> Amazon and Google have both been accused of similar practices, but they agreed some commitments rather than having a formal decision against them

Amazon is still full of fake reviews, it does not take a genius to spot them. And nothing happened. What would be the outcome of the investigation? legal paper saying we are committed to do better?

I hope it occurs to Google that they can use this feature to update the design of their websites and services

I wonder if they have been, the bugs in their UIs are increasing in frequency

> Gartner analyst Dennis Xu has half-jokingly suggested banning use of Microsoft’s Copilot AI on Friday afternoons, because he fears at that time of week users may be too lazy to properly check its possibly offensive output.

Aka: clickbait

Kai is hilarious, and what makes his videos really funny is that they are 100% true about what software engineering looks like in real life.

Another channel that I like a lot is Alberta Tech, which falls in the same category.

Posting a product on any of these sites will not have the same impact as it did before AI. Not because your product is not good, but because there is much more noise now.

This applies to social media posting, SEO, articles, you name it. AI has amplified the noise to the point where finding something useful is pretty hard now.

Building in public is and was always a fake trend. You see a few who made it a long time ago by posting their journey (personal choice), and then everyone jumps in to spam, which is back again to the noise, ending with a lack of value.

I feel for anyone trying to take a product to the market right now, while there are more tools to build, marketing has gotten a lot harder, consumers are struggling financially, and companies are trying to stay afloat due to a lack of growth.

It did not work even before AI. The rise of "indie hacking" in the late 2010s brought in thousands of hustlers creating similar lists, and many of them were simply selling shovels to other indie hackers (including the lists themselves). By the time of the pandemic, the "submit to every directory & community" strategy was already useless.

I've been creating and marketing software as an indie developer for over 20 years now, and the marketing part definitely feels harder than it use to. See also:

https://successfulsoftware.net/2025/12/22/is-the-golden-age-...

True story, yesterday I tried to get some feedback from an industry relevant subreddit for a real estate quick check calculation tool (automatically extracts listing data into calculation and enables sharing investment ideas). The pure mention of AI brought up a whole crowd of fed up bullies that talked it down as vibecoding trash - which it really isn't. All those places are flooded.

People, not bullies. I can sympathize with you because I've struggled with the same, but we can't blame those people. They're now being asked every two days to give feedback on yet another tool. That used to be once every 6 months. And the overwhelming majority of those new "tools" is abandoned within a month. And there is indeed a huge amount of vibecoded slop. I've put more time and thought into our product than the last 20 such tools that got posted into our industry-relevant subreddit combined, but I can't expect the mods and users to put their time into assessing that.

I think it’s strange to dislike vibe coded things. I’ve seen a lot of cool stuff that’s mostly vibe coded. fomo.nyc for example. The problem is mostly the intention. I think a lot of vibe coded stuff isn’t solving a problem someone has its someone trying to seek profit. It’s no different from when smartphones first came out and people wanted to make a for everything when most of them didn’t solve any problems. The difference is nobody is wowed anymore by anything so your app that turns your phone into a beer kind of thing doesn’t exist in the vibe coded world.

> I’ve seen a lot of cool stuff that’s mostly vibe coded. fomo.nyc for example.

Maybe because you're not an actibe member of one of those fields whose subreddits now gets these vibecoded tools every other day. Because for those, I can tell you that the overwhelming majority isn't cool. Even when they are trying to solve a problem and aren't (yet) seeking a profit. They have very little time, energy and thought put into them. They're made by people who are passers-by, who aren't personally invested and often haven't experienced the problem first-hand; it's just a problem they heard about, or assumed would exist. And that leads to things that are a waste of time. Often they have blatantly obvious problems that show they didn't even QA it for an hour.

And to top it all off, they frrquently even vibe code their reddit posts. There is absolutely nothing of interest to interact with those posts.

The amount of reddit posts I land on from the last few years which are noticeable longer with no added value for the increased length than they would have been in the past is getting very annoying.

While this is directionally correct it does come down to a tonality, that I think wasn't justified in that case. But hey, it's the Internet and I'm not naive either.

I'm saying that the tonality is justified due to all of the other slop. You and me are simply the unavoidable casualties.

I'm saying this while in a field that's even more anti-AI/tools than yours, FWIW. By virtue of anything that mentions AI - and much of the tools that don't even mention it but are suspected - getting auto-removed by automod on the subreddit. There's only one subreddit in our field, everyone's on it, and it blanket bans anything with AI, despite the best tools out there incoporating it.

It's not even something like art, design, coding and so on where people are scared of job loss or even hobby loss, it's nothing like that at all for the community. I do suspect the mods' friends might feel threatened, as they _do_ make a living off of the community, but in our case I don't even think that's their primary reason to blanket ban it.

Yes but your content is also part of the flood

Unfortunately the source doesn't matter when there so much. It is really hard to differentiate things when you are inundated. Did you try a Show HN here? It requires more luck than ever because of the same problem, but worth a try. I'll take an honest look if you do it (though hard to say if I am the target market).

Had basically the same thing happen. Posted in a side project sub, spam filter nuked it because new account. And in other subs now, anything that mentions AI gets hit with "vibecoded slop" automatically. Doesn't matter if you spent months on it.

