March: We’re officially famous, but also humbled

Could 2026 just chill out already? It’s not just us right? But my goodness, it’s been a flying start to the year. Literally for some Friday team members.

A certain co-founder (Emma) has finally returned back from her trip to Singapore and Australia. She also attended Sydney’s first AI agents club, and in a total plot twist, she loved it. John, and his very friendly pals, are hosting a second event next week. If you’re based in Sydney and have an interest in AI agents, go along!

At Friday we (aka Lauren and Max) are building our very own AI Agent Army. It turns out the hardest part is mass orchestration and getting AI to be deterministic. So if anyone is keen to deep dive into early 2000s CSIRO computer lore with us, shout!

We’ve also had the pleasure of being featured in Entrepreneur UK magazine’s Top 100 Start-up to watch list this month. The life of a small business owner is tough, payroll, clients, conflicting demands, building amazing technology solutions, you name it. So we hate to say it, but it’s really nice receiving public accolades every now and then. If you know another founder, consider this a reminder from us to give them a shout out, free coffee or introduction to a useful person whenever you can, because they, like us, sometimes need the win!

In less positive news, this month our new BFF Claude, went rogue and completely deleted everything off Lauren’s laptop. We’re both amused by the irony of a data governance company having AI go totally rogue but also highly cognisant that if it’s happened to us, it’ll definitely happen to you. We love Claude, but it was a good reminder that guardrails are essential (and sometimes even those don’t work so backup, backup, backup - and then keep that separate so Claude doesn’t use your backups to overwrite your newest versions).

Lots of news this month, read on below!

Olive on the loose

Australian supermarket Woolworths has had to tweak its customer service chatbot after users reported some very odd small talk. The virtual assistant, Olive, was telling customers stories about its mother and uncle while asking for routine details like birthdates and postcodes, including references to its mother’s angry voice.

Customers were understandably confused. One user said the bot began bonding with them over shared birth years before asking for their postcode; another said it made fake typing noises while pretending to look things up.

The explanation is surprisingly simple: years ago, a staff member scripted the chatbot to reference family members as a way to give it a quirky personality. In practice, it mostly confused customers, especially when the bot also gave incorrect product prices. It runs out that adding personality to AI is easy; making sure it is grounded in reliable data is much harder.

Woolworths has now removed the family backstory as it upgrades Olive into a more advanced AI shopping assistant. No notes, OBSESSED. #SAVEOLIVE

Consulting will never die

A new Wall Street Journal report highlights a surprising development: rather than replacing consultants, AI is currently increasing demand for them.

OpenAI and Anthropic are partnering with major consulting firms including McKinsey, BCG, Accenture, Capgemini and Deloitte to help businesses actually deploy AI in meaningful ways. Despite the hype, most organisations still haven’t scaled AI across their operations and say AI has yet to deliver significant financial returns.

The issue isn’t the technology, it’s implementation. Companies need help redesigning workflows, integrating AI into existing systems, and identifying real business use cases. That’s where consultants come in.

Who’d have thought that while AI may automate analysis, organisations still need people who understand business, governance and operations to translate capability into real value. Two years ago, the same paper was writing about how consulting is dead - ironically, that’s when we decided to start a consulting business. Our take, hire consultants, don’t hire consultants, whatever you pick, start with good data (and if you do hire consultants come talk to us first).

Coding is out, data is in

Get a coffee and settle in for a long read. A recent feature in the New York Times highlights how AI coding agents are transforming software development. Many engineers now spend less time writing code and more time describing what they want systems to do, while AI generates and tests the code itself.

But the more interesting question isn’t about coding. It’s about what sits behind it. For startups building new systems from scratch, AI can accelerate development dramatically. But for large enterprises, most work happens in complex ‘brownfield’ environments with decades-old codebases, tangled dependencies, and enormous volumes of data. In these settings, the real challenge isn’t producing code quickly, it’s understanding how existing systems and data interact.

