Awesome. I had a crack at the Xerox Travel Scanner 150, which had niche drivers for some reason, only shipped Windows and didn't make it into SANE. Wireshark and prior art from SANE got me most of the way. Getting the stepper motor working was really hard and I didn't manage to get it completely working Maybe I'll resurrect 15 year when it's a little older. I really fancied a Pi+Scanner pipeline with OCR/AI to file all my documents as they arrive. Maybe I should just take a photo?
My biggest worry is the small talk that you casually have in meetings. Comments about your pattern of life, family, locations, friends and health. Slurping all that up over 100s of meetings is concerning. Stored raw transcripts of meetings is a huge liability. Should redact small talk and only store useful extracts.
I think this is because all frontier AI providers integrate with GitHub as a default git forge through their GitHub apps platform and their free tier is too generous. They should either charge some of these huge partners access to GitHub apps or partition the free tier off and have a different SLA.
If the frontier AI used GitLab as a default I'm sure they'd be the ones suffering now.
I've been excited about OpenRouter's OAuth 2.0 PKCE flow for backendless BYOK inference. I just updated my Garmin watch AI app (https://untether.watch/en/apps/byok) to run the whole flow on the watch for a seamless BYOK experience - formerly you had to copy/paste your Gemini key into settings, which was a huge barrier.
If you're building client-side/frontend apps and want to let your users BYOK, OpenRouter's PKCE flow is great for that.
Otherwise I'm still working on Untether (https://untether.watch) - a suite of digital-minimalism apps that let you stay connected and do quick actions on the watch while keeping your smartphone out of sight (and your hands).
Nice idea. I recently published an article with a different twist. Static Vite+React site but all the backends are via OAuth PKCE and your customers bring their own. https://type2fun.net/infinitely-scalable-personal-apps I like the idea of building apps but skipping the infra burden/costs.
Author here. Built some personal apps using OAuth/PKCE for auth, hosted as static sites. OpenRouter adds the AI. No backend, no database, no growing bill, minimal maintenance. I believe these frontend "Pixie apps" are a great architecture for personal tools you can actually share. Keen to hear what people think.
For me it was reverse engineering an obsolete travel scanner (should have just taken a photo ) and automating a document workflow. Something I knew was possible but impossible for me.
This pulled me into BLE protocols and private APIs and integrating with things directly instead of using 3rd party apps that were bad or limiting.
I'm currently fighting Garmin's wonderful Connect IQ watch app platform (it's horrendous).
I'm working on <https://untether.watch>. Trying to shift 20-30 micro phone interactions to the wrist per day to ultimately reduce phone use. Dumbphones are too extreme - you need a smartphone for certain day-to-day activities (banking etc.)
The watch is a great form factor - it's got a crap screen (MIP), the ergonomics are awkward (rotate and look down), it has limited capabilities. But that's the point! Do essential quick actions and leave the phone out of site.
Requires Android companion app to do the heavy lifting. Use the (head)phone mic and STT to reply to any android notification and make notes. More features to come.
Garmin's SDK is seriously challenging. APIs are often broken across firmwares, limited developer tools and testing is tough.
The issue is every AI coding tool integrates as a "GitHub App" (OAuth, PATs, webhooks etc.) first, over other code forges. This load is coming through their 3rd party app integrations. I bet the web/git volume isn't getting smashed as much.
I had to begrudgingly use GitHub over my preference GitLab to use some 3rd party AI features.
The solution for GitHub is to charge or rate-limit some of these 3rd parties integrations and come up with an equitable solution.
You can make standalone Garmin Wifi Apps for free (you only have to pay if you want to charge).
Do note that it's very janky. By default web requests proxy via the phone - if you want to try wifi you have to switch into a "modal wifi mode" - do a batch of web requests and logic - then switch back into "normal" mode and then do something with the saved data.
You cannot push notifications from wifi but you could do:
1. request to open the app using a background service
2. toggle wifi mode
3. fetch the data and close the app
4. create watch notifications from the background service.
You may need to acknowledge the request to open the app with a button press - but I've also see examples of it just opening (not sure of conditions though).