What actually moved the needle was talking about data, not the product. I posted about my tool — crickets. Then I wrote about stuff I discovered while building it and people started engaging. Exact same product behind both posts, just "here's what I found" instead of "here's what I built." Night and day difference.

The producthunt noise was a thing on day 1 I feel like, I quickly stopped checking that website after they launched

the law of shitty click-throughs strikes again

What's unique about Vertex's privacy policy?

They don't read the things you send them, not even for "safety checks" or sys-admins accessing the system. Totally opaque (as it should be).

Keeping chat content around for 30 days might as well mean "forever." Anyone at the company can steal your customers chats.

My agreements with customers would prevent me from using any service that did that.

The problem is well articulated and nice story for both cofounders.

One thing I don’t get is why would anyone use a direct service that does the same thing as others when there are services such as openrouter where you can use the same model from different providers? I would understand if your landing page mentioned fine-tuning only and custom models, but just listing same open source models, tps and pricing wouldn’t tell me how you’re different from other providers.

I remember using banana.dev a few years ago and it was very clear proposition that time (serverless GPU with fast cold start)

I suppose positioning will take multiple iterations before you land on the right one. Good luck!

Hey Oras, thank you for the feedback! I think we definitely could list on OpenRouter but as you point out, our end goal is to host finetuned models for individuals. The IonRouter product is mostly to showcase our engine. In the backend, we are multiplexing finetuned and open source models on a homogenous fleet of GPUs. So if you feel better or even similar performance difference on our cloud, we're already proving what we set out to show.

I do think we will lean harder into the hosting of fine-tuned models though, this is a good insight.

Hi,

I'm wondering, how can one host a fine-tuned model on your platform?

I wasn't able to find any information on how to do that

Really weird, you'd think all API calls are tied to the user ID. Unless some hard-coded debugging ids crept out to prod. But again, code reviews? integration test?

Recently Lloyds Banking Group has upgraded their apps so that you can see details from all your accounts with them in every app (some functionality still needs the right brands app though).

There's obviously some magic to glue different accounts together without user input, so I can see my Scottish Widows pension and Halifax credit card balances in my Lloyds Bank app[1], even though I have separate logins for each brand's website and app. Possibly my National Insurance number and address I'd guess? But I can see it going badly wrong if they get the magic merging wrong...

[1] Clearly I need to open a Bank of Scotland account to get the full set.

Technically, not much magic, apart from Mortgage and Pension, OpenBanking is quite solid in the UK, and you can connect any bank account from any bank. This is now supported by many banks in the UK. You can connect all other bank accounts directly. There are apps for that, too, for a small fee.

But I suppose you mean connecting these directly based on the national insurance number, rather than relying on user consent to use OpenBanking to see them all together in one place.

Yes, this isn't OpenBanking, I'm sure the Lloyds app could do that at one point to link to my Halifax account (or others), but it was flaky IIRC (and seems to have disappeared from the app?), this is a separate automatic thing.

This is probably in response to the new App first banks like Monzo/Starling (I use Monzo) Where you can open an account, see savings, current account, business account, CC etc all in one.

My daughter opened an account using the app only (Scan passport, video clip to authenticate) got the card the next day and setup up savings, investing ISA's with a few clicks. Again incumbents not willing to innovate until someone disrupts, but probably too late.

You could see all your accounts with an individual brand in one website / app, and doing things like opening new accounts online or applying for credit cards isn't exactly new functionality either. I'm sure Monzo is slicker because it isn't talking to a bunch of COBOL in the backend, but traditional banks aren't behind so that basic functionality is missing.

The issue is that Lloyds Banking Group was basically bifurcated by brand, so that the Lloyds app and website only had your Lloyds current / savings / credit cards etc. whilst the Halifax website and app only had Halifax current / savings / credit cards. They're clearly trying to merge things so you only need one app / website.

I noticed my soon to be closed nearest Lloyds branch gained a "Halifax customers can use this bank." poster in the window, I'm not sure if that's just a reaction to the fact that physical bank branches are getting kinda rare, or if it's really taken them about 17 years to merge the Lloyds and Halifax backend systems to the point where branches are interchangable.

Yes the backend for Monzo is indeed very performant. I have a dev app connected using their api and can read all my accounts/transactions over an API. As soon as I make a transaction it appears in my web app within a few seconds. Compare that with older banks it can take a day for it to appear.

Maybe some session sharing issue? Wouldn't show up if you weren't testing multiple users at the same time.

Yes, hope they publish a post mortem of the cause, will be interesting to see how this could have happened.

"Judgement" is perfect for now. I use CC and Codex every day, and I agree they spit out something that will work on the first run, but they go wild about some system design choices. For now, I agree with you that judgement and system design/architecture is the distinction.

What I worry about and think will happen is we will see emerging platform that does full systems, and replit is a good example of that. I don't think these tools are enterprise ready yet (I might be wrong), as enterprise is always about integration with other systems including legacy ones and data silos.

I think we will be like COBOL developers now; we will maintain pre-AI systems until we retire.