Practical insights from the trenches of building and scaling technology businesses.
The technical companion to A Few Good Logs: which regulations apply to AI in regulated sectors, what evidence each one expects, and the engineering controls that produce it.
Responsible AI states what an organisation intends. Defensible AI is what it can prove when a regulator, a court or an aggrieved customer comes calling, and most organisations have yet to build it.
Most growth problems are targeting problems wearing a marketing costume. If your buyer has to be taught that your category exists before they can want it, you have taken on a cost that most startups cannot afford.
When software was expensive, the build was the filter. Now that it is cheap, the homework founders skip has become the most expensive thing they can skip, and founder/market fit matters more than product/market fit ever did.
AI has collapsed the cost of building software, but the barrier it has not touched is access to networks. For Black and ethnic minority founders, that makes the remaining gap sharper, not softer.
Most of the investment industry is using AI to build faster horses, not to reinvent the customer relationship. A talk at the TISA AI Conference on why the next two years will separate the reinventors from the rest.
AI coding agents can build in hours what used to take days. But what used to take hours to debug can now take days. The relationship between human and AI is more Batman and Robin than most people think.
What happens when you try to build a production-quality, privacy-first financial assistant using open-weight LLMs and AI-powered development tools? Six months of lessons from a recent client prototype.
In an increasingly fractured geopolitical landscape, whose AI is the United Kingdom actually running on? And what happens when access to it is no longer guaranteed?
Investment firms fear AI will replace human financial advisors. The real question is whether AI might actually grow the market for human advice, not shrink it.
The traits that make founders successful in the early days are rarely the traits needed to scale. Honest self-reflection, not more capital, is what separates founders who grow from those who get in their own way.
Despite America's dominance, a perfect storm of constraints threatens to reshape the global data centre landscape, with Latin America emerging as a formidable competitor.
Democratisation of AI access is positioned as inherently positive. But what happens when everyone becomes an amateur expert, and powerful tools are available without guardrails?
When adopting the use of GenAI, think about the long-term consequences of automating the easy work.
When demoing an AI-powered application, also remember to use private data and complex supply chain advantages to demonstrate a unique value proposition.
A framework of four key questions - where, how, what, and why - to guide your organisation's generative AI deployment decisions.
Organisations rush to identify AI use cases, but the critical first step is understanding genuine business pain points before reaching for the technology.
Drawing parallels with the 1990s web, generative AI is still in its engineering phase. Building products on unstable foundations carries real risks.
The OpenAI board drama is a stark reminder that misalignment between founders and investors causes more failures than lack of product-market fit.
A comprehensive guide to international expansion for technology companies - from motivation and market research to GTM strategy, hiring, and visa considerations.
Why being 'known' matters more than being 'great' - lessons from three startups that had superior products but lost to better-known competitors.
Drawing parallels with the web's evolution, 5G remains in its engineering phase. The real innovation will come when it transitions to its design era.
