Not on the Panel
Unpacking what works when building in ███ ██████, GovTech, and the █████████ market.
👋 Welcome
Welcome to our new readers for this week! Although it may be a new financial year, in government, we’re not quite there yet. With 36 days left in the reporting window, there’s still a steady stream of 2024–25 contract data rolling in. The big EOFY deals are only just starting to surface, and this week’s numbers reflect that.
While the rest of the tech world chases consumers, the real story lies in the power, scale, and opportunities hidden within the $2.3 trillion global government market.
Our thesis is built on a radical but obvious insight:
“It’s ridiculous that almost no venture-backed companies work closely with government, given how many billions it spends.” - Trae Stephens, co-founder Anduril.
Not on the Panel decodes the jargon, tracks the deals, and provides the map to Australia’s $80B+ B2G sector. We’re not just writing about it; our GovTech & B2G angel syndicate is putting capital behind it.
In this week’s edition:
• We find another Australian startup, quietly winning in the B2G market (we really need to update our map)
• Sixty million may not sound trim, but in government records management, it is business as usual.
And in our main feature:
• We take a look at OpenAI’s beachhead into the Australian GovTech market. In my view, it reads like a hallucinated AI mistake — the kind with an em dash addiction and no understanding of how government actually works.
🤝 If you value our unique perspective, help us grow by sharing it with your network. Every new subscriber brings us closer to delivering more in-depth coverage of the overlooked and undervalued B2G market.
📰 This Week in Global GovTech
🇦🇺 Australia
💰 Australian Gov Backs Deep Tech with $178M
The Albanese Government's Australia's Economic Accelerator (AEA) Innovate program awarded over $178 million to 39 research projects. Part of the $1.6 billion AEA, these investments in areas like hydrogen reactors and underground Wi-Fi signal significant future B2G opportunities.
Source: aea.gov.au
📉 The Tech Sector Shrinks, but OpenAI Thinks It Can Save the Day!
A new report from the ACS reveals that Australia’s tech sector shrank by 2.6% in 2023-24, the first decline in nearly two decades. Experts call this a "warning signal" for Australia to build its digital economy. OpenAI also released its "economic blueprint" for Australia, suggesting the use of AI for modernising government services. In other news, water is, in fact, wet. 👇See our feature article below👇
Source: ACS
🤝 Australia, India Partner on Undersea Surveillance
The Department of Defence announced a new science and technology project arrangement with India. This three-year joint research initiative focuses on enhancing undersea surveillance for early detection and tracking of submarines and autonomous vehicles, signalling future high-tech defence procurement.
Source: defence.gov.au
💡 Registrations Open for AI in Government Showcase
The Digital Transformation Agency announced registrations for the AI Government Showcase: Adoption in Action (July 31, 2025). The event will feature live demonstrations and booths from innovative small-to-medium AI vendors, offering direct engagement for GovTech startups in AI.
Source: dta.gov.au
📈 APS Data & Digital Foundations Plan Updated
The DTA published an update on the "Investing in the APS' data and digital foundations" mission. This highlights ongoing efforts to build a strong data and digital foundation within the Australian Public Service, investment in capability, talent, and fostering industry collaboration, which indicates an emphasis on the need for external expertise.
Source: dta.gov.au
🌐 Global
🇧🇹 Bhutan's Hackathon Fuels Decentralised ID Solutions
GovTech Bhutan's hackathon (June 25-27, 2025) yielded 13 decentralised ID solutions. Winning teams Cyberchain and DeepGov secured BTN₹250,000 (~ US$3,200) each. All 13 teams received BTN₹500,000 (~ US$5,800) in incubation funding from the Royal Government.
🇬🇧 UK Kicks Off "Digital Transformation in Public Sector Week"
TechUK launched its "Digital Transformation in the Public Sector Week 2025" (June 30, 2025), themed on technology reshaping UK public services. It highlights AI, resilient digital infrastructure, and data unlocking, indicating active demand for government technology.
🇺🇸 Castelion Secures $350M for Hypersonic Missiles
Castelion, a defence tech firm, secured $350 million to mass-produce hypersonic missiles. This significant Series B raise highlights major investor confidence in the defence B2G sector.
