requires calling specific endpoints like /v1/assistants or /v1/threads/runs along with tool definitions in the payload ("tools": [{"type": "retrieval"}, {"type": "function"}, {"type": "browser"}], etc.). Without this, the model simply processes the prompt as a standard conversation, without performing any actual web search.

In short, changing only the model name doesn’t enable real-time search — the endpoint and payload structure also need to support tools.

If there's any workaround within Pabbly to define a custom endpoint and request body, please let me know. Otherwise, I hope this feature will be added in the near future.

 

[Preeti Paryani]

Hello @tokyo2025,

Please try using AI Assistant (Pabbly): Ask Assistant feature and let us know how it goes.

 

[tokyo2025]

prompt:
Search Google for the search term “chatgpt image1” and list the titles that appear in the SERPs from 1st to 5th place.

This is the result that was output.



Unfortunately, with the current information, we are unable to pinpoint the top 5 titles in Google search results for the search term "chatgpt image1". However, there are some related articles about ChatGPT's image generation and image recognition capabilities.

 

[Preeti Paryani]

Hello @tokyo2025,

Thank you for your query regarding real-time search capabilities using AI. After reviewing the available options, we’ve found that real-time web search is currently not supported via the API versions of either OpenAI or Claude models.

As an alternative, we recommend trying the Perplexity AI: Create Chat Completion action to see if it provides the search results you require. While it may not fully replicate real-time browsing, it could offer relevant and updated information suitable for your use case.

Please feel free to test it and let us know how it works for you

 

[tokyo2025]

Thank you for your response.

Just to clarify — the limitation seems to be on Pabbly’s side, not on the API providers themselves.

According to official documentation:

• OpenAI explicitly states that gpt-4o-search-preview is designed to perform real-time web searches via the Chat Completions API:
  

  “GPT-4o Search Preview is a specialized model trained to understand and execute web search queries with the Chat Completions API. In addition to token fees, web search queries have a fee per tool call.”
  — source

• Anthropic also confirms in their official announcement that Claude now supports web search through the API:
  

  “Today, we're introducing web search on the Anthropic API—a new tool that gives Claude access to current information from across the web... by enabling the web search tool when making requests to the Messages API.”
  — source

So technically, both OpenAI and Anthropic support real-time search functionality via their APIs.

Regarding your recommendation to use Perplexity AI:
While I appreciate the suggestion, it's important to note that Perplexity provides summarized results from indexed data, not true real-time web querying. This distinction matters for use cases like SEO and trend monitoring where freshness is key.

That said, I now understand that Pabbly currently does not support these advanced AI features due to endpoint and payload limitations, and I accept that this functionality is not available at the moment.

Thank you again for your support — I hope this capability can be added in the future.

 

[Preeti Paryani]

Hello @tokyo2025,

Thank you for the update. You can submit your request along with the relevant research and API documentation to our roadmap. Our integration team will review it, and we’ll keep you informed if there are any updates.

Pabbly

Vote or request new Pabbly features. Subscribe to get updates about new features from Pabbly.

roadmap.pabbly.com

 

[tokyo2025]

I’ve submitted the official request to the roadmap.
Thank you for your support and guidance throughout this process.
Looking forward to seeing how things develop!