If you use wisprflow or another active voice / transcription app… what do you love and hate about it? Just coded my own local solution that works, diving into the UI next. What features would you want to see?

I built this because it was getting really annoying opening and reading Markdown documents and all the tiny little frictions that just take a few seconds here and there but done hundreds of times per day add up and get annoying. I kept navigating through folders in Finder (or VS Code, or whatever tool just happened to open it). And for some reason nobody can seem to render markdown correctly. I kept having math symbols not render correctly, or only 2 columns of a table show up and ugly horizontal scrollbars to see there rest of the table when there is plenty of room on the screen to render it, or it open in a window so small I had to resize it to be larger and increase the font size every time it opened because it didn't remember the last setting. I finally had enough so I made a list and built the perfect app for me. - It watches the folders where agents work - Surfaces documents as they are created or modified - Keeps an active set of documents stable and easy to switch between - Refreshes the document and keeps scroll position as the agent updates it - Optimized for fast and keyboard only navigation (don't worry mouse works just fine too). The goal is simple: spend less time managing Markdown files and more time understanding what is in them. I'm releasing it open source, MIT. Feel free to grab it, customize it, suggest improvements, etc. https://github.com/brennancheung/bmd One thing that is useful is that when an agent creates a file and it's in a folder you've added to the sidebar it automatically shows up. Or you can just tell the agent to call the CLI of the app and it will open it directly as well. So it's really optimized specifically for the best user experience when the agents create and edit markdown files. No editing. It's not meant for that use case. The entire thing was vibe coded in less than a day, and mostly in the background while I was working on other stuff. It's really impressive what you can build now.

I created a software development process using ICM. I noticed as Claude works through issues, it will write stuff to it's memory files, so it's available for other sessions. But I'm thinking, I'd rather have it update my workflow instead, in case I need to swap agents. Has anyone run into this and how have you solved it?

Everyone is building at AI speed right now. That is fine for most of the app. It is a terrible place to freestyle authentication. A half-baked login looks "done" in a demo and fails under real users: broken sessions, weak password flows, MFA bolted on late, multi-tenant gaps, cookies that only work on localhost. The model will happily invent a sign-in form. It will not invent a security posture. This is where a tool like https://clerk.com earns its seat. Not because it is magic. Because it stops you inventing the one surface that should never be improvised. Clerk is an excellent Auth skill and for new SaaS, prototypes, and startups where speed-to-market is the metric and the stack is modern frontend. It is a poor tool when at consumer-scale unit economics are the product constraints. This is not "never build auth." It is "do not invent the front door while the rest of the app is still a sketch." The principle: Velocity is only a win if the you don try and speed run the boring bits. //A<3

My Cousin @Ryan H. was tasked with making me a new logo for the group (the old fashioned way, not much AI help here) and I think he SMASHED it out of the park. Perfect nuance, what do you all think?