What a week it’s been. We’ve journeyed from the dizzying heights of the latest LLM releases, down into the trenches of building with AI, and through the murky waters of its societal costs. To wrap things up, I want to return to my natural state: optimistic and forward-looking. Let’s talk about where this is all heading.

The Next Leap: From Reactive Tools to Proactive Partners

The journey so far has mostly focused on AI as a reactive tool. You give it a prompt, it gives you a response. Even the RAG system we discussed on Wednesday still waits for your input.

After wrestling with GPT-5’s so-called ‘router’ all week, which seemed to have the directional sense of a lost tourist, it got me thinking. We don’t just need smarter models; we need  

smarter systems. This is Phase 2 of the AI journey: the shift to proactive, goal-based agentic workflows.

An AI agent is more than just a chatbot. It’s a system that can reason, plan, remember, and use tools to autonomously achieve a goal. Instead of just answering a question, you can give it a complex task, and the agent breaks it down, uses tools, and takes action. We’re already seeing glimpses of this future in research prototypes like  

Google’s Project Astra, a multimodal assistant that can see, understand, and act on your behalf. This is the next leap: from a tool you command to a partner that collaborates.

My Next Big Idea: A PA Named “Loraine”

This shift to proactive agents has inspired a product idea I’ve been kicking around. I call it Loraine.

Loraine isn’t just another app. It’s a vision for a true, context-aware Personal Assistant. If the “agentic mesh” is the operating system for an enterprise, Loraine is the “OS for your life”.

It would be a personalised system that manages the different facets of your life through specialised “modes.”

• Initial Setup: You’d start by answering 20 or so questions to profile your priorities and values.
• Work Mode: Manages your diary and messaging. Imagine an out-of-office that, when you’re on holiday, not only replies to customers but keeps projects moving!
• Kid Mode: Proactively suggests activities based on your location, the weather, and your child’s interests. Maintains score when you have to referee a match at the last minute. Add the latest tik-tok or Youtube meme to your playlist to have street cred [well at least be aware of what’s the latest trend – some of us are beyond street cred salvation]
• Partner Mode: Remembers anniversaries and birthdays, proactively suggests date nights, and organises weekends away.
• Friends Mode: A group scheduling agent that plugs into your friends’ Loraine apps to finally solve the eternal problem of trying to get five busy adults in the same pub at the same time—a task currently harder than achieving cold fusion.

And here’s the kicker, to keep the corporate vultures at bay and ensure Loraine’s always on your side, the whole thing would be set up as a charity. Well, mostly! The owner would take a modest cut, naturally, but the core mission would be locked in by contract to serve the user, not to maximise shareholder value.

The architecture would likely be a hybrid model: a lightweight agent on your phone for secure access to local data, connected to a more powerful cloud component that houses your long-running personal vector database—your life’s memory.

The Loraine Dilemma: Building an Ethical AI Companion

Of course, a system this personal is fraught with ethical peril. Building Loraine responsibly would be even more challenging than building her technically.

My little joke about this being a “charity” has a serious undertone. A system this powerful would need a governance model that guarantees its goals remain aligned with the user’s well-being, not a corporation’s bottom line.

• Privacy and Data Governance: The user, and only the user, should own and control their data.
• Bias and Fairness: How do we ensure the “Partner Mode” doesn’t reinforce gender stereotypes? Constant auditing for fairness would be paramount.
• Autonomy and Manipulation: There’s a fine line between a helpful suggestion and a manipulative “dark pattern.” Loraine must be designed to respect user autonomy.
• Accountability: If Loraine misses a critical medication reminder, who is liable? The lines of accountability must be crystal clear.

Though let’s be honest here, what have I just described with ‘Loraine’? Basically all I needed from Tim’s AI-phone 16 – Look at the requirements and there’s actually very little AI, Apple missed a trick here by failing to release some state of the art systems automation workflows and then follow up with the AI sauce once they acquire the right start up – could they still afford Perplexity?

The Glass is Still Half Full: AGI and a Better Tomorrow

This brings us to the end of our week and to “Phase 3” of the AI journey: the ongoing cycle of innovation pushing us toward the ultimate, if still distant, goal of Artificial General Intelligence (AGI)—an AI that can perform any intellectual task a human can.  

The timeline for AGI is hotly debated. Experts predict dates around 2040 or 2060, while some bullish entrepreneurs think it could be here by the late 2020s. Personally, I think the exact date is less important than the direction of travel.  

And for me, that direction is overwhelmingly positive. I started this journey as a kid with a Sony MSX computer, my imagination fired up by the promises of science fiction. For decades, it felt like that promise was just out of reach. Now, it’s finally starting to materialize.

AI has the potential to be the ultimate tool for human augmentation. It can free us from drudgery, giving us back time to spend with our families and communities. It can help us tackle humanity’s greatest challenges, as championed by “AI for Good” initiatives around the world.  

Yes, there are immense challenges. We must face them head-on, with honesty and a commitment to building a future that is equitable and sustainable. But I remain a steadfast optimist. My glass is, and always will be, half full. Fourty-four years after my parents first brought a computer into our home, the technology is finally starting to deliver on the dream. And I, for one, can’t wait to see what we build next.