AI Engineering Platforms: Kubernetes, GPUs, Agents and Observability
As AI features and agents move into core products, organisations need a platform for them, just as they needed one for microservices. We out…
Category
AI & LLMs
25 articles tagged AI & LLMs.
AI Engineering in 2026: Agents, Models, Tools, Evals and Observability
AI engineering has become its own discipline, distinct from both machine learning research and traditional software engineering. This mid-20…
MCP Security: The New Security Boundary for AI Applications
MCP has become the standard way AI applications reach enterprise systems, and that makes the MCP layer a security boundary in its own right,…
Agentic AI Architecture: Building AI Systems That Can Actually Act
Moving from AI that answers to AI that acts changes the architecture. Actions need identity, authorisation, state, reliability and accountab…
AI Agents in Production: Architecture, Memory, Tools and Evaluation
Agent SDKs, standard tool protocols and better models have made AI agents a production reality in 2025. This article sets out a reference ar…
Reasoning Models: Why AI Started Thinking Before Answering
OpenAI's o1 and o3-mini, DeepSeek-R1 and now Claude 3.7 Sonnet's extended thinking all spend extra computation reasoning before they answer.…
MCP for Developers: Connecting AI to Real Software Systems
The Model Context Protocol gives AI assistants a standard way to use your tools and data. Three months after its release, clients and server…
MCP Explained: The Missing Layer Between AI and Applications
Every AI application has been building its own integrations with tools and data. The Model Context Protocol, released as an open standard in…
Building Reliable AI Agents With LLMs
Agent demos are easy; agents that work reliably in production are not. After a year of building them, the patterns are clearer: constrain th…
Multimodal LLMs: Beyond Text-Based AI
Two weeks ago, GPT-4o showed a single model that sees, hears and speaks in real time. Gemini 1.5 reads hours of video, and Claude 3 reads ch…
RAG vs Fine-Tuning: Which Should You Use?
It is the most common question we hear from teams building with LLMs: should we use retrieval-augmented generation or fine-tune a model? The…
RAG Explained: Giving LLMs Access to Your Own Data
Large language models do not know your policies, products or documents, and fine-tuning is a poor way to teach them facts. Retrieval-augment…
AI Agents: Moving From Chatbots to Autonomous Systems
AutoGPT became one of the fastest-growing projects on GitHub within weeks of its release, promising AI that pursues goals on its own. Behind…
Building Software With GPT-4: From Prompt to Production
GPT-4 and the ChatGPT API have made it easy to build an impressive LLM prototype in a day. Getting it to production is another matter. This…
GPT-4 and the Rise of Multimodal AI
GPT-4 arrived last week with markedly better reasoning, a much longer context window and the ability to understand images as well as text. W…
ChatGPT: Why Conversational AI Suddenly Went Mainstream
ChatGPT reached a million users in about five days. Its underlying technology is not radically new, so why has it captured public attention…
Prompt Engineering for Developers: Programming With Natural Language
A week after ChatGPT's launch, developers are using it to explain code, write functions, draft tests and debug errors. Getting good results…
Generative AI Explained: From Text Generation to Image Generation
In six months, DALL·E 2, Midjourney and Stable Diffusion have turned text-to-image generation from a research curiosity into something anyon…
Prompt Engineering: The New Interface for AI
With instruction-tuned models like InstructGPT now the default in OpenAI's API, and research showing that 'let's think step by step' can tra…
Fine-Tuning Large Language Models: The Early LLM Playbook
Prompting a large model gets you surprisingly far, but many production tasks still need fine-tuning. As models grow into the billions of par…
Foundation Models: The Beginning of the LLM Era
Stanford researchers have given a name to the models reshaping AI: foundation models, trained once on broad data and adapted to countless ta…
Transformers vs Traditional NLP: What Changed?
Natural language processing used to be a pipeline of hand-built steps and task-specific models. In three years, pre-trained transformers suc…
Transformers Explained: The Architecture Behind Modern LLMs
BERT, GPT-3, T5 and now vision models all rest on the same 2017 architecture: the transformer. This explainer walks through how transformers…
Few-Shot Learning Explained: Teaching Models With Examples
Traditional machine learning needs thousands of labelled examples per task. GPT-3 has made 'few-shot' learning, where a model picks up a tas…
GPT-3: Why Scaling Language Models Changed AI
OpenAI's GPT-3 is more than a hundred times larger than GPT-2, and it can perform tasks it was never explicitly trained for, just from a des…