# Integrations

```{warning}
This chapter is partially covered. The runtime dependencies below are described
in the architecture pages, but they have no dedicated integration guides yet.
External business-data connectors are on the roadmap and not implemented.
```

## Current Runtime Dependencies

| System | Role | Where it is described today |
| --- | --- | --- |
| Energy Logserver / OpenSearch | Identity source for login, chunk and vector index, document catalog, retrieval pyramid projections, ingestion job queue. | [Architecture overview](../03-0-0-Architecture/overview.md), [Authentication and RBAC](../08-0-0-Security/auth-rbac.md) |
| Ollama-compatible LLM runtime | Answer generation, abstractive summaries, query expansion. | [Answering](../03-0-0-Architecture/answering.md), [Model coordination](../03-0-0-Architecture/model-coordination.md) |
| Valkey / Redis | Shared coordination plane for model placement, leases, residency state and scheduler state in multi-process deployments. | [Model coordination](../03-0-0-Architecture/model-coordination.md) |
| Docling | Document parsing during ingestion. | [Document ingestion](../03-0-0-Architecture/document-ingestion.md) |

## Planned

External data connectors are on the roadmap. The planned capability set covers
registering a connector with validated configuration and credentials, checking
connector health before synchronisation, synchronising external schemas into
usable snapshots, preserving partial sync results, and blocking unsafe or
unauthorised structured-data queries.

Until connectors ship, this chapter should grow one page per runtime dependency,
each covering supported versions, required configuration, health checks and
failure modes.