Build vs Buy: The AI Shift Begins

Friends,

our weekly AI briefing is here - designed to help you respond to AI, not react to the noise. No curveballs. No chaos. Just clarity.

🌟 An Extra Dose of AI Optimism

AI isn’t the enemy - our mindset is. This video dives into how leaders can shift from doing to being as the cost of doing drops to zero.

If you’re curious about how to stay human while scaling with AI, hit play.

And yes, the algorithm appreciates full watchers - so please stick around, comment, and subscribe for the next one.

📰 This was the week that was...

Jensen Huang met with the King of England to discuss the AI race between the US and China. Meanwhile, a short trader famous for the Big Short placed bets against AI.

But the real story? Microsoft's aggressive Copilot bundling tactics. When you don't own your AI, you're not in control of it. The build versus buy decision that's been firmly "buy" throughout 2025 is shifting.

🔥 Urgent Priorities

✅ No fires to fight

✅ Strategic foundations shifting

✅ Time to plan for ownership, not just adoption

This week doesn't demand urgent action - but it does demand strategic thinking about 2026.

🎯 Strategic Insight

Tension: The market is flooded with AI tools to buy. But AI is so wide and deep that most leaders don't know enough to make good procurement decisions. And if you don't own your AI, you're not fully aligned with it.

Optimistic insight: 2026 is the year to prepare for digital self-sovereignty. The path is becoming clearer - tools like Ollama let you run local large language models, and training is the key to making informed decisions.

What's shifting: On Monday, Tuli and I hosted charity trustees on an AI Governance course, outlining a framework for ethical AI governance. At Wednesday's AI Morning Salon, we discussed running local LLMs. The build versus buy decision has been firmly "buy" for most of 2025. That's about to change.

Why this matters now: Training is critical to AI adoption plans. Plan in 2026 to assess how you make the most of rapid development and prepare for when you'll need to move to digital self-sovereignty.

👉 Takeaway: Reach out to discuss your training needs. Start mapping what you'd need to own your AI infrastructure.

🤓 Geek-Out Stories

1️⃣ Google's Project Suncatcher: Data centres head to space

Google announced Project Suncatcher - AI data centres on solar-powered satellites. Solar panels in orbit generate up to eight times more power than on Earth. Two prototype satellites launch in 2027.

📖 Read more

Why it matters: AI data centres could consume 3% of global energy by 2030. This addresses the fundamental infrastructure challenge facing adoption - energy and sustainability. The energy constraints aren't permanent, they're being solved. When planning your AI strategy, factor in that compute economics will dramatically improve. But consider: do you want to be locked into providers' pricing as economics shift, or positioned to benefit from falling costs?

2️⃣ Vibe coding becomes mainstream: Collins names it Word of the Year

Collins Dictionary named "vibe coding" its Word of the Year for 2025. The term describes using AI to convert natural language into working code - "programming by vibes, not variables."

📖 Read more

Why it matters: Non-technical people are becoming technologists, hugely empowering for innovation. But leaders face three challenges: total cost of ownership (vibe-coded apps become expensive to maintain), governance structures (who can build what?), and skill development (when to involve professional developers). The winners in 2026 will create the right structures to harness benefits whilst managing risks.

3️⃣ LibRAG: Vector-free retrieval transforms knowledge engineering

LibRAG is a new open-source RAG architecture. Instead of embedding models and vector databases, it uses end-to-end LLM reasoning. Claims over 90% accuracy out of the box versus under 50% for traditional approaches.

📖 Check it out on GitHub

Why it matters: Knowledge engineering is how you build AI tools that actually work for your business. Until now, this required specialist teams and expensive infrastructure. LibRAG changes that - it's simpler to run and maintains itself. For the first time, mid-sized organisations can own their AI knowledge systems rather than renting them. If you're planning your 2026 strategy, this technology makes ownership viable where it wasn't before.

🎨 Weekend Playground

Perfect your prompting with AI-powered tools

This weekend, try Prompt Perfect and Prompt Cowboy - tools that optimise your prompts for you.

Your mission: take a prompt you use regularly and see how these tools transform it.

🧪 Try Prompt Perfect 🧪 Try Prompt Cowboy

Why this matters: Prompt engineering is becoming automated. These tools show where we're heading. Notice how they interpret your intent and whether results are actually better. That feedback loop is the skill that will matter most in 2026.

📢 Share the Optimism

If The AI Optimist helps you think more clearly, forward it to someone else navigating the shift. If it's not quite landing, hit reply and let me know - I read every message.

Stay strategic, stay generous.

Hugo & Ben