I think the social experiment is a cop-out used after it failed. If the PR was accepted, we'd probably see a blog post show up on HN saying that agents can successfully contribute to open source.
Most of the people I know with money are difficult to convince to spend it. e.g. rich people don't buy designer bags; poor people do. My wife makes all of our food; we do delivery or go out to eat maybe once every year or two. We have no recurring subscriptions (other than utilities). Our phone bill is $20 for both of us. etc.
We also live in an area where outdoor ads are banned (which tends to be the case in wealthy areas IME), and I block ads on our computers, so we rarely encounter them. Consumerism is gauche.
I think that's debatable, there's arguments like quantity over quality to be made, but I also think it's somewhat beside the point of "ad supported services are a favour to the poor."
Which is why a lot of things are moving to "pay w/ ads". Not only do you get paid twice, your ad space is more valuable because you've weeded out the people who can't pay.
I agree. I think the main problem is personalized advertisement that incentivizes companies to record as much data as possible. I'd prefer if they worked like they do in print magazines. Every reader sees the same.
Lets say I'm reading a laptop review. Show me adds from the laptop manufacturer or of websites that sell said laptop. People reading the review are likely in the market for a laptop so it makes sense to show it. At most you could probably narrow it down to the country so a German doesn't get shown a Best Buy ad but thats as far as I would go.
> Sales figures collated by The Game Business last month showed that U.S. Switch 2 sales over the holiday period were down around 35% versus the Switch 1's first holiday sales performance back in 2017. In the UK, a similar comparison saw Switch 2 lagging Switch 1 by 16%. Even in Nintendo's homeland of Japan, Switch 2 holiday sales couldn't match Switch 1, and were down by 5.5% over the year's final nine weeks.
> In France, 2025's final tally of Switch 2 sales was down by "over 30%" versus the amount Switch 1 notched up back in 2017
Switch 2 is the fastest selling console of all time right now. [0]
They sold more than 17 million units in less than a year. The Wii U only sold 13 million over its entire lifetime. The Switch 1 took 2 years to reach 20 million, and the Switch 2 will very likely reach that number in less than half the time.
Nintendo may have expected even higher sales numbers but saying that
I find that with more complex projects (full-stack application with some 50 controllers, services, and about 90 distinct full-feature pages) it often starts writing code that simply breaks functionality.
For example, had to update some more complex code to correctly calculate a financial penalty amount. The amount is defined by law and recently received an overhaul so we had to change our implementation.
Every model we tried (and we have corporate access and legal allowance to use pretty much all of them) failed to update it correctly. Models would start changing parts of the calculation that didn't need to be updated. After saying that the specific parts shouldn't be touched and to retry, most of them would go right back to changing it again. The legal definition of the calculation logic is, surprisingly, pretty clear and we do have rigorous tests in place to ensure the calculations are correct.
Beyond that, it was frustrating trying to get the models to stick to our coding standards. Our application has developers from other teams doing work as well. We enforce a minimum standard to ensure code quality doesn't suffer and other people can take over without much issue. This standard is documented in the code itself but also explicitly written out in the repository in simple language. Even when explicitly prompting the models to stick to the standard and copy pasting it into the actual chat, it would ignore 50% of it.
The most apt comparison I can make is that of a consultant that always agrees with you to your face but when doing actual work, ignores half of your instructions and you end up running after them to try to minimize the mess and clean up you have to do. It outputs more code but it doesn't meet the standards we have. I'd genuinely be happy to offload tasks to AI so I can focus on the more interesting parts of work I have, but from my experience and that of my colleagues, its just not working out for us (yet).
I noticed that you said "models" & not "agents". Agents can receive feedback from automated QA systems, such as linters, unit, & integration tests, which can dramatically improve their work.
There's still the risk that the agent will try to modify the QA systems themselves, but that's why there will always be a human in the loop.
Should've clarified in that case. I used models as a general stand-in for AI.
To provide a bit more context:
- We use VS Code (plus derivatives like Cursor) hooked up to general modals and allowing general context access to the entire repository.
- We have a MCP server that has access to company internal framework and tools (especially the documentation) so it should know how they are used.
So far, we've found 2 use-cases that make AI work for us:
1. Code Review. This took quite a bit of refinement for the instructions but we've got it to a point where it provides decent comments on the things we want it to comment on. It still fails on the more complex application logic, but will consistently point out minor things. It's used now as a Pre-PR review so engineers can use it and fix things before publishing a PR. Less noise for the rest of the developers.
2. CRUD croft like tests for a controller. We still create the controller endpoint, but providing it with the controller, DTOs, and an example of how another controller has its tests done, it will produce decent code. Even then, we still often have to fix a couple of things and debug to see where it went wrong like fixing a broken test by removing the actual strictlyEquals() call.
Just keeping up with newest AI changes is hard. We all have personal curiosity but at the end of the day, we need to deliver our product and only have so much time to experiment with AI stuff. Nevermind all the other developments in our regulatory heavy environment and tech stack we need to keep on top off.
Fair, however at some point of a companies size/spending the complexity of integrating with a SaaS becomes as large as the one to run your own open source tool.
Beyond that, and Im aware this is very much application/company dependent, theres plenty of SaaS companies that offer horrendous or no support no matter what you pay. We used to use splunk for monitoring and logging. Paid a ton of money because we were handling financial data and needed tracibility and reliability. We constantly had to put out fires that were caused by their unreliable platform. It was not a good experience.
Ultimately, we jumped ship to Prometheus. We pay a fraction of the price and spent less time on it.
> We realized 1 million context window is not enough to explain all facets of Klarna
> Every new thread of AI is the same employee starting from scratch again. First day at the job.
Agents are limited. With you so far.
> This week we get a demo of a vibe coded frontend that is more beautiful and easy to use than any ticket management system I have seen
Again, totally matches my expectations. Agents totally make pretty stuff that looks like working software.
I just haven't drunk enough management wine to connect the dots and figure out these facts support a jira replacement.
It also gets me wondering, if Atlassian leaned more heavily into AI (More vibe coding, more agents, more layoffs) would they have been able to keep the Klarna contract?
> Are you sure about that? Because thats exactly what Klarna is doing/has done.
That link does not say that they are switching away from a system that requires armies of consultants to implement.
AFAICT, they are switching away from Jira (Atlassian/confluence products). Those are not ERP systems.
Once again, I must point out that the these sorts of assertions reveal that the person making the assertion has never been involved in an ERP rollout, neither a big one nor a small one.
And, again, I reiterate, the only threat is to small players in the market, who don't have a community to hire from. Because to become a big player, you need to gain traction as a small player, and if every small ERP system can be replaced with an AI generated system, non single one is ever going to gain traction (Why pay $10/user/month for a basic system when you can have AI generate that for a once of fee and some employee time?)
Workday is not an ERP? Beyond that, they're effectively replacing major stacks of traditional SaaS tools with in-house ones. Considering the scale and complexity of what Klarna does and the regulations it has to follow across many different markts, I'd say its a valid concern. Now, I don't think SAP etc are going anywhere, especially in traditional businesses where most of the company is reliant on it, but it seems there is a way to do it.
That said, plenty of banks still run on mainframes and use COBOL.
Well lets see how it works out for them - they're ending the partnership for HR software in order to build their own, but they say they haven't built anything yet!
Our knowledge is constantly expanding, allowing us to build things differently than we used to. Modern cryptography, which makes things like multi-sig possible, is only a few decades old; it didn't even exist when the current banking industry was being established.
reply