🇳🇬 Nigeria Promotes B2B & B2G Platform for AfCFTA
Nigeria is accelerating the implementation of the AfCFTA, with a focus on developing a B2B and B2G matchmaking platform for the upcoming Intra-African Trade Fair. This aims to foster trade and investment between businesses and governments across the African continent.
💼 This Week in Australian B2G
(Week Ending 4 July)
📄 $2.235b in total reported contract spend
🖥️ 236.98m in software + digital services + I.T
Breakdown by Service:
💾 Software: $140.39m
☁️ SaaS (Cloud): $1.20m
🛠️ Software maintenance & support: $17.42m
🧱 Platform SaaS: $289k
⚙️ Software/hardware engineering: $58k
💻 I.T $177.61m
Breakdown by Procurement Method
📢 Open Tender: $68.92m
🗃️ Prequalified Tender: $0
📩 Limited Tender: $168.05m
GovTech Flex (Our selection of last week’s contract announcements)
NQRY, an Adelaide-based company that builds investigative software for law enforcement and intelligence, has secured its third AFP contract. The new $3.58 million deal brings total government revenue to $6.34 million.
iCognition may not catch the eye of most people, and a $413,000 contract rarely makes headlines. Yet this Canberra-based company has quietly accumulated almost $60 million in government revenue since 2006, all from managing the one piece of software more than 100k public servants cannot function without: OpenText Content Manager, also known as TRIM.
🧱Build Better
A Manifesto for a Government That Doesn’t Exist
OpenAI’s new blueprint, released this week, promises that artificial intelligence will revolutionise the public sector. But the public sector is not waiting for revolution. It is managing complexity. And OpenAI has misunderstood the job.
The Digital Déjà Vu
Every few years, a new piece of technology arrives with the promise that it will finally solve the government's problems. In the 1990s, it was Web 1.0. Interactive webpages, online forms and downloadable PDFs were going to reduce call volumes and eliminate paperwork. They did not. A decade later, Web 2.0 brought logins, real-time chat and transactional portals. The promise returned. Faster services. Better access. Greater transparency. Again, much of it remained unrealised.
Then came the cloud. Then came chatbots. Then came service design. Each time, the technology improved, but the core experience of government changed very little.
Now we are told artificial intelligence will succeed where all else has failed. According to OpenAI’s Australia Economic Blueprint, AI can improve productivity, reduce wait times, streamline correspondence and modernise the public sector. It is a seductive idea. But it makes the same mistake as every digital reform pitch that came before it.
It treats the government as a machine that needs to operate at a faster pace. What it actually needs is better judgment.
A Real Pain in the APS
OpenAI writes:
“AI can help public servants better serve constituents by summarising information, drafting responses and analysing data.”
This line appears thoughtful and even helpful. But it reveals how far removed this vision is from the actual rhythm and reality of public service work.
Take ministerials, the formal letters written in response to public inquiries, complaints or campaign emails. To an outsider, they appear to be a perfect use case for AI. But to anyone who has worked inside government, they are something else entirely.
These letters are not a paperwork problem. They are a political risk. One poorly chosen phrase can land an official in Senate Estimates or generate headlines that derail an entire policy agenda. Ministerials are slow, not because people cannot write. They are slow because people are afraid of what happens when language is misread or misused.
Writing them is, quite honestly, a pain in the APS. And yes, that is precisely what it sounds like. Not just a bureaucratic frustration, but a persistent institutional ache. A kind of muscular tension caused by a policy that is unclear, rigid or poorly communicated. The real issue is not the letter. It is what caused the letter to be written.
When citizens are confused or feel mistreated, they write in. Sometimes one at a time. Sometimes in orchestrated waves. The correspondence does not arrive because the website is broken; it comes because something more profound is. AI will not change that. It will only help public servants respond more quickly to the consequences of failure.
That is not transformation, it is containment.
The Illusion of Speed
At one point, the blueprint states:
“Governments should act swiftly to harness AI while ensuring public trust through transparency and accountability.”
This is precisely the kind of sentence that sounds reasonable in a keynote but is completely unworkable within government.
Governments do not act swiftly for good reason. The APS is not optimised for speed. It is optimised for caution. That is not a flaw. It is an institutional memory. Ask anyone who lived through Robodebt, the pink batts rollout, or any other large-scale policy failure that damaged trust, harmed lives, and forced ministers to apologise on camera.
