journal

Journal

Fashion vertical first release

Developed with industry partners, our fashion-industry vertical brings sector-specific terminology, data and decision logic into a dedicated AI context layer and is now entering its fine-tuning phase.

, , ,

fig. 03 — location layer

Overview

We released our dedicated industry vertical for the fashion sector. Developed over the past nine months with four development partners from the industry, the vertical brings fashion-specific terminology, data structures, processes and decision logic into a dedicated AI context layer.(fashion.vertical.3.1)

The vertical is now entering its fine-tuning phase, in which the underlying context is being validated against additional company data, workflows and operational use cases. Fashion companies interested in joining the development programme can contact our sales department.

What defines an industry vertical

A verticallm industry vertical provides the foundational context an AI system needs to operate within a specific sector. It contains the industry’s terminology, core entities, relationships, process structures, business rules, performance indicators and common document types.1

This context is organised into a structured knowledge layer that connects operational data with the meaning, logic and dependencies behind it. It enables the AI to understand not only individual terms or data fields, but also how products, suppliers, customers, locations, processes and decisions relate to one another.

Inside the Fashion Industry vertical

Over the past nine months, the Fashion Industry Vertical has been enriched with context covering collections, seasons, styles, colourways, sizes, SKUs, bills of materials, fabrics, trims and supplier structures.

It also includes the operational logic behind collection planning, sourcing, production, assortment management, allocation, replenishment, inventory, sell-through, markdowns, returns and omnichannel fulfilment. This provides a sector-specific foundation for applying AI to fashion planning, supply chain management and operational decision-making.

References

  1. Each vertical is developed with domain partners over an extended engagement, capturing sector-specific terminology, data structures, processes and decision logic. This context is encoded into our proprietary knowledge graph, where entities, relationships and constraints are connected into a queryable layer the model reasons across.” ↩︎

Related content

Why LLM’s are really poor at strategy

Ask leading AI models the same strategic question and you may expect different perspectives. Research suggests the opposite: they often converge on the same fashionable answers.

Read more

The downfall of cloud-AI frontier models

general-purpose model intelligence is becoming increasingly abundant, cheaper and interchangeable, while proprietary company context remains scarce and valuable.

Read more

Small language models.

Small Language Models (SLMs) are compact, often task-specialised language models that require less compute, cost, and infrastructure than frontier-scale LLMs.

Read more