Back to Home
Writing

Articles & Thinking

Writing about technology since 2015. System architecture, AI, data engineering, cloud, and everything in between. Search, filter by year, or browse by topic.

Filter by year

Filter by topic

Showing 46 of 46 articles

Tech Industry·6 min read·4 April 2026

The Tariff Shock Is Here. What Does It Actually Mean for Tech?

Tech stocks are in freefall. Supply chains are being repriced overnight. Everyone is talking about tariffs. But underneath the noise, there are some quieter questions worth sitting with.

Tech IndustryOpinionGlobal TechAIEconomy
AI·6 min read·3 April 2026

Everyone Is Building AI Agents. But Is Anyone Actually Using Them?

AI agents are the most talked-about thing in tech right now. Every company is announcing one. But when I look at what people are actually doing with them day to day, the picture is much quieter than the announcements suggest.

AIAgentsOpinionTech IndustryFuture of Work
Future of Tech·7 min read·2 April 2026

The End of IT Dependency: How AI Is Turning Everyone Into a Builder

Ten years ago, businesses called an IT vendor when something needed building. Today, an individual with a laptop and an AI tool can do what an entire outsourcing team once could. The dependency chain is collapsing. Most people have not noticed yet.

AIOpinionTech IndustryFuture of WorkAutomation
AI Security·9 min read·1 April 2026

The Claude Leak of 2026: A Post-Mortem on AI Secrecy and the Dawn of the Autonomous Wild West

The March 31st exposure of Anthropic's internal architecture is not merely a corporate data breach. It is the definitive collapse of the "Security through Obscurity" model that has governed Silicon Valley for the last decade.

AISecurityOpinionAnthropicCybersecurity
AI Tools·8 min read·31 March 2026

I Spent a Week With Perplexity Computer. Here Is What No One Is Telling You.

Perplexity Computer launched in February 2026 at $200 a month. I used it every day for a week across research, writing, and code tasks. Here is what the reviews are not telling you.

AIToolsReviewPerplexityAgents
AI and Healthcare·7 min read·30 March 2026

AI in Healthcare: Not Just Hype, But Something We Need to Get Right

AI in healthcare sounds perfect on paper. Faster diagnosis, less admin, smarter systems. But when you look closer and see how hospitals actually operate day to day, things get complicated fast.

AIHealthcareOpinionTechnologyNHS
AI and Healthcare·7 min read·20 March 2026

The Future of AI in Healthcare System Architecture

Healthcare AI is at an inflection point. The architecture that served hospitals for decades is colliding with AI capabilities that demand a different approach. Here is what that means in practice.

AIHealthcareSystem ArchitectureData
Developer Tools·6 min read·8 March 2026

What AI Tool Should I Use as a Developer? The Honest, Unbiased Answer.

GitHub Copilot, Cursor, Claude, ChatGPT, Gemini. Every developer is asking the same question right now. I am not going to tell you which one to pick. I am going to help you think about it properly.

AIDeveloper ToolsOpinionCareerProductivity
AI·10 min read·5 March 2026

The State of AI in March 2026: An Honest Assessment

Three years after ChatGPT, the AI landscape has clarified considerably. What has worked, what has not, what the current frontier looks like, and what I think happens next.

AILLMState of the Industry2026
Career·6 min read·4 March 2026

Do You Actually Need a Certification to Work in AI? Here Is My Honest Answer.

Everyone seems to be chasing an AI certification right now. But I keep asking myself the same question: does a certificate actually prove you can do the work? My honest answer might surprise you.

AICareerCertificationsLearningOpinion
Architecture·8 min read·15 January 2026

LLM Integration Patterns for Enterprise Systems

After two years of integrating language models into enterprise systems, clear patterns have emerged for what works and what causes problems. From prompt management to cost control to compliance.

LLMEnterpriseIntegrationArchitecture
Architecture·10 min read·10 November 2025

Designing Resilient Distributed Systems: Lessons from Failures

I have been involved in enough production incidents to have strong opinions about what makes distributed systems resilient. The patterns are not complicated. The discipline to apply them consistently is.

Distributed SystemsResilienceArchitectureSystem Design
AI Engineering·7 min read·15 June 2025

MCP: The Protocol That Made AI Agents Actually Connect to Things

Anthropic's Model Context Protocol landed in late 2024 and by 2025 had become the standard way to connect AI models to tools and data sources. Here is why it matters and how it works.

MCPAI AgentsClaudeAnthropicIntegration
AI·8 min read·10 April 2025

Reasoning Models: What o1 and DeepSeek Actually Changed

OpenAI's o1 series and DeepSeek's reasoning models demonstrated that thinking before answering produced significantly better results on hard problems. The implications for how we use LLMs are still being worked out.

o1DeepSeekReasoningAILLM
AI·9 min read·18 February 2025

Agentic AI in Production: What Works in 2025

Two years after AutoGPT made agents seem imminent, the production reality is clearer. Narrow agents with human oversight work. Fully autonomous agents are still research. Here is the honest picture.

