Is this really something to be proud of? I make heavy use of AI, but for anything that is eventually going to production there's still a human reviewer. Based on this I assume Anthropic is shipping code with no human review? Is the future of our industry going to be oneshotting prompts with out manual code review and then having AI fix the bugs found in production buy customers? I'm not trying to be a Curmudgeon, but am I missing something?
They review based on the intent of the change not jz blindly.. I think it's mentioned in the article where the job of a developer is to provide high level design decisions and saying no to claude so that it doesn't over engineer solutions.. As the demand for good, but not perfect, software is the bottleneck for anthropic and access to better internal models, I feel that's good enough for their workflow as they've started repeatedly that this is just suitable only for them and they want to share their knowledge like a good software company does.. you talk as though code reviews were not a joke in lots of companies where only specific teams/companies does detailed reviews.. many teams/companies jz blindly approve based on tests passing or just do a high level overview check before merging..
Where do you draw the line, and where does someone take final responsibility? I don't have an issue with allowing AI to do everything with limited to no human verification for hobby projects. I have an issue with allowing AI to write and ship medical device software with no human review. What's the middle ground?
You still take the full responsibility, I don't think leveraging AI reviews removes any layer of responsibility. The challenge is building the feedback loop (or the agent that builds the feedback loop) to make the review process actually catch what it needs to catch.
I think the line is wherever you feel comfortable with pushing the changes yourself. You feel like adequate testing had been done, basing it on previous testing periods or how successful AI runs have been.
Whether you or AI wrote the code, you decide when it's safe enough to put your name on it. Like it's always been!
Are there API's that provide simple information on if it's a good night for astronomy? I am totally blind, but would like to know when it's a good night for my kids to use there telescope. I don't use a Mac as my primary machine, and xasteria is not accessible.
Hi jareds - your reply prompted me to look at what accessibility options are already in Nightwatch, as I'm ashamed to say I'd not done enough in this respect.
Keyboard navigation, the ability to change text-size, and voiceover support will all be in the next version.
Don't ruin this for me. As a blind developer WSL is amazing. Windows accessibility is significantly better then Mac in my experience. I use WSL, everyone else uses a Mac, and all the same tooling and scripts work identically.
As a blind software developer this looks interesting, but the output descriptions are not useful. I asked it "create a diagram for a rest api with a database, app server, and cache." The diagrams alt text was "609 / 100k tokens used" The text it returned was "Created a simple REST API architecture with three core components: a web app server, a cache layer, and a database. Organized them into logical tiers." If it can generate the diagram it would be nice if the alt text was a detailed description of the diagram.
Great point, I hadn't considered this use case. I've added richer, screen-reader-friendly diagram descriptions to the roadmap. Thanks for the feedback!
I also am looking forward to Apple smart glasses. I use Meta glasses because thre useful since I'm totally blind. I'd much rather have on device recognission though if possible. I also trust Apple more then Meta. At least I'm technically enclined enough to realize I shouldn't wear them in the bathroom or bedroom. At least when humans look at my AI prompts there maily seeing food boxes or computer bios screens.
Ever since Be My Eyes introduced there AI features it's my understanding that there's been a lot less need for volunteers. I'm totally blind, and started using the app once they added in AI. It works great for for things like reading food labels after my kids have movved things around, determining if the tv has been left on, etc. I think I would use the volunteer feature if I still lived alone, but I don't.
Thanks for creating this. I'm blind so the $16 USB tester off amazon to sort through my drawer of cables is not an option. This will stop me from needing to buy a sbc just so I have something running Linux to test cables.
Curious whether you've tested the popover with VoiceOver yet. Menu bar SwiftUI apps usually nail the status item by default, but the popover content tends to be hit-or-miss depending on row layout. A stack of Text views walks cleanly, but custom-drawn rows often read as one blob. Worth confirming for jareds' use case.
What's the current situation for coding with Local LLM's on decent hardware? I have an M3 Max with 64 gb of ram and am thinking I should start looking at Ollama and Opencode? Is this a useful stack for smaller personal projects?
It’s getting there. You could give a try with qwen 3.6. It’s worth paying for better models in the cloud, but local models are now better than nothing.
One nice development recently was ollama's support for MLX optimization on Mac hardware. It's not obvious how to know you're using a model that works with it, yet, so it's rough around the edges.
I wasn't planning on buying one, but I'll add this to the list of app enabled coffee tech I refuse to buy. As someone who's blind I'm getting really tired of app enabled coffee equipment with no open source integrations or protocol documentation. Fellow also doesn't appear to make any effort to make there apps accessible. They have had there Aiden out for over a year and I still don't see any notes about accessibility in there app update. I'm not going to buy one and use the home assistant integration since that could break at any time. Luckily I'm more of a coffee drinker instead of espresso so the Ratio Four works well enough for single cups and half pots.
How long until the $10 Github Copilot subscription goes away? That was a great deal for my limited personal programming. The only reason I switched from it to Claude was to get coding and general ai in a single bill.
I think Github Copilot is in the process of slowly winding down right now. They've been putting very, very long (multiple day) rate limits on users for various esoteric reasons for weeks now and just yesterday or so paused signups.