I had made a few draft posts that I'm deciding to post here tonight. These are disjointed, slightly out of date and all over the place. I'm full steam ahead on Lazurite & am now working a lot on
What I've built for @flipper.social so far (called jank.mp4) This uses my rust-y readability/article parser, lectito. It's meant to combine what I love about hypothes.is, @margin.at, @semble.so and create a way to manage learning and reading a lot like Readwise!
Being a cover band on the ATmosphere
Observing the atmosphere conf and learning about all the cool projects being made on top of the protocol has me wondering about where I fit in to the community. This is a part of a bigger question I've been asking myself the past few months as I observe software engineering and programming fundamentally change thanks to LLMs and agents. I think they're force multipliers, highlighting strengths and weaknesses of the engineers and teams that use them, myself included.
I'm not as well versed in the discourse around what's going on with open source, licensing, and people copying other people's work but I'd like think that while my major projects are in the pre-release phase, I've been doing my best to give credit to inspirations. My process is always to look at features in other apps and architect them in a way that makes sense to me and fits in my codebases.
None of my major projects really reinvent the wheel in any way.
Lazurite for both mobile and desktop is just another BlueSky client, borrowing and building upon features in apps like pds.ls, skeetdeck, witchsky, and clearsky. My personal spin on mobile is using Flutter to create a polished experience mostly for Android users like myself and can act as a teaching tool for engineers wanting to try out flutter and the dart language. On desktop I want to highlight Jacquard's utility as an integration backbone for AT Proto apps that use Rust. Tauri is also a phenomenal tool for desktop app development1 and it helps me adhere to my obsession with local-first as a guiding principle for application development. Other projects in my quiver are Murmur is an audio transcription application that uses Apple Intelligence or Ollama (Windows & Linux) and Twisted/Twister which are just a mobile front-end and search API sitting on top of Tangled. Finally there's Commonplace which I'd call iAWriter fan art with some integrations and PDFs (I have no idea if iaWriter exports PDFs).
Speaking of cover bands...here's one of of my favorite songs ever, covered by some of the most talented musicians I've ever seen.
Twiste{r/d}
I deployed Twister on Railway a few weeks ago, in what felt like some kind of possession, overcoming this fear of shipping and a little bit of imposter syndrome. It feels like a big deal.
The whole system was a few web services in a project, all using a libsql database on Turso cloud. It ended up being a fairly painless process, as managed/PaaS deployments tend to be. But within a day of experimenting with indexing and adding microcosm.blue integrations for better document construction, I was burning through usage. I used up much of the hobby plan credits I purchased and nearly ran through all of the read and write quota on Turso's free tier. So I was left with two options:
- 1.
Add credits to Railway and upgrade my Turso account
- 2.
Use an affordable VPS
I went with option 2 on a VPS with Coolify. This ends up being a fixed ~$10 per month without the risk of ballooning up to the estimated $30 for just the web services.
Since I released this, tangled added search! Twister will live on as the API I use for Twisted.
Murmur
Murmur is a local-first AI-powered transcription application I've been working on for the past month and a half. Inspired by talat.app, I decided to try out using Apple Intelligence bindings for the Mac deployment in place of using Ollama with Gemma 3 for metadata (really summary) generation. The embedding and transcription models stayed as is, using nomic-embed-text and whisper.cpp for embedding generation and transcription, respectively. It ended up improving my development experience, as I've been working on a MacBook Air for a few weeks. I can never seem to keep my free space over 20gb.
There are a lot of neat applications of these technologies with respect to audio and communication. Excited to see where this one lands.