AI AgentsLLMProductionAI Engineering
AI·8 min read·30 August 2024

AI Engineering: The New Discipline Nobody Had a Name For

Something new emerged in 2024 that sat between software engineering and machine learning but was not quite either. Building reliable products on top of language models required skills and patterns that had not existed before.

AI EngineeringLLMOpsML EngineeringAI
AI Tools·7 min read·22 May 2024

Running LLMs Locally in 2024: Ollama and the Democratisation of AI

Ollama made running large language models on your own machine genuinely accessible. One command to download and run Llama 3. For data-sensitive applications, this was significant.

Local LLMsOllamaLlamaAIPrivacy
AI·7 min read·14 February 2024

GPT-4V and the Year Vision Came to Language Models

GPT-4 with vision capabilities arrived in late 2023 and became widely available in early 2024. Being able to show the AI a screenshot and ask questions about it changed more than I expected.

Multimodal AIGPT-4VComputer VisionAI
AI·8 min read·8 September 2023

LLM Agents in 2023: Impressive Demos, Harder Reality

AutoGPT went viral in April 2023. Agents that could autonomously break down goals into tasks and execute them looked like the next leap. Six months later, the gap between demo and production was clear.

AI AgentsLLMAutoGPTAI
AI Infrastructure·7 min read·20 June 2023

Vector Databases: What They Are and Why They Suddenly Matter

Pinecone, Weaviate, Chroma, Qdrant: vector databases went from niche to essential infrastructure in 2023. Here is what they actually are, why they exist, and when to use one.

Vector DatabasesEmbeddingsAISearch
AI Architecture·9 min read·12 April 2023

RAG: The Architecture That Made LLMs Actually Useful for Business

Retrieval Augmented Generation solved the problem that made LLMs risky for most business applications: they confidently hallucinate facts. RAG grounds responses in real documents and it changed what enterprise AI actually looked like.

RAGLLMVector DatabasesAI Architecture
AI·8 min read·10 December 2022

ChatGPT Launched and Then Everything Changed

ChatGPT launched on 30 November 2022. I tried it within hours. One million users in five days. One hundred million in two months. I have been thinking about what happened ever since.

ChatGPTOpenAIAILanguage Models
AI·7 min read·15 September 2022

Stable Diffusion and What Open Source AI Actually Means

Stability AI released Stable Diffusion openly in August 2022. Within weeks it was running on consumer hardware, generating images that would have seemed impossible a year earlier. The implications were not simple.

Stable DiffusionAIImage GenerationOpen Source
Developer Tools·8 min read·18 November 2021

GitHub Copilot: Six Months of Using AI to Write Code

I got GitHub Copilot access in the technical preview in June 2021. By November I had strong opinions about what it actually was and was not. It is not a replacement for a developer. It is something different and more interesting.

GitHub CopilotAIDeveloper ToolsProductivity
Industry·9 min read·25 August 2021

Web3 in 2021: Separating Signal From Noise

In 2021 it was impossible to work in technology without having a view on Web3. Here is mine, formed after actually building things with Ethereum and Solidity rather than just reading about it.

Web3BlockchainEthereumDeFi
Tools·7 min read·8 October 2020

Low-Code Platforms: Serious Tool or Expensive Toy?

Every vendor was selling low-code in 2020. Some teams I worked with used it to genuinely transform their processes. Others created unmaintainable nightmares. The difference was in how they approached it.

Low-CodePower AutomateDigital TransformationNo-Code
AI·8 min read·20 July 2020

GPT-3: The World Changed and Most People Were Not Looking

OpenAI released GPT-3 access to beta users in July 2020. I got access in August. The first hour with the API was unlike any other hour I had spent with a technology. Something had genuinely shifted.

GPT-3AILanguage ModelsOpenAI
Industry·8 min read·15 April 2020

COVID-19 Forced Digital Transformation Nobody Had Managed to Do

Years of digital transformation strategy documents achieved less in five years than COVID-19 achieved in five weeks. Here is what that tells us about how organisations actually change.

Digital TransformationRemote WorkCloudHealthcare Tech
Infrastructure·9 min read·14 May 2019

Kubernetes in Production: What Nobody Tells You

Running Kubernetes in a lab is manageable. Running it in production for a year is an education. Here are the things I wish someone had told me before we started.

KubernetesDevOpsInfrastructureProduction
AI·7 min read·21 February 2019

GPT-2 and the Day AI Writing Became Uncomfortable

OpenAI released GPT-2 in February 2019 and refused to release the full model, citing misuse concerns. I read the generated samples and understood why. Something had changed.

