Fractal Neurons
Contact

Technology

Meaning has structure.

A single idea runs through everything we build: meaning has structure. Retrieve on that structure — not on shared vocabulary — and you reach two problems keyword and vector search cannot solve: finding the same mechanism across domains, and giving agent fleets memory that is durable and governed.

Proof on real public data

The engine finds the same cause in a language it was never given.

We trained a structural bridge on ~250 example links between two incident records that share no vocabulary, format, or industry — NASA aviation safety reports and federal pipeline filings — then, for aviation incidents it had never seen, asked it to retrieve the pipeline accidents with the same underlying cause.

21.9% Chance shuffled control
38.4% Text similarity words only, no learning
56.6% Structural bridge learned correspondence

same-cause precision@5 · 307 held-out aviation queries → 1,251 pipeline incidents · +18.2pp over words alone · top-20 exhibit 18/20 same-cause, human-audited

one pair · match 0.90 Here is a single retrieval, whole. Read the two records first — then watch why every search you own would rank them as strangers, and what the engine read instead.

01 The two records

Aviation Weather

Encountered unforecasted severe turbulence at FL250. No convective activity was present, and it was a relatively clear evening … the aircraft abruptly experienced large pitch and attitude changes … a continuous battle to keep the wings level.

NASA ASRS · AV-2086578

Pipeline Natural Force Damage

Water contained within the body of the valve expanded into ice as temperatures dropped; the expansion resulted in a hairline crack of the valve.

PHMSA · PL-20230047

02 The vocabulary gap

Two accidents, not one word in common. The language never overlaps — so anything that matches on language is blind here.

Aviation
  • turbulence
  • FL250
  • convective
  • clear evening
  • pitch
  • attitude
  • wings level
  • aircraft
Pipeline
  • valve
  • water
  • ice
  • temperatures
  • expansion
  • hairline crack
  • body

03 Why search misses it

Keyword and vector search rank by how much language two records share. With zero overlap, this true match scores like noise — the pipeline record never surfaces for the aviation query.

Words only near chance

Matches on shared vocabulary — of which there is none.

Structural bridge match 0.90

Matches on the cause structure — which is identical.

04 What the engine read instead

Strip the words away and the same three-step cause is left standing in both records. That skeleton — not the vocabulary — is what the engine retrieves on.

  1. A latent force, unforecast

    AV clear evening, no convective activity — turbulence that was never predicted

    PL water inside the valve, temperatures quietly dropping

  2. acts on a system that read as normal

    AV the aircraft was flying fine at altitude

    PL the valve was intact and holding

  3. and it fails, abruptly

    AV large pitch changes — a battle to keep the wings level

    PL the expansion cracks the valve body

what the engine saw Clear air that turns violent; still water that turns to ice. Zero shared vocabulary — the match is the causal structure.

The agentic layer · associative memory

Discovery is a property of the memory — not a query you run against it.

The substrate that surfaced those cross-domain pairs is the same memory we give to agent fleets. Agents store the experience they live, recall it by situation, and link experience to knowledge — and those links get stronger the more they earn their keep.

  1. 01

    Store what it experiences

    Decisions, findings, corrections, and outcomes are written to a per-agent store as they happen — passed through a novelty gate so repeated experience merges instead of piling up.

  2. 02

    Recall by situation

    When an agent meets a situation it has been in before, the memory recognizes it and brings back what was actually used last time — personal experience layered over shared, immutable knowledge.

  3. 03

    Links that strengthen with use

    Every time a recalled memory is used, the link that surfaced it is reinforced; links that surface and get ignored weaken and fall away. Memory that improves with experience — never by overwriting what is shared.

Governed for fleets

  • Global knowledge stays static and shared; personal experience grows per agent — so one agent's experience never pollutes the base every agent reads.
  • Ownership, provenance, permissions, and human oversight travel with the knowledge and stay visible.
  • Knowledge is promoted deliberately — personal, then team, then shared — never leaked upward by default.

Point that same associative memory at a body of knowledge instead of an agent's task history, and the links that make recall better become the correspondences no keyword search would surface. That is where the aviation ↔ pipeline pairs came from.

The deployment model · Dedicated Research Instance

A dedicated instance that runs inside your own walls.

We work with each partner inside a Dedicated Research Instance: a confidential, per-partner environment where their permitted internal material sits alongside selected external knowledge — bounded so that only they can reach it, with source boundaries preserved at every step.

  • A private, bounded environment provisioned per organization.
  • Confidential data stays inside your walls — never pooled, never used to train a shared model, never leaving the instance.
  • Provenance and scope attached to every result.
  • Security described as inspectable properties — never as slogans.

Who we partner with

Experts with unique data
and high discovery value.

We work with a handful of partners in confidential technical domains. If you hold unique data with high discovery value, we'd like to hear what you're working on.

info@fractalneurons.com Company on LinkedIn

Confidential details only after a secure route is agreed.