AI Resources

In the late 1990s, the concept of a burn rate was used to justify the stocks of loss-making companies. Investors prioritised user growth over earnings, assuming that profitability would follow once the market was captured. We are seeing this pattern again in the generative AI sector. OpenAI generated 13 billion dollars in 2025, yet

Artificial Intelligence or AI isn’t just a buzzword anymore, and it most definitely is not something reserved for the future. It’s already here, woven into our daily lives in ways we often don’t even notice. It’s everywhere!

There is an important paper, which is about smaller language models. It has a few implications that are also quite important. It’s about the increase in capability per parameter in large language models over time, called Densing. Densing makes smaller models more competitive and CPU deployment more realistic even for tasks that depend on generative capa…

We’ve all been witnessing the rise of powerful chatbots that shocked all of us with this spark of language that we thought only humans are capable of and there was a lot to learn and to catch up on. One of the most powerful things however was to learn about human value in the world and about things that you can’t automate.

Let’s face it: Chatbots and LLMs can speed up extracting and synthesising information a lot. It’s like when Wikipedia just came out and was widely discouraged as an authoritative source for research. Today, we are slowing coming to terms with what AI can do for us, and we are finding out

Large Language Models (LLMs) are sold on a promise of leverage: they let individuals do more, faster, with less friction. In software and knowledge work, the pitch is that AI collapses the cost of creation.

The honeymoon phase of Generative AI is over. We are now entering the Implementation Gap—that frustrating valley between the viral demo your CEO saw and the broken prototype your engineering team is currently debugging.

The promise of agentic AI is seductive. It suggests a move from simple text generation to autonomous research and long-horizon problem-solving, which involves a lot of decisions. We expect a digital worker that plans, executes, and corrects its own work. For businesses, automating tasks that are laborious and repetitive can help save on labour costs an…

It’s the end of 2025 and time for an end of year summary and some predictions.

I frequently speak with teams at large companies where the mandate comes from the top. The boss tried ChatGPT, found it intuitive, and now expects an autonomous agent that solves complex workflows. The technical team has been trying it for months, the reliability is poor - even if the accuracy is acceptable (which is rarely seems to be), it fails specta…

The evidence is overwhelming across domains. As a model takes more steps, its reliability falls. As we grant it more autonomy, it fails more often. Agent demos may look intelligent, but real evaluations paint a harsher picture.

In his book from 2008, Lawrence Lessig famously described two types of culture. He called them “Read-Only” and “Read-Write”. Radio is a perfect example of Read-Only culture. A few professionals broadcast the signal and you listen. You cannot change the show nor can you talk back. The flow of information goes one way. Lawrence Lessig argues that this cha…

Privacy is the moral space individuals need to think, choose, and become themselves. It is not simply a legal checklist or a mechanism for compliance. It is a fundamental moral right that protects our capacity for autonomy. Without a private sphere, we cannot freely form beliefs or pursue our own conception of a good life. This understanding of privacy …

The internet has a problem. There’s a flood of low-quality, machine-generated text that now fills search results. It is grammatically correct. It is also boring, repetitive, and obviously fake. Users are learning to spot it instantly. They hate it and when they see it on your website, they leave.

We are deep in an AI Hype Cycle. The potential for AI to automate human labour is a subject of profound societal interest and concern. We constantly hear claims about AI’s rapid progress on complex reasoning benchmarks, which has fuelled a narrative that a new wave of “agentic AI” is ready to automate large parts of the workforce.

We built a tool that tells you exactly why your website visitors are leaving. We’ve built lots of different reports that use AI and different kind of heuristics, and a detailed dashboard that shows you problems and recommendations. Now, we need your help to test it.

ChatGPT has been a turning point in technological innovation. Even though other commercial providers have been catching up and overtaken OpenAI in a few areas, it was when ChatGPT entered the scene was when people took notice of the potential of AI. While the initial phase of AI adoption was characterized by renting generic capabilities from a handful o…

GPT-5 Is Here, and My Clients Are... Underwhelmed

The "AI Efficiency" Smokescreen

According to analysts at Morgan Stanley, Goldman Sachs, and McKinsey & Company AI was going to massively redefine more than the work force within a few years automating tasks and replacing a number of jobs. The predicted ranges of displacement or disruption still vary widely.

Today, AI is no longer just an “assistant” to humans—it’s rapidly becoming the core executor within systems. Cosimo Spera and Garima Agrawal propose flipping the traditional “human-led, AI-assisted” paradigm; build AI-First systems, where AI leads and humans provide strategy, ethics, and oversight in the loop.

Remember when everyone was getting fired because of AI? Plot twist: the robots are still figuring out how to fill out expense reports.

Introduction

Imagine you’ve asked an AI model to draft a critical business proposal, and it generates a complete—but subtly off—document. Without a human reviewing it, that small misalignment could lead to missed opportunities or misunderstandings. This is where Human in the Loop (HITL) becomes not just helpful, but essential.

As a Maths student at Imperial working through problem sheets, coding labs, and club activities, I’ve noticed AI assistants becoming part of my daily routine. Within minutes, the AI suggested a corrected loop structure, recommended vectorised NumPy operations, and even drafted clear comments.

Why Prompt Precision Matters

Have you come across this experience? Reading through social media, found most of the posts are obviously LLM-produced. This seems to be the trend: social media platforms like ‘LinkedIn’ and ‘X’ have AI integrated for writing posts.

Recently, Builder.ai, a company founded in London and backed by Microsoft whose valuation relied on their revolutionary coding AI Natasha, was discovered to use 700 Indian engineers instead of an AI. Moreover, these engineers, it seems, often churned out rather buggy code. They are also accused of having inflated their revenue figures.

Retrieval-Augmented Generation (RAG) has evolved dramatically since its introduction in 2020. While simple implementations can deliver impressive results, today's most challenging use cases demand more sophisticated approaches. In this article, I'll show you how to leverage LangGraph—a powerful extension to the LangChain ecosystem—to build advanced RAG …

Practical strategies for moving beyond simple ChatGPT usage to sophisticated enterprise LLM applications. Learn how to address common challenges like context management, security, and evaluating outputs at scale.

Large Language Models (LLMs) have made impressive strides in understanding and generating text. Yet, when it comes to tackling complex, multi-step problems, traditional prompting methods like Chain-of-Thought (CoT) can fall short.