What Perplexica is, how it runs, and the open-source system to own next

What Perplexica is

Perplexica is an open-source answer engine: a program you host yourself that takes a question, searches the web, and writes a reply with the sources it used. People often describe it as an open-source take on Perplexity, the hosted answer service. The project now ships under the name Vane, so the Vane repository is the same codebase, and the old Perplexica GitHub address redirects there.

What Perplexica does with a question

Perplexica builds its answer from a fresh web search rather than a stored index. It sends the query to a search backend, fetches the top pages, splits and reranks their text, and hands the top passages to a language model that writes the answer with citations. SearXNG powers the search in the standard install, with Tavily and Exa listed as coming. Effort is adjustable: a speed mode answers quickly, a balanced mode reasons a little, and a quality mode runs a longer search-and-scrape loop for deeper research. You can limit sources to the web, academic papers, or online discussions, and you can also ask questions about files you upload.

How Perplexica runs

Perplexica runs as a self-hosted web app, and the documented install is a single Docker command. That command starts an image bundling the app and its SearXNG search backend, after which you open localhost:3000 and point the app at a model: a local model through Ollama, or a hosted provider such as OpenAI, Anthropic, or Groq with an API key. The project says it runs entirely on your own hardware, and SearXNG is what keeps the search itself private.

Why people look for something else

People look past Perplexica for two different reasons, and the right fix depends on which one applies. Some want a different answer engine: a chat window over their own documents, another search backend, or a tool that asks no model choice of them. Others find that a search box does not cover the work they have, which is a company memory that persists, connections to the tools they use, and a person reviewing what an agent changes. The rename from Perplexica to Vane sends some people looking for the old name too.

Two directions: another answer engine, or a system you own

Once you leave Perplexica, the choice is between another answer engine and a system you own. Another answer engine replaces one search box with another and changes little else. A system you own adds the parts a search box cannot have: agents that do the work, memory that persists, and connectors to the apps you already run.

Kortix is the open-source AI Operating System: your agents, their skills, your company memory, and every connector in one git repo you own, with the agents working on real cloud computers. Any model provider with your own keys, or the ChatGPT plan you already pay for. Run it on Kortix Cloud, in your VPC, or on your own on-prem network, and self-host is free.

Work in Kortix does not land on its own. Session work reaches main through a change request, and merge is default-deny for agents, so a person reviews each change before it becomes part of the company.