shados

The dice roll prediction is about the way the prompt is setup misunderstanding how Jev works (they treat the confidence score as a probability score, which it isn't).

If you instead give it a list of probability for each number and ask it whats the probability of each number, the result will be accurate.

I honestly was worried for them. Now, even with all the clones, they generally still come up on top in price and latency, but "good enough" is often sufficient.

Guess the (investment) market has spoken.

Is this playing in the same space as the like of gvisor?

D1 is their bet there but considering hyperdrive, a real Postgres would be cool.

I also wish they expanded jurisdiction more. I worked at companies that couldn't use Cloudflare because of specific location requirements in contracts

The whole nerfing narrative puts in the spotlight now crazy supertitions come into being. The group think every day that everything is falling apart is crazy.

Yup. LLMs can do almost anything. Can they do it at the speed, cost and confidence of a model like Jev is a different story. In Jev output tokens are straight up free because it's not doing text token generations.

There's a bunch. I got pissed during one of the last GitHub outages and had an AI agent one shot a clone with PR, stacks, actions, and full compatibility with the git client running on Cloudflare workers and R2. Only limitations is it was https only because CF workers don't have tcp to implement and without an enterprise account.

Works great to this day. And I was far from the first.

This whole mess brought to like just now little moat there is left in the AI world.

Like, everyone talks about how you can just prompt a new CRM instead of paying for Salesforce.

But in this case, I raced with Fable and Sol and had time to build a full featured, fully functional CI/CD workflow system that has most of the features of GitHub actions (not the ecosystem obviously) and costs less than the GitHub runners per minute, while being able to scale to millions of workflows... And I had it working end to end before the outage was resolved and was already running my apps deploys through it.

It would take a little longer to do in a company with red tape and more investments in the GitHub ecosystem, and of course it only replaced actions. But it works.

There is no moat.

in the world of AI, forking is so much easier.

I was messing with a closed source tool from a vendor, and turns out they didn't support my use case (they were Mac only, I needed a Linux version. And I wanted it in a different programming language than they had used for maintenance and performance).

Pointed Fable and Sol at the binary and the symbols (most of the work was through http, to be fair), and a few hours later I had a fully functional version with additional features.

now of course the legality of that being closed source is iffy, but it was for personal use. In this case its open source. Need to run this on another platform ? You probably can get very far with a prompt or two.

Cloudflare is hit or miss there.

They move pretty quick so they tend to ship a lot of product very quickly that have a lot of potentials. But they are very much MVPs, and you'll hit a lot of limitations that aren't always obvious at first blush. And then it can take years before they mature, at which point they'll pour a ton of resources and they'll go from useless to top tier in like 2 months.

It's really hard to tell.

The Vite plugin for Cloudflare Worker was one of those. It was really rough at first and I wasted hundreds of hours getting stuff working as advertised. And then in a week they fixed all the bugs.

AI Gateway is another. We used it at first because it seemed to check all the boxes, and turns out it had tons of limitations and almost none of the features of other offerings, with very few updates months+. And then recently they basically added all the features it was missing in a few weeks/months and now its good.

Agent SDK too, it was always behind with outdated dependencies and missing features all other agent frameworks had. And then one day they started iterating super fast and now its legitimately good.

You can never tell reading the docs, so you have to try it, curse because you only find the limitations once you're running the thing for real in production, abandon it, then a few months later its everything you could ever want.

I have a love hate relationship with Cloudflare. Mostly love, but still.

SOC2 is just "the process we say we have, is what we do in practice". The process can be almost anything. Some auditors will push on stuff as "required", but they're often wrong.

But all it means in the end is you can read up on how a company works and have some level of trust that they're not lying (too much).

It makes absolutely zero guarantees about security practices, unless the documented process make these guarantees.

CC would explode even further if they had official Team/Enterprise plan (likely in the work, Claude Code Waffle flag), and worked on Windows without WSL (supposedly pretty easy to fix, they just didn't bother). Cursor learnt the % of Windows user was really high when they started looking, even before they really supported it.

They're likely artificially holding it back either because its a loss leader they want to use a very specific way, or because they're planning the next big boom/launch (maybe with a new model to build hype?).

Yup, but Article 3 point 2.a has a fairly strict definition, where for an entity outside of the US to be considered as "offering service" to EU members requires some kind of strict ties. The de facto examples is offering a product by specifically mentioning payment in Euro, or having presence on an domain with a top level TLD of a member state. If there's no ties that shows the offering is made to EU members, it doesn't apply.

Very very little tie is required (eg: just having one employee in the EU in a 50,000 people org would do it right there), but the law has been fairly consistently interpreted as such.

I get where you're coming from, but this isn't a if or but or theoretical. Its just how GDPR gets applied. I probably confused things by trying to introduce poor analogies, when the law itself is fairly clearly interpreted a specific way.

Its correct purely because of jurisdiction. EU laws don't apply for people with no presence in the EU, unless there was some kind of treaty where one country agrees to enforce another's.

That's just how laws, any law, works. The EU can "fine" all they want but it would be entirely symbolic.

That's like if US restaurants had to enforce EU food safety laws when on US soil because a EU citizen is eating there.

