NERVE is live across local, private and approved-cloud deployments. Book an evaluation
Where the line sits
Acquisition is kept separate from assessment on purpose. The record of what was collected stays independent of any later judgement about it, and either product can be adopted on its own.
Information infrastructure
NERVE
Owns everything up to the point where information becomes searchable and citable.
Acquires approved sources, from archives and devices to repositories, feeds and streams.
Preserves the original bytes and records a hash so byte identity can always be re-checked.
Normalises decoded content into one canonical model, with the Original, Canonical and Derived layers kept distinct.
Publishes typed records with lineage, so every result resolves back to accepted bytes.
Carries the dataset handling rule with every record and filters retrieval by the searching account’s authorisation.
Intelligence lifecycle
CORTEX
Owns everything from a stated requirement through to a released, reviewable product.
Captures direction and requirements, then tasks collection against them.
Assesses source reliability, lineage independence and identification risk before corroboration is claimed.
Applies structured analytical technique, explicit assumptions and recorded challenge.
Produces reports with exact-span annotation that never overwrites the original.
Controls release and feedback under separate permissions from collection.
Why the split matters
NERVE never interprets. It does not grade a source, assert that two records describe the same person, or decide whether something is significant. Those are CORTEX decisions, made under separate permissions and recorded separately from the acquisition evidence.
The interface
Typed records
The canonical model
Documents, messages, media, records and events arrive as typed records sharing one common envelope, so CORTEX reads a message from a mailbox and a row from a database through the same fields.
Lineage
Citation back to bytes
Records carry the acquisition they came from and the processing that produced them. An annotation made in CORTEX can therefore be traced to an exact span of the exact accepted original.
Handling
The rule travels with the record
The dataset’s purpose and handling rule are attached to each derived record, so downstream analysis in CORTEX inherits the same constraint rather than re-declaring it.
Authorisation
Filtered before counts are built
Retrieval is filtered by the requesting account’s authorisation before results, counts and facets are assembled, so unauthorised material never shapes a CORTEX view. Broad administrative permission does not automatically grant protected content access.
Reprocessing
Improvements propagate
When extraction improves, NERVE rebuilds derived records from the retained original. Citations stay stable, so CORTEX products made earlier remain resolvable after reprocessing.
Failure state
Partial is visible, not silent
A job that completes as Partial is reported as Partial. CORTEX sees which items failed, so coverage gaps are an input to assessment rather than an unknown.
Running them together
01
Register NERVE as a holdings source
02
Agree the record contract
03
Carry identifiers and lineage through
04
Enforce handling at query time
05
Keep products resolvable over time
What neither product claims
A hash proves bytes, not people
A hash match confirms that two files are byte-identical. It does not confirm who created them, when, or why.
A matching identifier is not an identity
Two records sharing an identifier are two records sharing an identifier. Deciding they describe the same person is an assessment made in CORTEX, recorded as such.
A location observation is not presence
A recorded location is an observation attached to a record. Concluding that a person was there is a separate analytical step with its own confidence.
Environment and accreditation are shared
Both products run inside an environment you control. Hosting, network boundary and accreditation remain a shared responsibility between AIC and the operating organisation.
Adoption
NERVE on its own
Run NERVE where the problem is getting scattered holdings into one searchable, citable, governed place. It serves your existing tools through the same service contracts, with no requirement to adopt an analytical workflow.
CORTEX on its own
Run CORTEX where the problem is the intelligence lifecycle itself: direction, collection, assessment, product and controlled release. It can consume sources other than NERVE.