About EDEX

Filesystem-native retrieval for AI teams who cannot miss.

EDEX replaces vector stacks with deterministic filesystem indexing, so your models can traverse source repositories, knowledge bases, and regulated archives with perfect provenance. It runs as an MCP server, speaks tool calls natively, and stays deployable inside your perimeter.

Vector-free

0 embeddings

Latency

sub-second search

Cost

70% lower tokens

Filesystem-first indexing

Hierarchy-aware ingestion respects folders, permissions, and structured metadata without flattening context.

Tool-call retrieval

Agents request exact slices through deterministic tools, keeping answers grounded and auditable.

Deployment control

Run EDEX beside your data, with on-prem isolation and policy gates for every request.

How it works

01

Mount the knowledge surface

EDEX connects to your repos, drives, and archives without reshaping the data.

02

Query with intent

Models call tools that resolve paths, permissions, and context before retrieval.

03

Deliver traceable answers

Every response includes a provenance trail so teams know exactly what was read.