Fortunatelly, unlike US laws, GDPR, by virtue of being EU law, is actually readable by normal human beings, so its fairly straightforward:

https://gdpr.eu/article-3-requirements-of-handling-personal-...

Right, the point Im trying to make is that if someone has a requirement, they can build it on their own, for themselves, at much lower cost (because they don't need sales and marketing for themselves).

It was always possible (hire software devs to do it), but the bar and cost is much, MUCH lower.

Assuming they don't have an EU presence of some sort, EU law doesn't apply to them.

Now if they want to open up shop in the EU, or use a payment processor to charge money that has EU presence, things change.

Correct. But you can build this thing on your own, for yourself. LinkedIn didn't, and you don't trust this third party. So if you had the problem its trying to solve, you could just spend the 2-3 hours and build it for yourself, even if you don't have all the necessary skills to do so.

That's the whole point.

Correct, but the amount of people who can do it has drastically increased, and the amount of time it takes for most people to build these things has drastically decreased.

Yup. But my point is you can build this yourself, FOR yourself. So if you're not comfortable with using this one, you can build one on your own that you can trust (because you built it yourself).

Thats the whole point. In an AI world, you're no longer bound by the limits of what 3rd parties do or don't do, plus or minus some datasets (like in this case, the job postings).

It really shows how AI is going to change the entire industry.

Let's imagine tomorrow a Product Manager at LinkedIn wants to introduce this as an official functionality? They're going to have to run it by management or their pod (or find the PM in charge of that area if its not them), finish existing project, wait for resources to be ready, have legal/marketing/compliance involved, get it developed, go through all the other red tape, etc.

I don't know exactly how LinkedIn works internally, but I'm sure some of this is accurate.

So maybe, MAYBE they'll have it in a couple of months? But someone can build it in a few hours, even if they're not super good at this stuff.

It changes everything about how we think about products and SaaS software.

Yup. People act like they are geniuses there, but they were pulled kicking and screaming to allow native apps.

Honkai Impact 3rd does have pay2win ladder. It's pretty hardcore too, even as a whale it can be quite rough to get to the top. There's a LOT of top tier whales in that game and you need some pixel perfect timings to utilize it at the top level.

It's not just that. A large portion of IT people who work in these industries find Windows much easier to administer. They're very resistant to switching out even if it was possible and everything the company needed was available elsewhere.

Even if they did switch, they'd then want to install all the equivalent monitoring crap. If such existed, it would likely be some custom kernel driver and it could bring a unix system to its knees when shit goes wrong too.

You highly overestimate the capabilities of the average IT person working for a hospital. I'm sure some could do it. But most who can work elsewhere.

The backend for frontend pattern. Something most apps where frontend and backend are in the same organization (fullstack or otherwise) should have. It does wonder to maintainability and performance.

Even though we use GQL here, we still have a B4F, so it's browser -> B4F -> GQL -> Database.

The tooling we use make it trivial so it doesn't add any development overhead (often reduces it. No CORS bullshit for one), and our app goes ZOOOOOOM.

Not really. The usual strategy for persisted query is to only use them in production or adjacent areas. You build your app using regular graphql doing whatever you need, then after you tested things and it all looks good, when you ship, you parse the codebase, store the queries, and ship them.

It doesn't work if you don't have control over the client (eg: exposing to third parties), in which case query complexity limits are easy to implement and do the job, too.

A lot of the points in the articles are issues anywhere, it just depends on how well they are understood in the context and how mature the tooling is. Eg: rate limiting in REST is rarely problematic because we have no end of WAFs and gateways and circuit breaker libraries that understand REST very well.

It's very true that a few years ago the gap was there, and people implementing GQL had to solve all these problems without access to these mature solutions.

But these days, they're mostly solved problems. Authorization by type that propagate through the graph is pretty simple. Rate limiting and complexity limits on queries is often built in the frameworks. Query parsing is solved with persisted query all the major frameworks have as first class citizen. Performance: lots of GQL caching solutions out there, and the "multi threaded by default" model used by most frameworks helps a lot.

Etc etc etc.

I jumped companies a lot, did both. For a while I too thought GQL was redundant, then went back to a more classic world after a successful GQL implementation. I had forgotten how many problems were solved, especially organizational and people communication issues, through GQL. And so we're moving to GQL again.

Someone in the comment mentioned GQL is a tech solution for an organizational problem. I'll say most good software tooling and frameworks ARE tech solutions for org problem, because technology is usually a solution to people problem. GQL is the peek of that, and it does it quite well. Don't use GQL to solve tech problems, there's better tools for those.

Canada has/had that too. Eg: Montreal used to be a pretty small part of the Montreal metro until they merged all the cities on the island. Even then it doesn't include it's north and south chores.

Consider how the FAA handled the Nextgen project and continually gaslight anyone negatively impacted over the last 10 years, I would be against any airport built within 10 miles of where I live too. Not surprising people would be against it. It doesn't have to be so bad, but it is.

The wcag color contrast algorithm isn't great (and there is a proposal for a better one), but sufficient color contrast is necessary for a variety of visual conditions. That it's not good enough for some doesn't mean it's not more accessible than not doing it at all though.