NERVE and CORTEX: How One Ingestion Layer Makes Every Approved Source Searchable

Most intelligence teams are not short of data. They are short of a reliable way to get data in, keep it trustworthy and find it again. AIC is addressing that with NERVE, an intelligence data fabric that sits beneath CORTEX and turns approved source material into preserved, normalised, searchable information.

The division of labour is the point: NERVE gets information in, and CORTEX turns information into intelligence.

Why a Separate Ingestion Layer?

Device extractions, mailboxes, file shares, databases, licensed feeds, sensor streams, geospatial files and legacy data lakes all behave differently. Integrating each one directly into an analytical tool creates a permanent maintenance burden: bespoke importers, duplicated mapping logic and a provenance trail that degrades every time data moves.

NERVE inverts that. A source is integrated once, through a versioned connector or import path. The original is preserved before anything is interpreted. Transformation runs then map source-specific structures into canonical messages, documents, calls, contacts, locations, files and transactions, while keeping the source-specific detail that may matter later.

Where NERVE Stops and CORTEX Starts

NERVE handles data and information. CORTEX owns intelligence judgement. In practice, NERVE can record that a message contains a telephone number and that two records share the same normalised number. Whether those records support a judgement is an analyst decision, made in CORTEX, with confidence, competing hypotheses and review attached to it.

That boundary keeps both products honest. NERVE solves heterogeneity, scale, provenance and discoverability. CORTEX solves requirements, analysis, products and accountable decisions. Neither has to pretend to be the other.

How the Two Actually Talk to Each Other

The integration is service-mediated. CORTEX does not query NERVE’s object store, metadata database, search index or broker. It calls stable search and evidence contracts, so NERVE’s internals stay private implementation detail and can evolve without breaking the analyst experience.

Two things travel with every analyst search: the CORTEX workload identity, which proves which application is calling, and the end-user authority context, which decides what that particular analyst may see. A trusted service identity cannot be used to bypass an individual’s current information scope.

Results come back as stable evidence references rather than copies. CORTEX keeps the reference as a citation, and resolving it later recovers the exact object revision, the locator for the matched passage and the lineage back to acquisition. Because references survive reprocessing, an assessment written before a parser upgrade still resolves afterwards.

Independent, But Better Together

NERVE does not require a CORTEX requirement in order to work. Standing datasets, licensed feeds, enterprise repositories, sensor streams and historical archives can all be ingested, preserved and made searchable on their own terms. Where a requirement or legal authority does exist, it is recorded as a reference rather than assumed.

Equally, CORTEX remains the intended analyst-facing consumer. The combination means an analyst can search holdings from inside the lifecycle tool they already use, without learning every underlying forensic schema or source system.

AI as Enrichment, Not Authority

OCR, transcription, translation and embedding generation are optional enrichment stages, each tracked as a derivative with its own provenance. Machine-derived text and labels improve discovery, and remain visibly derived. Core ingestion and search do not depend on public cloud AI, which matters for local, disconnected and sovereign deployments.

What We Are Claiming, and What We Are Not

NERVE is currently published as an intended-state specification. It describes designed behaviour rather than delivered, tested or accredited capability, and performance or scale figures will follow measured acceptance evidence. We would rather be precise now than revise claims later.

If you want to understand the analytical half of the picture first, read more about CORTEX, or see the full NERVE product overview. To discuss your own source landscape, get in touch.