Ix Platform

Shared infrastructure around DSG-JIT.

Ix provides platform services that build and maintain persistent, versioned system models across many domains. Deterministic ingestion builds structure; LLMs are optional at query time.

Layered Architecture

Platform services + domain layers. Documents become versioned subgraphs backed by content-addressed blocks. Relationships come from deterministic extraction + resolution.

Sources (Enterprise Knowledge + Artifacts)

File systems • Docs/Wikis • Tickets • Chat • Code repos • PDFs/CAD/Datasets

Filesystem / Drives

Docs / Wikis

Issues / Tickets

Chat

Repos / Artifacts

ingest

Domain Layer

File System & Internal Knowledge Layer

Discovery → Normalize → Chunk (CAS) → Extract + Resolve Links → Index Updates

Discovery

Connectors enumerate sources; collect paths, owners, timestamps, types.

Normalize

Canonicalize formats; extract text/structure; normalize identifiers and refs.

Chunk + CAS Write

Content-addressed blocks with provenance anchors; versioned subgraphs.

Extract + Resolve Links

Deterministic entity/ref extraction; evidence-backed link resolution + merges.

Incremental Index

Update graph indexes + optional vector index; maintain audit log.

writes

Core Platform Services

Ix Shared Platform Services

Common infrastructure used by all domain layers and applications

DSG-JIT (core engine)

Graph + Versioned Docs

Document subgraphs (rev lineage)
CAS evidence store (blocks)
Anchors + provenance
Diffs / time-travel queries

Storage + Indexing

Metadata store + ACL tags
Vector index (optional)
Graph indexes (neighbors/types)
Audit logs

Query & Reasoning Runtime

Hybrid retrieval (graph + vec + meta)
Evidence-grounded outputs
Tooling APIs for apps
Optional LLM integration (not required)

Telemetry

Event-based, content-free metrics

• ingest health/lag • connector errors • query latency • feature usage counts • stack traces (redacted)

No node/edge semantic payloads. No document text. No private embeddings exported.

Ix Telemetry API

Collector

serves

Application Layers

Robotics/Autonomy • Neuroscience • Genomics • Digital Twins • Embodied/Agentic AI

domain writes

Robotics & Autonomy

Neuroscience

Genomics

Digital Twins

Embodied / Agentic AI

Notes: The graph is built without LLMs. Links come from deterministic reference extraction (IDs/URLs/paths), entity extraction (rules/dictionaries/classical NLP), and resolution/merging with evidence anchors. LLMs can be layered later in the Query Runtime, but they are not required to build or maintain the graph.