Modernisation is a trust exercise: Andy Tripathi on information foundations, AI, and what comes next
In conversation with Andy Tripathi, Chief Information Officer, Australia's Digital Transformation Agency. A practitioner-led discussion on responsible modernisation, information stewardship, AI, and the long-horizon risks government can't afford to defer.
Some of the most important lessons in digital government come from the worst nights — when the system you've built is the only thing standing between communities and a crisis they didn't ask for.
Andy Tripathi returns to one of those nights early in our conversation. The Black Summer bushfires. He was leading the digital services team. The main website — which had become the single source of truth for evacuation centres, road closures, and emergency alerts — went down under an unprecedented surge in traffic. People were waiting on those updates to make life-and-death decisions.
What stopped that becoming a public failure, Andy tells me, was preparation nobody outside the team would normally see. "We had 'back of the backups' — redundant platforms and alternative communication channels — ready to activate." The community kept getting reliable information. Trust held.
That story sets the frame for everything else he says. Modernisation in government, he argues, is "not just a technical exercise — it's a trust exercise." And the test of it isn't what gets deployed. It's what holds together when it matters.
This conversation, more than most, is about the unglamorous foundations beneath the AI conversation: information governance, citizen outcomes, trust, and the discipline of sequencing. Andy is unusually clear-eyed about all four — and about what CIOs should be doing on Monday.
Citizen outcomes are the only test that matters
When Andy describes ICT reform, he doesn't lead with platforms or programs. He leads with what should change for the people on the receiving end.
His outcomes list is short and unfussy. Less effort for citizens and businesses. Consistent and reliable service delivery. Clarity in interactions and decisions. Fair and inclusive access. Timely outcomes aligned to real-world needs. If a reform doesn't shift those, he says, it risks becoming "internally focused" — a polite way of saying expensive and beside the point.
What stays with you is the image he uses for getting this wrong. "It's just like an architect who designs a house without ever speaking to the people who will actually live in it. The result can be a system that looks impressive on paper but doesn't meet real needs."
The land planning modernisation he led is the antidote. The team didn't start with the system. They started with the community — information sessions, public pop-ups, surveys, real conversations about real frustrations. From there, they built personas that reflected the actual diversity of need, then designed around them.
This is what human-centred design looks like in delivery, not theatre. And it's why Human Experience sits as one of the three pillars of the FGX Framework, rather than as a chapter at the back of a strategy document.
The information test: AI amplifies what's already there
The strongest line in the conversation is also the simplest.
AI does not fix poor information management — it amplifies it.
Andy is precise about the non-negotiables that decide whether AI in government scales benefit or scales risk:
Clear accountability for information assets
Defined authoritative data sources for key decisions
Known and managed data quality
Strong lifecycle governance, including disposal
Traceability to support transparency and review
Without these, AI becomes a high-speed pipeline for bad decisions in areas where mistakes affect entitlements, compliance, or public confidence. With them, it becomes "a powerful enabler of better, faster, and more consistent outcomes."
The records management reform he's led recently doesn't try to do everything. It zeroes in on three things and stitches them together: comprehensive training, implementation of Information Asset Registers, and the integration of privacy, cyber security, and information management into a single governance posture. Most agencies treat those as separate domains run by different teams. Andy treats them as one problem.
He's also blunt about the moment we're in. "There's a distinct 'before AI' and 'after AI'." That language matters. It signals that AI adoption isn't optional, but it also signals that the discipline behind it has to be deliberate, ethical, and grounded in what the organisation actually knows about its own information.
Strategy as a relay race, not a sprint
Government leaders are routinely asked to deliver cyber uplift, AI adoption, and legacy modernisation at once — often against legacy systems and constrained funding. Andy's frame for staying coherent is unusually practical.
He talks about defining a small number of clear, outcome-driven objectives, integrating cyber, data, and digital into a single delivery strategy, naming trade-offs explicitly, and building feedback loops between policy, delivery, and operations. Then he uses a metaphor that lands.
"Developing and executing strategy should be viewed as orchestrating a relay race rather than an individual sprint. If each participant focuses solely on their segment without considering handoffs, collective success is jeopardised despite individual proficiency."
The implication is that digital strategies don't fail because any one team underperforms. They fail at the handoff. Between policy and delivery. Between cyber and AI. Between strategy and operations. Coordinated practice on the transitions is what keeps the baton moving.
