michaelbuckbee

I kept making a hacky eval tool to try and compare the outputs of different models, model configs, prompt versions, etc and finally rolled it up into a web app + downloadable app (kind of like Postman or Insomnia, but for AI).

Free, holds your keys in localstorage and makes direct calls to the APIs (unless there's a CORS issue), at https://evvl.ai if you want to try.

It's great! But what makes it even more so is that it's not a time capsule, but still updated. There's pages for Alien Romulus, the new Predator movies and more.

The Wright Brothers memorial in North Carolina is a very special place. They've renovated it somewhat recently and it now conveys a lot more of the struggle that they went through.

So much debugging of prototypes, crashes, redesigns and high stakes testing.

There's also something undeniably cool about standing right where other humans did something for the first time did something and walking the distance of their flights on the field.

https://www.nps.gov/wrbr/index.htm

Depends on what it is, the quick bash script I write to verify something or the core of SaaS that's being launched have different levels of strictness, style, testing, etc.

This seems very cool, but I'm not sure of the utility of using CF workers for it? Is that just a framework or is this being deployed to the internet?

To try and overcome my own personal tendency towards negative criticism on HN, I try and reframe my comments from "why this won't work" to "how can we make this work".

Not a panacea, and I'd be interested to hear others ideas on how to better comment and give feedback.

Late to the party, would love to know more of your workflow with how you're using beads.

I had an odd experience that rhymes with this, where I was trying to figure out where I had seen the actress Julie Dreyfus (from Kill Bill), whose Google results are utterly crushed by Julia Dreyfus (of Seinfeld and Veep fame) despite having demonstrably different spellings to their names...maybe it was that simple.

I think it might be better framed as "college doesn't seem like it's worth it".

Some of this is back of the napkin math of degree costs having risen so much, but also likelihood of using your degree or getting a professional job having fallen.

The only "real" justification is that this is a long term play to figure out a UI and interaction shift that will work for general augmented reality devices (aka whatever device Apple releases five years and two iterations from now based on the Vision Pro.)

I think this is one of those things where if you're married (like me) you only have the most peripheral sense of the popularity of these things and if you're single they potentially occupy way too much of your thoughtspace.

I run several add-ons in the Heroku ecosystem and having dipped my toe into both Laravel Cloud and Vercel+Next...it's hard not to look at the generalized Heroku ecosystem and see it suffer in comparison.

I worry that Heroku is now in an uncomfortable middle position between much easier to use systems and much more enterprise devops systems like AWS.

As a long time Ruby dev it's really hard to separate from Rails...and there has been both drama (which is ok to an extent, you want passionate people to care about things), but it's unfortunately led to some severe stagnation on the part of the Rails ecosystem versus rivals.

At a time where other frameworks (thinking mostly of Next+Vercel and Laravel here) are merging their frameworks with hosting, deployment, services and ops the Rails ecosystem has gone the other way and tried to bring back "you can host it yourself on a VPS!".

While there are merits to both approaches, the market really seems to have spoken and it prefers the integrated approach.

Analog purchases have become much more of a signaling mechanism than for direct consumption.

In my family group there were a good numbers of vinyls gifted this past christmas and none of them are going to be regularly listened to as the majority of music consumption they do is "on the go" in the car or mobile.

Similarly, I'm seeing them make more purchases of "trophy books" where they read the book on their phone or listened to the audiobook but liked the book so much that they want to have it on their shelf (there are also special editions with elaborate edge decorations, etc. that seem to feed into this).

The lack of director commentaries and behind the scenes content on streaming has always baffled me as the rights to that must be much cheaper to acquire and would result in more minutes of streaming watched for less licensing money.

SQLite is known for having a lot of testing. Per their docs around 600x as much test as application code.

https://sqlite.org/testing.html

The "garish" statues are more akin to a false color image of mars that shows topography or something. That they're a visual representation of a particular portion of the pigments found and are not supposed to be an accurate recreation of how the statue looked at the time it was created.

A lot of the comments here feel like they're disappointed that this is a "Docker with unnecessary LLM crap thrown in" when I think what they're really going for is more "LLM workflows with a higher degree of observability and sanity".

I think a more interesting point of comparison is the Claude Code Github Action, Co-Pilot code reviews, etc.

Absolutely, if nothing else it's a fantastic entry point for new people to watch the show (especially kids).

Note also, there's another season that's on the streaming services as individual movies post the Netflix reboot with much of the same team.

Fantasy series seem like they've gotten longer and longer and it's often years between volumes. Many authors have started doing recaps of their previous books at the start of later volumes, but not all.

I could see this being useful for that.

"Comeback" is probably the wrong word, maybe "uptick" in usage?

Worktrees just fit particularly well for the scenario of developing multiple different features in parallel on the same codebase, which is a pattern that devs doing a lot of AI assisted development have.

(not a joke) I wonder to what extent the ability to produce a Rush Hour 4 will effect the deal.

https://www.cnbc.com/2025/11/25/trump-pushed-paramount-reviv...

The usage of worktrees is seeing a big comeback in the era of AI assisted coding.

I have a script that takes Github issues and spins them out into their own worktrees with corresponding stack.

I can then run individual instances of Claude Code in each and easily flip between them.

Consider Google's search results page (setting aside the ads and dark patterns for a moment) as a form of generative UI.

You enter a term, and depending on what you entered, you get a very different UI.

"best sled for toddler" -> search modifiers (wood, under $20, toboggan, etc.), search options, pictures of sleds for sale from different retailers, related products, and reviews.

"what's a toboggan" -> AI overview, Wikipedia summary, People Also Ask section, and a block of short videos on toboggans.

"directions to mt. trashmore" -> customized map of my current location to Mt. Trashmore (my local sledding hill)

Google has spent an immense amount of time and effort identifying the underlying intent behind all kinds of different searches and shows very different "UI" for each in a way makes a very fluid kind of sense to users.

That's really neat.

Even there, didn't they recently make some changes to the CS go skins ecosystem to devalue much of the aftermarket sales.

It's not going to pay off for everybody, this is a land grab for who will control this sort of central aspect of AI inference.

Amazon also uses Claude under the hood for their "Rufus" shopping search assistant which is all over amazon.com.

It's kind of funny, you can ask Rufus for stuff like "write a hello world in python for me" and then it will do it and also recommend some python books.

David Simon's earlier work, "Homicide" had a lot of interesting switching between film + video and aspect ratios as well. I think it's something that he's been interested in for a long time.

I really like how you framed this as the takeaway or learning that needs to happen as what should be in the alt and not a recitation of the image. Where I've often had issues is more for things like business charts and illustrations and less cute cat photos.