Connect The Dots
|
PLUS: Open source AI might be about to hit the mainstream "This is the biggest technological revolution of my life. This is clearly bigger than the internet. The comps on this are things like the microprocessor and the steam engine and electricity. We're three years into what is effectively an 80-year revolution." - Marc Andreessen
Morning All, In the 1930s, the pioneers of computing faced a fundamental choice. Build computers in the image of the adding machine (calculator) or build computers modelled on the human brain. Neural networks. They chose the adding machine. And that single decision shaped everything...mainframes, PCs, smartphones, every dollar of global wealth the computer industry created over the next 80 years. The vision for AI existed as far back as 1946. That vision was kept alive by a small academic movement of researchers and scientists whilst everyone else just assumed it was never going to happen. But those scientists kept working. None more so than the researchers at Google. In 2017, 8 of them discovered the breakthrough that makes modern AI possible. They discovered transformers, the T in ChatGPT...and everything changed forever. Most people are treating AI like another technology cycle. Something to adapt to, ride, and wait out. If Marc Andreessen is right, that would be a misguided way to frame what we're living through. We are not adapting to a new cycle. We are standing at the very beginning of the most consequential transformation in human history. The current AI industry is structured like the image below: 1. Infrastructure Layer (Foundation and Physical Backbone)This layer forms the “power plant and highway” of the industry. 2. Data Layer (Raw Material)This layer defines the scale, scope, and ceiling of AI capabilities. 3. Model Layer (The Intelligent Core and Deep Learning Frameworks)Foundation model providers train the “engines” of AI systems. 4. Application Layer (User-Facing Interfaces and AIGC Products)This layer delivers generative capabilities directly to end-users (AIGC products).
5. Agent & Tools Layer (The Multi-Agent Paradigm)The most transformative layer of 2026. It transforms passive AI into proactive systems capable of planning and environmental interaction. 6. Vertical Solutions Layer (Industry-Specific Applications and Closed-Loop Value)The top layer integrates everything into domain-specific solutions where commercial value is fully realised. There are thousands of companies and hundreds of thousands of products across every layer of that stack. Now, it is my belief, when all's said and done and the dust settles, you will come to rely on a handful of individual companies...and one company will be standing taller than anyone else...GOOGLE! As we sit here today, Google is the only global company that has a viable footprint on every layer of the AI pyramid.
My analysis is that Google is running a two pronged AI strategy that positions them brilliantly to capitalise on whichever direction the AI industry moves in. And collect even more of your money! Prong 1 is Gemini (closed, proprietary, paid, cloud only). Prong 2 is Gemma (open source, free, runs anywhere). Both are built on the back of the same research, Gemma 4 literally uses Gemini 3's architecture. Here's the genius play: Gemini captures enterprise revenue while Gemma competes with China to dominate the entire open-source ecosystem. While developers build on Gemma, the hope is they get locked into Google's ecosystem. With options at every level of the stack (AI Studio, Edge Gallery, Firebase, Vertex etc) the proposition would be compelling. Apple have run the same playbook...and it fuelled the growth of the app store. The best phone on the market right now is probably an iPhone 17 Pro Max. Would you rather pay money to get that phone today or would you wait 6 months and get the same iPhone 17 Pro Max for free? That's a similar situation to the one we're in with closed and open source AI models right now. The open source alternatives are getting so good they're not that far behind frontier AI labs. Why this matters: Frontier model providers aren't going to keep subsidising cheap tokens on subscriptions forever. Especially if the trend towards 24/7 autonomous coworkers continues. In fact just this week, Anthropic essentially banned the use of OpenClaw with Claude Code. This type of protectionism is probably only going to increase, along with the price of using the service/product. As a result I'm extremely optimistic about the future of open source and local models. Imagine having the power of ChatGPT running directly on a £600, 16GB Mac Mini you bought last week. No subscriptions, no pay as you go credits. All running on your own machine. For free. You don't have to imagine anymore, it's already possible. There are already some amazing products being created in the open source community. Here's just a few examples... CareerOpsWorried about AI coming to replace you? Job hunting can be so time consuming that it feels like it could be a full time job itself. Why not use AI to find your next role? CareerOps is an AI-powered job search pipeline that turns Claude Code into a full job search command centre. Instead of manually tracking applications in a spreadsheet, you get an AI-powered pipeline that:
GraphifyPattern recognition and the ability to make connections that aren't obvious is what sets humans apart. The most successful of us are the ones that are best at that. Now AI is helping to level the playing field with Graphify. Graphify is an AI coding assistant skill that reads your files, builds a knowledge graph, and gives you back structure you didn't know was there. Drop in code, PDFs, markdown, screenshots, diagrams, whiteboard photos, even images in other languages and graphify uses Claude vision to extract concepts and relationships from all of it and connects them into one graph. MemPalaceAI systems are incredibly smart but have the memory of a goldfish. The reasoning, the tradeoffs, the "we tried X and it failed because Y" - all trapped in chat windows that evaporate when the session ends. That's every decision, every thought process, every back and forth debate. Gone. You start over every time. Giving an AI system the ability to remember what you talked about turns it from a bad intern into a genuine senior colleague. MemPalace is The highest-scoring AI memory system ever benchmarked. And it's free. It starts with a wing. Every project, person, or topic you're filing gets its own wing in the palace. Each wing has rooms connected to it, where information is divided into subjects that relate to that wing. So every room is a different element of what your project contains. Project ideas could be one room, employees could be another, financial statements another. MemPalace takes the approach: store everything, then make it findable. MemPalace keeps every word, and the structure gives you a navigable map to find what you need. Those 3 are examples from just the past week. If indeed we are only three years into what is effectively an 80-year revolution...just imagine 5 years from now, the supreme levels of intelligence we'll have available at our disposal...all for free. What a time to be alive. P.S. Here's The ShortlistOther stories I think are worth your time...
|