This is where Government 3.0 thinking shows up most concretely. Connected, citizen-centred government is, in operational terms, a series of well-engineered handoffs.
What to do on Monday
Some of the best advice in the interview is the simplest. Andy gets very specific about what a CIO or deputy should do in the first 90 days to lift information governance.
Focus on the most critical information assets. Assign clear ownership. Understand data flows and duplication. Reduce redundant and obsolete information. Make the Information Assets Register usable — not a document that lives in a folder, but a working artefact connected to lifecycle decisions. Align with security and privacy controls.
The shift, in his words, is "from static documentation to active information management."
Then there's the metaphor that earns its keep.
"Think of it like keeping an organised garage. If you just keep dumping things in without a proper inventory, when something goes missing during a break-in, you won't know what was stolen. Having a clear register and lifecycle process means you always know what you have, where it sits, and how it's being used."
It's a small line. But it's exactly the kind of plain language that gets executives across a room agreeing on the next move. The Information Assets Register stops being a compliance artefact and starts being an operating tool — embedded in the authority to operate process, so every new system entering the organisation is logged before it goes live.
Trust is engineered, not communicated
The FGI position is that trust is something you build into the design, not something you patch on at the end. Andy agrees, and goes further.
"Security is a safeguard. But trust is something deeper — citizens feeling confident that they understand how their data is handled, that they can relate to the people and processes behind government services, and that their needs are genuinely considered." He names the difference clearly. "It's the difference between locking the doors and knowing exactly who's inside and what they're doing."
That's SPRITE — Security, Privacy, Resilience, Inclusion, Transparency, Ethics — translated into a single sentence. Security is necessary. It isn't sufficient. The gaps Andy sees most often are in the other five: how decisions are made, who's accountable when things go wrong, whether services behave consistently across channels, whether design genuinely includes everyone, and whether emerging technologies are being used ethically.
Citizens trust services that are predictable, understandable, and fair. That's an engineering job, not a communications job.
The horizon: AI now, post-quantum next
The interview includes one of the more unfashionable but important warnings in cyber strategy right now. While AI dominates the boardroom, post-quantum cryptography is being underinvested. The risk is the "harvest now, decrypt later" scenario — adversaries collecting encrypted data today, knowing they can break it later. Particularly damaging for governments that retain sensitive records for decades.
Andy's response is to design for crypto-agility now, prioritise protection for high-value, long-life data, and integrate post-quantum considerations into current investment decisions. Not as a future project. As a constraint on today's choices.
"The challenge is not choosing between present and future priorities, but recognising where delay creates irreversible exposure." That sentence belongs above the desk of every Chief Information Security Officer in government.
Capability beats technology
Closing the conversation, Andy is clear about what will differentiate high-performing digital government teams over the next decade. Not specific platforms. Capability.
Strong information stewardship and governance. Broad AI literacy across the workforce. Flexible and adaptive technology architecture. Integrated approaches to risk — cyber, privacy, and ethics held together. Mature human-centred design practices.
He's just as clear about what leaders should stop doing. Over-investing in bespoke solutions where shared ones are viable. Measuring success through activity rather than outcomes. Treating information as a by-product rather than a strategic asset. Deferring difficult decisions on legacy systems and risk.
That last list is, in many ways, the harder one. It's where political courage shows up.
On collaboration, he's pragmatic about the limits of one-size-fits-all. "Something that works well for one agency may not suit another's business requirements." The shift he wants is towards "build once, use many times" — common standards, shared patterns, reusable capabilities, and aligned incentives — supported by governance that makes collaboration the easier path.
The patterns are consistent across his answers. Modernisation is judged by what it delivers under stress, not by what it deploys on a good day. Citizen outcomes are the only credible test. Information governance is the foundation AI either rewards or exposes. Trust is engineered into the design, not added at the end. And the long-horizon risks — post-quantum, legacy debt, capability gaps — get harder and more expensive the longer they're left.
Connected, citizen-centred, capable of continuous improvement. That's the Government 3.0 frame Andy is working inside, and it's already the foundation for the Gov 4.0 horizon — agentic, AI-native, intelligent government — that the next decade will demand.
He puts the bar in one line near the start of the conversation, and it's the right place to leave it.
Modernisation is not judged by what we deploy, but by how consistently we deliver when it matters most.




Comments