Over the past few days — three full days and almost two nights in between — I threw myself headfirst into the world of AI. Not just reading about it casually, but diving deep into the mechanics of Chat-Bots, app development around AI, and the more technical aspects like persistent memory and smart data storage for language models. It was intense, a bit chaotic at times, but incredibly rewarding.
I started with the basics, trying to truly understand how these large language models tick. How they handle inputs, how they structure outputs, and especially — how they can remember past conversations when persistent storage is involved. This was one of the more fascinating areas for me: enabling LLMs to store memories, recall stories we talked about, and build context that spans across multiple sessions. It’s not just about smart replies anymore, it’s about continuity. And that opens up a whole new dimension of interaction.
Next came the real nerdy stuff: building in multiple search methods so the models aren’t working in isolation. I explored ways for LLMs to fetch fresh data from the web, expand their knowledge dynamically, and support responses with real-time information. It’s kind of wild to think about — a model that doesn’t just rely on pre-trained data, but can actively seek out new knowledge on the fly.
Of course, no journey like this comes without hurdles. I read a mountain of documentation (some clear, some not so much), watched hours of video tutorials, experimented with demo scripts, and pushed various setups to their limits. There were moments where things broke for no clear reason, where I got stuck in endless error loops, and where I had to walk away for a breather. But every frustration came with a breakthrough — and slowly, piece by piece, things started to click.
Now, I’ve got a working set of demo scripts running flawlessly. LLMs connected to persistent storage, integrated with internet search, able to “think” with context and pull in new insights. Seeing it all come together was one of those moments where you just sit back and say, *yeah, this is cool*.
So what’s next? Well, now it’s time for something a little more fun.
I’m going to experiment with automating some of my Social Media activity — not with generic schedulers or basic bots, but with some incredibly smart, adaptable LLMs. The kind that can write posts, analyze engagement, adapt tone based on audience, maybe even reply in threads. Nothing serious yet, just playful testing. But with the kind of foundation I’ve set up, who knows where it could go.
I’ll be keeping track of the journey, the tests, and the results — all on my personal blog at https://about.tomdings.com/. If you’re curious about AI, automation, or just want to see what these models are capable of in real-world scenarios, feel free to follow along.
This is just the beginning.
