Skip to Content
DocumentationAI and EnginesRevornix AI

Revornix AI

Revornix AI is no longer just an isolated chat box. It is built on top of several layers:

  • A default Revornix AI model
  • Document and section context
  • Vector retrieval
  • Knowledge graph expansion
  • MCP client and MCP server capability

1. Default model slots

To use Revornix AI properly, you need at least:

  • A default Revornix AI model

And many deeper workflows also depend indirectly on:

  • A default document-reading / summary model

That is because Revornix AI is not only generic chat. It shares the same knowledge production pipeline used by documents, sections, graphs, and podcast-related flows.

2. Context sources

Revornix AI does not rely on a static prompt alone.

For document- and section-scoped scenarios, it can combine:

  • Document or section Markdown
  • Related document excerpts
  • Vector retrieval results
  • Knowledge graph expansion results
  • MCP-returned capability output

That is why Revornix AI behaves more like an operator over your knowledge base rather than a standalone external assistant.

3. MCP is still a core part of the stack

On the client side, Revornix supports:

  • HTTP MCP
  • Std MCP

On the server side, MCP capability is built into the API and mainly organized under:

  • api/mcp_router/common
  • api/mcp_router/document
  • api/mcp_router/graph

Because of that, Revornix AI can both call into document and graph capabilities itself and be called by external MCP clients in return.

4. Official hosted and self-configured resources now coexist

Revornix AI does not depend on a single resource source.

You can:

  • Configure your own model providers and models
  • Use public community models after forking them
  • Use official hosted models in official deployments

Official initialization seeds:

  • The official provider Revornix
  • The default seeded official model entry gpt-5.4

But that should not be read as a guarantee that all official routing is permanently hard-bound to one single upstream model name.

5. Plan access and runtime limits

Default Revornix AI model selection, actual runtime access, and some official hosted resources are connected to plan restrictions.

So a more accurate mental model today is:

  • Revornix AI is a composite capability built from default models, default resources, knowledge context, and MCP
  • Availability depends not only on whether you configured an API key, but also on whether the selected resources are accessible under the current plan
Last updated on
Section Sharing and Co-creationCustom Model