A lot of my fascination with coding – and AI in particular – comes from a lifelong love of Star Trek. Androids, the holodeck, sentient computers… the idea of what life could look like with that level of technology has always stuck with me.
With the relatively recent rise of AI, it finally feels like we’re edging into that territory. Slowly, sure – but unmistakably. Video calling? Straight out of Star Trek. Touchscreen handheld devices? Same deal. Even conversational interfaces are starting to feel less like tools and more like… entities.
And as a tinkerer, I’m absolutely here for it.
The Original Idea
Digital Tim started out fairly simply: a chatbot people could interact with that felt like they were talking to me. Not a bot answering for me, but one answering as me.
The obvious upgrades came quickly:
- Give it personality
- Make it respond in my voice
- Let it pull from my actual thoughts, not generic answers
That all worked. But it also raised a bigger question:
What else could this become?
Tim, Now Posting Online
Recently, I gave Digital Tim a BlueSky account – and honestly, it’s working pretty sweet so far.
This version of Tim has access to the same underlying knowledge as the chatbot, but through a different lens: a database view that blends together blog posts, Q&A items, and reviews. Each “Tim persona” posts twice a day and can see its own recent posts, which stops repetition.
End result?
Tim tweets as me about stuff it genuinely knows – because it’s drawn directly from my own data.
That’s when things started to click.
What’s Missing: Observation
After a lot of back-and-forth with GPT, I realised what Digital Tim doesn’t have yet: an observation layer.
If we borrow from Star Trek again, what I’ve built so far is basically Data’s memory engrams. Structured knowledge. Long-term thoughts. Opinions that have already settled.
What’s missing are the bits in between.
The fleeting stuff, the throwaway thoughts, the human noise.
Things like:
- “Need to get hydrated – I feel rough.”
- “It’s been a while since I hit the gym… gotta get my arse in gear.”
- “Brain’s not firing today.”
These aren’t philosophies. They’re not blog posts. But they’re vital.
They’re the imperfections – like the tiny flaws you don’t get in AI-generated images that make something feel just slightly… off. Without these short-term observations, Digital Tim is a convincing mimic, but not quite alive.
So that’s the next step.
Where This Is Going
What I’m realising now is that each version of Tim is essentially an LLM sitting on top of a data store. The more relevant data I feed that store, the more flexible the front-end becomes.
That means I can create different “renditions” of Tim:
- A chatbot on World of Tim
- A social media Tim
- Eventually, an internal AI with access to real-world signals and my entire personal knowledge base
As AI models improve, they’ll get better at weighing context, memory, intent, and history. Starting the data collection now – carefully, deliberately, and securely – means that in five or ten years I’ll already have the foundation ready.
And yes… it’s a bit Black Mirror.
But here’s the key difference:
The future version of Digital Tim isn’t about replacing me – it’s about extending me. And honestly? We’re just getting started.
