AI literacy foundations
How large language models work, where they fail, and the principles for safe, effective use. A shared vocabulary and a shared floor of competence across your organisation.

Managed sovereign AI · In service
Zara LM is Phoenix ICT's managed sovereign AI service, built and operated on Australian Government controlled infrastructure, trusted to operate at Secret and in service across government.
Zara LM combines a secure chat client, a retrieval-augmented knowledge layer grounded in curated authoritative sources, and an agentic orchestration layer that carries complex work through to a review-ready draft.
Every output is verified and approved by an accountable officer. The platform is designed to produce a draft that is between 70 and 80 per cent complete, so that human judgement is preserved.
All processing remains inside the accredited boundary. No prompt, document or telemetry leaves the environment.

Zara LM brings together conversational AI, grounded knowledge and orchestrated agents, each layer built and sustained by Phoenix ICT.
A secure, multi-model conversational interface that lets users work naturally within the accredited boundary.
Retrieval-augmented generation over curated authoritative knowledge bases, with citation and source attribution on every output.
Task-specialised agents for analysis, briefings, submissions and the organisation and searching of large data sets.
Zara LM is a sovereign AI platform built for the Australian Government, deployed on the latest models and technology, secure by design, and trusted to operate at Secret.
Australian-built and operated on Australian Government controlled infrastructure, with no reliance on foreign-hosted AI services.
Deployed on current-generation AI models and a modern, maintained technology stack that keeps pace with the frontier.
Security built in from the start, with no prompt, document or telemetry leaving the accredited environment.
Hosted entirely on-premises in a secure environment, on cloud, or in an air-gapped environment, giving flexibility and security to match the classification.
Trusted to operate at Secret within a secure, accredited boundary.
Trusted by the Australian Government for sensitive, consequential work where assurance and provenance matter.
Automates the preparation of briefings, submissions, committee material and other high-volume administrative products through configurable, template-driven agents.
Organises and searches large data sets, surfacing relevant material and grounding answers in curated source documents.
A modular node architecture that scales to over 200,000 users, from a single production node to a multi-node fleet.
Every output is verified and approved by an accountable officer before it is used: AI accelerates, humans decide.
The return on a sovereign AI capability comes from how it is adopted, not from the model alone. Zara LM is priced up-front on infrastructure Government already owns, and delivered alongside the workflow redesign that turns capability into measurable return.
Government owns the hardware, so there are no usage-based charges. Cost is fixed and known up front: run the infrastructure you already hold to its full capability, with no metering, no per-query billing and no surprises as adoption grows.
The largest gains come from redesigning high-volume processes around the capability. Phoenix ICT works with your teams to identify the workflows worth targeting, build configurable agents for them, and measure the time and effort returned.

Zara LM is trusted to operate at Secret. All processing remains inside the accredited boundary.
Sovereign AI only delivers value when the workforce knows how to use it well. Zara Assisted is our AI literacy and training service for executives, leaders and the APS workforce, delivered by practitioners who build and operate Zara LM every day.
We combine foundational literacy, secure-use training, role-based curricula and a workforce review that aligns teams to the new ways of working AI unlocks.
Talk to us about training
How large language models work, where they fail, and the principles for safe, effective use. A shared vocabulary and a shared floor of competence across your organisation.
Working with sensitive and classified information on sovereign platforms. Accreditation boundaries, data handling, prompt hygiene and the accountable officer model.
Board and SES-level sessions on AI strategy, risk, governance and the decisions leaders need to make about adoption, assurance and workforce impact.
Tailored curricula for executives, delivery teams, analysts and technical roles, each mapped to the daily work AI will change first.
Practical, hands-on skills: prompting patterns, tool use, agent design and how to evaluate and trust AI-assisted output before it's actioned.
A structured review of roles, processes and team designs to align your workforce to the new ways of working AI enables: where the gains are, and how to organise for them.
Talk to us about delivering a secure, accredited AI service, and uplifting the people who will use it, into your environment.
Talk to our team