LLM engineering
Model selection, context engineering, fine-tuning and distillation for specialised domains.
XTO is an Australian AI research and development company. We explore, prototype and test AI apps, intelligent web services and AI-powered technologies that could move people and industries forward.
Optimistic · Technical · Creative · Curious · Relentless
We believe the next era of human progress will be written in intelligence, and that Australia should be one of its authors. XTO exists to build frontier AI that is genuinely useful, safe by design and beautifully engineered. From Australia, for the world.
Model selection, context engineering, fine-tuning and distillation for specialised domains.
Autonomous agents that plan, use tools and complete multi-step work, with human oversight built in.
RAG pipelines, vector search and knowledge graphs that ground AI in trusted sources.
Detection, segmentation, inspection and visual understanding for the physical world.
Deployment, monitoring, versioning and scalable inference across cloud and on-premise.
Red-teaming, evals, guardrails and governance aligned with Australian standards.
AI applications. Experimental apps with intelligence at their core, exploring software that can understand, reason and act on behalf of the people using it.
iOS · Android · Web · DesktopAI web services. Research into APIs, platforms and cloud infrastructure that put intelligence on tap, with reliability, security and scale designed in from day one.
APIs · Platforms · Cloud · EdgeAI-powered technologies. Our R&D engine: custom models, agent architectures and deep technology that underpin everything we explore.
Models · Agents · Vision · R&DWe're exploring how frontier AI can work in the hands of the frontline: on-device language models, agents, real-time vision and voice, prototyped as secure apps for iPhone, iPad and Android.
Compact, distilled language models that run on the phone itself: instant, offline-capable and private by default.
Agents that act across ERP, CRM, email and documents from the home screen, with approvals built into every action.
Inspection, defect detection and safety checks from the device camera, at the edge and without a signal.
Hands-free copilots for field crews, with speech recognition, summaries and translation in real time.
Snap invoices, forms and site reports; multimodal models extract, validate and sync them to core systems.
Encrypted on-device inference, least-privilege access and Australian data residency for sensitive workloads.
We map the problem, the data and the value at stake, and define what success looks like in numbers.
Working prototypes in days, not months, tested against real inputs so decisions are made on evidence.
Production architecture, security, evaluations and interfaces, built to scale and built to last.
Staged roll-outs with guardrails, monitoring and human oversight from the first user onward.
Continuous evaluation and improvement as models, data and our understanding grow.
Human, societal and environmental wellbeing: AI should benefit individuals, society and the environment.
P.01We respect human rights, diversity and individual autonomy.
P.02Systems are inclusive, accessible and tested for unfair bias.
P.03Privacy rights are upheld and data is protected at every layer.
P.04AI operates reliably and as intended across its lifecycle.
P.05People know when they are engaging with AI and can understand it.
P.06There is a timely way to challenge outcomes that significantly affect people.
P.07Humans remain accountable for AI systems and their outcomes.
P.08Guided by Australia's AI Ethics Principles: not a poster on the wall, but engineering requirements, tested like any other. We never train models on personal information without explicit permission, and human oversight is built into every system that makes consequential decisions.
Read our Responsible AI PolicyXTO is not currently trading, so we aren't taking on client work or recruiting. If you have a question or comment, send us a message and we'll respond within two business days.
General enquiries & privacy