This is where AI’s limits become clearer. Generating code is relatively easy (in fact, too easy some might say given how many scripts are on our local drives); ensuring it works safely inside complex organisational data environments is far harder. Enterprise systems depend on accurate data models, governance, version control and careful integration across legacy systems. Without a clear understanding of that underlying structure, faster code simply creates faster problems.

In other words, as AI makes coding easier, the real bottleneck shifts upstream. The critical capability becomes not writing software but understanding the data, systems and relationships that software operates on. After all, software is only as good as the data ecosystem it sits on.

Heated Rivalry

The competition between OpenAI and Anthropic is heating up, and it’s becoming increasingly personal. Anthropic, founded by former OpenAI researcher Dario Amodei, is rapidly gaining ground, with revenue expected to reach $19 billion this year, strong enterprise adoption, and a reputation for ethical AI.

The rivalry recently escalated over a Pentagon contract. Anthropic refused terms that could allow its AI to be used for autonomous weapons or domestic surveillance, losing the deal. OpenAI stepped in to secure its own agreement, triggering backlash from some tech workers. Claude has been receiving lots of support from enterprise customers in the aftermath, reaching number one download status on the Apple Store.

We think this is remarkable for so many reasons. Firstly, as mentioned above, Claude (Anthropic’s baby) went rogue on us recently. Can you imagine what it would be like with nuclear codes? What baffles us more is that instead of being congratulated for pointing out weaknesses in their solution, they’re being punished. The world according to the US government is belonging more and more to total grifters. Bizarre. If you want a more macro look at the whole issue, this piece by the Economist is excellent.

(E)U ready for this?

In positive news according to the Economist, Europe still lags far behind the U.S. in producing tech giants, but there are signs the gap is narrowing. Venture investment in European startups has grown sharply over the past decade, more established founders are backing the next generation, and policymakers are under growing pressure to strengthen the region’s tech base.

Part of that shift is geopolitical. Donald Trump’s hostility towards Europe, combined with changes in the U.S. and China, is making Europe relatively more attractive to both investors and talent. The piece argues this is especially visible in climate tech, defence tech and deep tech, where European startups are starting to build real momentum.

It is not a claim that Europe will overtake America any time soon. But the mood is changing: Europe is becoming a more credible place to build ambitious tech companies than it was even a few years ago.

We’re talking to founders every day, and the vibrancy in the UK and Europe is incredible. We’re lucky to be a part of the cohort.

AI Fail, Again, lol

AWS recently suffered a 13-hour outage after engineers allowed one of its AI coding agents to make changes to a system environment. The tool reportedly decided the best fix was to “delete and recreate the environment”, which caused the disruption.

Amazon says the problem was user permissions rather than AI autonomy, but the incident, the second involving its internal AI tools, highlights the risks of letting AI agents take actions in production systems.

So our take, and experience (see above) is that while coding tools can speed up development, they can also break things very quickly. Use with smart guardrails (but also be prepared for the worst).

Ads Are En Route

OpenAI has started testing advertising inside ChatGPT, marking a major shift for the company. Ads will appear below responses for some free users, with advertisers paying high minimum spends to participate.

The move reflects the growing pressure on AI companies to generate revenue as the cost of building and running AI infrastructure soars. If successful, it could open a huge new channel for digital advertising as people increasingly search, shop and plan through conversational AI.

Just as search engines became ad platforms, AI chatbots may be the next big battleground for digital advertising.

Counting Sheep

Australian abattoirs are using AI-powered cameras to automatically count sheep as they arrive for processing, replacing a manual system that has caused disputes between farmers, truck drivers and processors for decades.

The technology captures images as animals move through the facility and converts them into structured data. Verifying numbers, identifying sick livestock and checking that the right cuts are labelled and packaged correctly.

We love this use of data and AI. Reminding us all that the real value comes when messy real-world processes become an actual solution. Even if it is for sheep.

Watercooler Chat

A section of the things we like that keep us sane while running a small business…

• This interview in the NYT with Bobby Carnevale — He talks about life with his first pet, a bearded dragon.

• Essential Antwerp — Emma is very tempted by this dress specifically.

• If you, like us, love to know what people get paid, read this from New York Magazine.