
AI Development.
The Group develops its own AI models, and builds bespoke model systems for institutional partners. XSIS and Qurec are the Group’s platform expressions of the practice.

The premise.
Most enterprise software written before 2023 assumes deterministic inputs and rule-based decisioning.Foundation models invert that assumption. Categories built on the old assumption cannot be patched into the new world; they have to be rewritten. The Group rewrites them — both for its own platforms and for partners that engage us to build under their brand.
How we operate.
We build at three layers: the model layer (proprietary fine-tunes and from-scratch model development), the workflow layer (structured-output orchestration that replaces forms and SaaS UIs), and the decision layer (audit-ready outputs with traceable provenance).Engagements are delivered with on-premise, VPC, and air-gapped deployment options for regulated counterparties.
What the practice does.
Proprietary model development
Development and fine-tuning of language, vision, and structured-data models against the Group’s own infrastructure or a partner’s data perimeter.
Production decisioning
Structured-output orchestration replaces legacy form-based workflows. Every decision is logged, versioned, and auditable.
Multimodal understanding
Text, vision, audio, and tabular ingestion against a unified semantic index.
Bespoke partner engagements
Selective engagements where the Group designs, trains, and ships AI systems for institutional partners under their brand.
The Group’s platforms in this field.
The other fields.
See the full Group portfolio.


