GPTAINLPLanguage Models
Frontend·7 min read·7 November 2018

React Hooks: Rethinking How We Write Components

Dan Abramov announced React Hooks at ReactConf 2018. The room went quiet, then the Twitter responses started. Hooks changed how we write React components more profoundly than anything since the library launched.

ReactHooksFrontendJavaScript
Data & Privacy·8 min read·25 May 2018

GDPR Changed How We Build Software, For Better and Worse

GDPR came into force on 25 May 2018. The software industry spent the preceding year either preparing carefully or panicking. I saw both approaches close up. Here is what actually changed.

GDPRPrivacyComplianceData
Frontend·7 min read·15 March 2018

JAMstack and the Static Site Renaissance

Static sites were considered old fashioned. Then JAMstack arrived and suddenly static was the sophisticated option. Pre-rendering, CDN distribution, and decoupled architecture made it interesting again.

JAMstackStatic SitesFrontendPerformance
Frontend·6 min read·22 August 2017

CSS Grid Arrived and We Could Finally Stop Fighting the Web

Fifteen years of hacking layouts with floats, then flexbox. Then CSS Grid landed in browsers in early 2017 and two-dimensional layouts became something you could just do.

CSSFrontendLayoutWeb
Data·6 min read·20 May 2017

Python Became the Language of Data in 2017

R had been the data scientist's language for years. Python had always been a strong contender. By 2017 the debate was effectively over. Here is what tipped it.

PythonData SciencePandasAnalytics
AI·7 min read·14 February 2017

TensorFlow Made AI Accessible to Engineers, Not Just Researchers

Google open-sourced TensorFlow in late 2015. By 2017, it had become the default framework for machine learning. What made it special was not the algorithms. It was the engineering.

TensorFlowMachine LearningAIPython
API Design·8 min read·5 October 2016

GraphQL vs REST: What the Hype Got Right and Wrong

Facebook open-sourced GraphQL in 2015. By 2016, the developer community had split into two camps. I was in the sceptical camp. Then I built something with it and changed my mind, partly.

GraphQLRESTAPI DesignBackend
Frontend·7 min read·12 July 2016

Progressive Web Apps and the Promise of One Codebase

Google pushed hard on Progressive Web Apps in 2016. Service workers, offline support, push notifications, and home screen installation. The vision was compelling even if the execution was uneven.

PWAMobileFrontendPerformance
Infrastructure·8 min read·18 April 2016

Why Kubernetes Won the Container Orchestration Wars

In 2016 there were three serious container orchestration systems. By the end of the year, Kubernetes was clearly winning. Here is why it beat Docker Swarm and Apache Mesos.

KubernetesDockerInfrastructureDevOps
Frontend·6 min read·15 November 2015

ES6 Was the JavaScript Update We Had Been Waiting For

Arrow functions, destructuring, promises, modules. ES6 arrived in 2015 and suddenly writing JavaScript felt like writing a real programming language.

JavaScriptES6Frontend
Frontend·8 min read·8 September 2015

Why React Felt Different From Everything Before It

React was not the first JavaScript framework. It was not even obviously the best one. But something about the way it thought about UI made it click in a way Angular and Backbone never quite did for me.

ReactJavaScriptFrontend
Architecture·9 min read·3 August 2015

The Honest Truth About Microservices in 2015

Microservices were everywhere in 2015. The conference talks made them sound inevitable. The reality I saw was more complicated: teams solving distributed systems problems they did not have before.

MicroservicesArchitectureSystem Design
Engineering Culture·6 min read·20 June 2015

The Rise of DevOps Was Not About Tools

Everyone thought DevOps was about Jenkins and Ansible. The teams that actually succeeded figured out it was about removing the wall between the people who build software and the people who run it.

DevOpsCultureAutomation
Infrastructure·7 min read·12 March 2015

Docker Changed Everything and Most People Missed It

In early 2015 I watched a colleague containerise an app in twenty minutes that had taken three days to set up on a new server. That was the moment I understood what Docker actually was.

DockerContainersDevOps
DevOps·7 min read·18 February 2015

Configuration Drift and Why Manual Setup Keeps Failing

Everything works perfectly until you try to recreate it somewhere else. That is when the problems start. I have been noticing a pattern and it has a name: configuration drift.

DevOpsConfigurationAutomationInfrastructure
Infrastructure·6 min read·12 January 2015

Virtual Machines vs Bare Metal: Where Performance Still Matters

Everyone around me keeps saying the same thing lately: just spin up a VM, it's easier. And they're not wrong. But I've started noticing something strange — sometimes things feel just a bit slower, and nobody can really explain why.

InfrastructureVirtualisationPerformanceBare Metal