TRMNL X - integrating into my existing setup
I thought it would be fun to share some of my experience with the terminal X, specifically around the ease with which I was able to integrate it into an existing setup.
I created a set of five custom plugins covering things like:
- AI-powered daily briefings and tasks / messaging / calendar triage
- my personal health metrics
- status of my smart home
- personalised weather forecast
I had pre-existing infrastructure (somewhat cheekily called RafeOS), which is really, a set of tools, APIs, data, and second brain context that I was already using in my personal AI/agentic setup. For various reasons, I opted to use the webhook approach to get data into the TRMNL platform and then used the framework and design system to get to what I thought was good-looking screens in a very short amount of time. Not quite at the level of quality that I see in some of the plug-ins being featured here, but still very satisfying for me personally.
So right now my TRMNL X is cycling through screens from RafeOS, a personal ecosystem / agentic system I've been experimenting with this year. It shows an AI-powered daily briefing ("get moving Rafe"), my latest health metrics ("also get moving Rafe"), the status of my smart flat ("it's hot everywhere and the windows are open"), current and forecast weather ("it's hot now and tomorrow"), and so on (the current UK heatwave is not helping with differentiated testing conditions).
What's interesting for me is that it has become another surface, one that I've used more than I expected. I really like the passive glance use case, complementing what comes up on my laptop, my phone, wall tablet etc. and pleasingly, it means I've spent a little less time looking at other screens, which is never a bad thing. For what it's worth, the things that have been most useful to me have been smart home information, health trends / scores, and live weather. I think that's because the daily briefing comes up on other services, whether that's in a smartphone app or delivered via email or messaging.
I was able to pick up the TRMNL platform very easily, thanks to the quality of the platform, tooling, and documentation. Being transparent, I used AI (a combination of Claude and Codex) to do a lot of the work for me. I was particularly impressed by the agent friendliness of both the documentation and the tooling, which meant I was able to go from nothing to a five-screen set of custom plugins and scheduled updates in a handful of hours. That's testament to both the platform itself and what you can do with directed AI building / development. That said, it's obviously helpful to have some pre-existing technical product and design skills because it enables you to have a certain amount of rigour and make sensible choices. It really is a good example of where the value of judgement comes in an AI-assisted world.
I've had no problems with the hardware, which I'm very happy about, having read some of the other posts on Reddit. I did get caught out by the UK custom thing. Overall, I've been really impressed with the experience and the quality of the hardware. I wasn't completely convinced by the puck and the magnets, but in day-to-day use, but it did make it easy to get mounted on the wall, so I guess what I'd say is when it works, it works really well.
I've written up some of my early experience here: https://rafeblandford.com/e-ink-screen-on-the-wall-trmnl-x/ and here https://rafeblandford.com/developer-experience-trmnl/