Context, without the guesswork.

No. ODE generates the associative model from cues in the data itself, this means that entity resolution and relationships emerge from overlapping context, without predefined schemas or domain-specific ontologies. The only schema you need to provide is the data itself.

No. ODE works as a virtual associative interface over your existing sources. Ingest is non-destructive; sources stay where they are and remain untouched. You can safely tinker, modify, and delete models in ODE without affecting the sources at all.

ODE operates on REL-1's associative data format, a universal data abstraction. You can bring your own tabular, graph, vector, document, and NoSQL data. ODE's associative engine is format-agnostic, so as long as you select a mapping for the incoming data (we have several presets for tabular, etc.) the data gets associated and ODE can be used to reason across them through the same API and MCP.

Every piece of virtualized data carries lineage by default, the file, row, column, and timestamp information is captured when it gets virtualized. From any entity in the system, you can trace back to original data point and source system. If you're storing memories or agentic data, you can configure sources at ingestion to signify whatever is most helpful to you (a source by session, or source by agent-archtectype, a source by workflow, etc.).

Yes. ODE is a developer toolset that connects to REL-1 which is a deterministic associative engine; the API and SDK don't require a model in the loop. You can even manually run MCP commands if you'd like without an agent if that's your thing. Agents and LLMs are augmented and catered too, not prerequisites.

Yes. The MCP is lightweight and ships with deep performance and security customizations so you can run it inside your own infrastructure.

REL-1 is the deterministic associative engine that holds all of contextual virtual data; ODE is the developer toolset, the API, MCP, and SDK, that allow connection with and operations within REL-1. Since the associative data works in its novel data format, ODE gives the tools to bring data in, do operations, and have current systems interoperate within and across it.

Most context engines are both AI-driven and AI-optimized. That means they can introduce bias when their agents determine what's important or true, hallucinations when agents are doing mapping and at extraction, and typically only can be read-out effectively with AI systems, often only their agents or selected providers.

The associative engine doesn't use AI to map data semantically, instead it relies on repeated context cues within data to build a contextual map procedurally. As it only uses the information from within the data and metadata, all associations in REL-1 are deterministic in that they surface a contextual link inside the data, but don't assign semantic prejudice. The virutal model also is easy to interoperate with humans and other systems as well via api or JSON.