AI adoption & integration
Strategic roadmaps and tactical integration of AI, with outcomes rooted in value capture, not vendor narrative.
Capability · AI, ML & Neural Architectures
Evolving AI Systems.
AI and machine learning now span a wide spectrum for enterprises. We service the full range as a research-focused firm.
01 Capability
At one end — foundational models and agentic workflows are accessible to everyone. At the other end — custom AI, knowledge systems, and distributed neural architectures. We operate across the full range with active research into the software of evolving neural architectures.
02 Domains
Agents & automation
Agentic systems and workflow automation that use AI LLMs as a primitive.
Search, RAG & retrieval systems
Retrieval-augmented generation over proprietary knowledge. Semantic search, vector indexes, hybrid retrieval pipelines.
Information architecture & knowledge graphs
How data is shaped, structured, and connected. Ontologies, knowledge graphs, and AI reasoning structures.
Specialised model designs
Custom model architectures and fine-tuning — LoRA, adapters, specialised LMs, multimodal indexing across rich data streams.
Graph neural networks
AI systems that leverage graphs for transparency, agility and cost reductions in training.
Recommendation engines
Personalisation at scale. Graph-based retrieval, ranking models, multi-stage intelligence pipelines.
On-device intelligence
Local inference on Apple Silicon unified-memory architectures. Private, offline-capable, zero-copy fast.
Distributed intelligent systems
Multi-model AI pipelines coordinating across economic boundaries.
03 Our position
The role of AI in the world has changed dramatically in the last few years, and things are not likely to slow down. Our approach revolves around a few grounded truths:
Data as the primary enabler
How the data is represented is the most consequential decision in determining the value of any AI. This is true across foundational LLMs, custom inferencing on self-trained models, and multi-model architectures alike. Start with the data representation.
Size and cost will drive markets
Size is not the determiner of value — a well-shaped knowledge architecture can make a small model outperform a large one. Our view is that the next decade belongs to smaller, lower-cost, specialised AI systems that feed on structured knowledge — often private or proprietary — not ever-larger monolithic models.
Trust as a systemic requirement
As intelligent systems are increasingly used as sources of truth, verifiability of the information they build from (proof of lineage) and how they are treated (data sovereignty) become architecture concerns at the root of the system. Such systems will appreciate in value.
The decade belongs to smaller, low-cost, adaptable systems — not to the largest model. Sovereignty and information guarantees will be paramount.
Smaller.
Low-Cost.
Adaptable.
Trusted.
Low-Cost.
Adaptable.
Trusted.
04 In practice
Where this work shows up.
On-device intelligence
An on-device search and intelligence engine built ground-up for Apple Silicon's unified memory. msearch executes composable pipelines built from intelligence primitives — embed, search, infer, reason, train, index — in parallel across CPU and GPU. A zero-copy architecture means hyper-fast, each operation executing directly on the hardware it prefers.
Complex data across network nodes
Where msearch executes intelligence pipelines on-device, mgraph synchronises the state those pipelines use — graphs, vectors, tensors, streams, documents — across devices, edges, and datacentres. mgraph is designed for distributed intelligence systems, treating collaboration as a generalised problem, solving it as a primitive.
AI SaaS for property managers
AI productivity platform for property management — from proof-of-concept to production across a multi-national operation. Encompassed feedback data pipelines for training, AI agentic automation, Graph/RAG search.
AI in tourism
Designed and delivered AI-powered search, discovery and ticketing application for Gold Coast tourism. AI-driven search on domain modelling, recommendation engine and embedded personalisation capabilities.
Works with
Other engagement models
01
Engineering & Development
When the system needs an architect who has shipped retrieval and inference pipelines into production.
Read more →02Strategic Advice
AI adoption strategy, model and IP positioning, technology due diligence — informed by what we ship.
Read more →03Partnerships & Equity
When value is best captured with a co-builder.
Read more →Capabilities
01
Software Design
The runtime under the model — infrastructure, transport layer, real-time systems.
Read more →02Architecture & Infrastructure
Where inference runs — cloud, edge, on-device — and how data flows between.
Read more →03Technology Strategy
When the AI bet is upstream of an architecture decision.
Read more →Contact
Tell us what you're working on.
We reply within two business days. If a call would be faster, book a thirty minute conversation.
Prefer to talk?
Book a thirty-minute call