AutoSearch

Combine Perplexity Sonar API high performance search with AI enabled Automation Tool

Comment

Inspiration

Automation Tool provides Low Code feature to make easy integration with external service API such as Perplexity Sonar API and can offer web frontend interface to the user to enable better user experience.

  • One simple solution with N8N Automation Tool could be to introduce and configure an HttpRequest Node and appropriate settings.

  • However, ModelContextProtocol (MCP) is a better approach to integrate API with multiple endpoints in a unified way, like USB-C as physical connector.

Perplexity Sonar API MCP Server describes integration and usage from well known MCP clients like Claude Desktop or Cursor .

Claude Desktop option has been used at the beginning of development of backend service.

Finally, Sonar API MCP Server integration has been leveraged with automation tool platform such as N8N, to benefit of low code combined with automation features and multiple Tools/Nodes such as MCP client as well.

Hence it will be possible to benefit of Perplexity Sonar API High Perfomance Search operated with Automation Platform

What it does

Perplexity Sonar API MCP server is used as special "node" search service which receives search requests either interactively from users either automatically triggered thru web form submission , either later on periodic schedule either as webhook (chat or social network)

Some other features such as image upload analysis with base 64 encoded data or from url have been added to the existing Perplexity Sonar API MCP server codeset.

The user can also interact about financial services thanks to Stripe API MCP integration

How we built it

We have chosen N8N automation platform for low code development environment and designed autosearch application as N8N custom workflow with following features :

  • Web Form Frontend enabling user to specify : either text query with image url, either image file upload selection

  • Backend processing Multimodal (text and/or image) user query after routing + file size check followed by calling image analysis extension of existing Perplexity Sonar MCP server module.

  • Special Router Module has been introduced to parse the user multimodal request ( text, image thru text url either thru file upload) and redirect to appropriate backend module

  • Image File size is controlled earlier to get better user experience and avoid useless sonar api computation

The custom workflow has been developed on my local self hosted dockerized N8N configured with communautary n8n nodes mcp extension

Challenges we ran into

  • integration of custom image analysis extension of existing Perplexity MCP perplexity-ask Node JS service on N8N > build a custom N8N self hosted docker image by preinstalling the Perplexity MCP perplexity-ask module and overwriting the index.js with my custom version implementing the image analysis

Accomplishments that we're proud of

  • Benefits of mixing Low Code Automation Tool with Perplexity Sonar API High Performance Search using MCP (Model Context Protocol) nice to have technology

  • web interface generated by the automation tool

What we learned

  • Perplexity Sonar API various search criteria and its MCP proposal usage

What's next for AutoSearch

  • implement different triggers such as webhook, periodic batch, chat(bot) or social network app user interface to get more visibility

  • finalize the Proof Of Concept proposal by taking into account potential future evolutions around Perplexity Sonar MCP and/or its integration

Built With

Share this project:

Updates