EOGIYAFACTORY

Technology

An AI team with a facility
attached to it

Most software for this industry is written by people who have never had to keep something alive overnight. Ours is written next to our own tanks and proven on our own floor before it reaches anyone else.

Two engines

VISION · VLM

Machine eyes on the tank

Condition, freshness and distress, read from video. A person can check a tank three times a day; a model checks it continuously, and every judgement it makes becomes another row in the mortality dataset.

SMALL LANGUAGE MODEL · sLLM

A model that has read the rulebook

Quarantine, customs and hygiene regulation, answered with the clause attached. Half of live export is paperwork, so we built the reader — and reading regulation turns out to transfer.

what grew out of them
Live

SFC operations

Water-quality soft sensing, WMS, export document automation

In build

AI-HACCP

Food-safety plans and daily checks, answered from the standard

In build

Aquaculture SaaS

Subscription water and trade platform for farms and distributors

Delivered

Public-sector AX

City administrative AI, industry matching platform

Engine 01 · Vision

Judging something that moves

Industrial vision inspects objects that hold still. Ours watches animals that swim, hide and change by the hour. It is the harder problem, and that is where we think the value of the dataset lies — a long record of watching living things is not something you can buy.

  • Continuous monitoring per tank — ammonia, salinity, pH, UV
  • Anomalies raised to a person while there is still time to act
  • Nine species handled, turning into a per-species mortality record
  • Judged on outcome, not accuracy claims: 15% → 5% across 46 shipments
Water quality dashboard
Warehouse management

Engine 02 · sLLM

Small, because the answer has to be right

Quarantine and hygiene regulation is narrow, deep and unforgiving — the kind of problem where a general model is confidently wrong. That is why we chose to build small. It also involves documents that must never leave the building.

  • G-WMS — export documentation on an SLM, in production today
  • Citations, not assertions — every answer points at the clause it came from
  • Portable — quarantine rules, then food-safety rules, then administrative rules. Same pipeline, different corpus

In the field

Things that already touch cargo

Standard live container
Packing

A box that is the same every time

Standardised live containers replaced improvised polystyrene. Freight can only be optimised once the unit stops changing.

Sourcing app
Sourcing

BISFE app

Built with public-sector partners to open sourcing channels and find principals.

WMS
Operations

WMS & G-WMS

Stock, movements, counterparties and export paperwork in one system.

Public sector

A regulation-reading team,
widening its subject matter

We are often asked why a seafood company bids on civic AI. We did not start from seafood. We started from documents that carry legal weight, and those exist in every agency. What we are widening is not our industry but the range of documents we can read.

ProgrammeDomainOur roleStatus
Municipal administrative AIRAG question answering for civil servantsPlatform build & operationDelivered
Industry matching platformB2B company–investor matchingPlatform buildDelivered
Port state control supportBallast water inspection intelligenceService designIn design

Client names and contract values are published only within the scope each agency has agreed. Full references available on request.

4
Patents
3 · 3
Registered designs · trademarks
5
In-house research institute
ISO 9001
INNO-BIZ certified

Build something with us

Consortium bids, deployments and proofs of concept. We bring what we have already run on our own floor.

Start a conversation