Yet speed is not impossible. The JobKeeper program was designed and delivered quickly. Emergency payments during bushfires and COVID lockdowns were processed within days. The difference is familiarity. The APS knows how to define payments and how to distribute money. That is its bread and butter. These are well-understood processes with relatively stable risk profiles.
AI is none of those things. It is untested. It is probabilistic. It struggles to explain itself in plain language. It may be helpful in many contexts, but it is not a candidate for rapid deployment inside a system where accountability cannot be deferred to a black box.
To tell the government to move fast and preserve trust is to misunderstand both speed and trust. Trust is not gained through urgency. It is earned through consistency. And preserved through reversibility.
Where AI Might Belong
There is a better role for AI in government. But it is not in ministerials. It is not in summaries. And it is not in creating answers to questions triggered by broken policies.
The value of AI is not in saving time. It is in improving judgment.
The valid constraint in the APS is not communications capacity. It is analytical overload. Policy officers must work across dozens of datasets, legislative frameworks, and program constraints, often while responding to shifting political direction with unclear timelines and intent. The cognitive burden is real.
This is where AI could help. Not by replacing decision-makers, but by extending their sightlines. Surfacing hidden trends. Revealing the long tail of outliers that might otherwise be ignored. Simulating the downstream impact of a rule before it becomes an entitlement. Interrogating the unstructured data that never fits neatly into a table or a dashboard.
In other words, AI belongs upstream. It belongs at the point where complexity enters the room and before a single word of policy is drafted.
The Flywheel of Failure
To deploy AI after the fact to automate responses to citizens harmed or confused by poor policy provides absolutely no public value. Faster replies to a growing pile of complaints do not signal progress. They signal entropy. The machine runs, but nothing improves.
And so the real question is not whether AI can help the government work faster. The question is whether AI can help the government think more clearly. Not whether it can respond to policy failures more efficiently, but whether it can help reduce failure in the first place.
OpenAI’s blueprint describes a government that acts boldly, swiftly and with complete public trust. That government does not exist. And perhaps it should not. What exists is a system that carries the weight of memory. That moves slowly for reasons we forget at our peril. That values clarity over novelty. And that measures risk not by how good an idea looks on a whiteboard, but how explainable it is on the record.
AI will not transform the government through speed. It might, if applied carefully, help it make better decisions. And that is a far more ambitious goal.
🗺️ Our Government Market Map
For founders, the journey begins with a simple truth: the government can buy from you.
To prove it, we've built Part 1 of our market map: a live logo board showing the Aussie startups, by category, that are currently engaged in government contracts. Not all are landing huge deals; some are just getting started. But it’s the definitive proof that a pathway exists.
We're already working on successive layers of intelligence, such as adding contract values so you can see the scale. We are also still building a sister map of the overseas competitors you're really up against.
This market deserves sharper focus, better data, and real momentum. We are here for all of it.
📣 Are you a Startup Selling to the Government, or want to?
Each week, over 1,500 federal contracts are published on AusTender. We track what we can, but we cannot review every supplier to identify which companies are startups.
If you’re:
Tendering for government work
Already delivering to a government customer
Recently awarded a contract
We would like to hear from you.
You don't need a press release; leave a comment below or get in touch with us.
We’re always looking for stories that help explain what is actually happening in the B2G and GovTech markets.
📄 Method and Scope
AusTender publishes thousands of contract notices each week across a broad range of categories. While we rely on this data as our primary source, we cannot guarantee the accuracy or timeliness of individual listings.
We do not aim to cover every sector or supplier. However, if a startup is awarded a contract and we can verify it, we will include it, regardless of the classification. These stories help highlight where momentum exists and where future opportunities may lie.
✍️Meet the Editor
Hi, I'm Mat, a Startup advisor, former bureaucrat, investor, and lifelong procurement tragic.
Throughout my career, I’ve worked on four of the Commonwealth's most significant non-defence contracts. I remain frustrated that early-stage companies are often excluded from the government market.
This Substack is part of how I’m building in public. I work with founders and investors who see the $80+ billion business-to-government opportunity in Australia.
We also support founders in essential but often overlooked areas, such as governance, risk, and strategy.
If you’re a founder looking to break into government or seeking opportunities to back generational companies in this space, please don't hesitate to reach out. I’m always up for a coffee.