Who we help
We help agencies adopt AI and prove that it works.
You are under pressure to use AI responsibly and to show it is delivering something.
What we bring
You get the delivery discipline of our commercial work.
Public sector AI has a habit of producing frameworks and pilots while nothing reaches a user, because nobody is held to a date. Our commercial clients hold us to a named owner, a measurable outcome and something running in production, and we bring that into the APS alongside the governance fluency you would expect from people who have worked inside it.
Our people have worked inside Australian government agencies and built the strategies, frameworks and products now in use across the APS. At the Department of Home Affairs we wrote the AI strategy and the Data and AI Governance Framework, which the Secretary endorsed and which names the Chief Data and Analytics Officer as the Accountable AI Official.
Knowing the obligations is only half of it, because the other half is making sure something gets delivered. The strategy has to become a roadmap and the roadmap has to become a system people use, and that is where most agency AI programmes stall.
The pressures
These are the pressures we hear about most often.
You have a mandate to use AI and a duty to prove it is safe
Those two pull against each other unless the strategy resolves it, so we write governance that is proportionate to risk, which lets you say yes more often.
Everything you publish gets read by someone looking for problems
Legal, cyber, privacy, audit, the ANAO, the Ombudsman and Senate Estimates all read your work, so we write for those readers from the first draft.
Your pilots keep stopping short of production
Almost every agency has run a pilot and far fewer have anything in daily use, and the blockers are rarely about model quality.
Capability has to stay when people move on
Staff rotate and contracts end, so capability that walks out with a vendor leaves you with a governance problem on top of a delivery one.
How we help
Here is how we work with government clients.
We write strategy against the policy you answer to
Everything maps to the APS AI Plan, DTA guidance and the whole-of-government AI Policy, so your strategy can be read directly against the frameworks you are measured on.
We build the governance instruments as well as the principles
That means AI Impact Assessment pathways, an AI register, risk tiers and support for your Accountable AI Official, which are the artefacts audit asks to see.
We deliver in twelve weeks inside your own environment
Work runs in your tenancy and under your governance, with a named business owner and an agreed path into business as usual.
We lift the capability of your workforce and your executive
All-staff sessions, SES literacy programmes and role-based curriculum leave the knowledge with your people once we have gone.
Services
Most engagements start in one of these four places.
We have ordered them by where government clients usually begin.
Evidence
This is work we have done in this sector.
Building AI capability at two scales: the enterprise and the community
Policy does not make people AI-capable. We pair whole-of-agency sessions that reach everyone with a small, open Canberra practitioner community co-founded with an Australian university, because confidence is built by trying things rather than by reading guidance.
- One agency heard the same message at once
- Whole-of-staffOne agency heard the same message at once
- The partnership keeps sessions evidence-based
- UniversityThe partnership keeps sessions evidence-based
- AI in the Shadows meets regularly in Canberra
- ~30 peopleAI in the Shadows meets regularly in Canberra
A whole-of-department AI Strategy, built on four pillars
A department operating at national scale needed an AI strategy that could satisfy its executive, its assurance functions and the public interest at once. We delivered a four-pillar strategy covering 2026 to 2029, linked end to end from mission risk through to costed milestones with named owners.
- We advanced all four pillars in parallel
- 4 pillarsWe advanced all four pillars in parallel
- We scoped, costed and assigned each milestone
- 18We scoped, costed and assigned each milestone
- The roadmap sequences the investment
- 2026 to 2029The roadmap sequences the investment
Contact
Tell us what you are trying to achieve.
We will tell you whether AI is the right tool for it and what the work would realistically take.