Perplexica is a free and transparent search engine that understands what you ask it and finds exactly what you are looking for, all boosted by the latest generation of artificial intelligence. Sounds familiar, right?
Oh yeah, it’s like Perplexity.ai except it’s free, open source and allows you to search the web in depth to find the answers to all your questions. To do this, the tool uses machine learning algorithms based in particular on searching by semantic similarity. Basically, it is able to understand the meaning of your question and find the most relevant sources.
This way you get clear and sourced answersserved on a silver platter without having to spend hours sifting through pages of results to find the information you are interested in.
No cookies that spy on you or resale of personal data on the sly and your searches remain confidential. In terms of features, Perplexica offers several very practical search modes:
-
The mode Copilot (still in development): it generates queries to find the most relevant Internet sources. Rather than using only the context provided by SearXNG, it directly visits the top results to find the most relevant sources for your question.
-
The mode Normal : it processes your query and performs a classic web search.
-
THE Focus Modes : specialized modes to meet specific needs, such as academic research, YouTube search, calculations via Wolfram Alpha or even Reddit search for discussions and opinions.
And on a technical level, it uses a well-designed meta-search engine called SearXNG. It is he who retrieves the results, sorts them, cross-references them, to keep only the cream of the crop. The advantage is that you always have up-to-date information, without latency. And as I said in the intro, it also relies on local language models like Llama3 and Mixtral, via the Ollama API which allows it to refine the results even further.
And to install it, it’s super simple:
- Clone the GitHub repository of Perplexica :
git clone - Rename the file
sample.config.tomlinconfig.tomland fill in the necessary fields (API key, etc.) - Run the command
docker compose up -d - Wait a few minutes for the installation to complete
- Access Perplexica via
from your browser
Once it’s in place, you can even use Perplexica as default search engine in your browser. Just go to settings, add a new search engine with the URL /?q=%s and voila !
Afterwards, of course, it’s still a young project, so there are little things to refine and the devs are working in particular on finalizing the Copilot mode, adding a settings page, features like search history, and support for local language models.
But even as it is, I think it’s definitely worth testing Perplexica, if only to see what an AI search engine that is both ethical and transparent might look like ^^.
Special thanks to Lorenper for the info!