- Would you like to do that to encrease quality - build architecture manually and then ask AI to refine and code it, add more automation testing with help of AI.
Tell why you would like to do that, the answers can be different.
With Anthropic's prices it soon will be cheper to hire real people then "hire" their models. If you use it for codding I'm pretty sure you will not be able to find any model which can be hosted on local machine and produce more or less acceptable results. Options - use something more cheaper, Deepseek for example, or if company big enough think about hosting some open source on you private DC with GPUs, I know that some companies develop their own solutions, but I don't have good enough expirience to tell how mach it may cost
Half a year ago it was really hard to use any LLMs except Cloud Sonet, all others produces un-acceptable results, did a lot of changes which I did not ask to do, could not modify more then one file at a tiem, etc. Sonet was the one which really helped to develop faster. Currently almost all major providers including Deepseek passed the barrier when use of them speed up the development. I'd say, depending on the project complexity of cource, but more important becomes cost of the model. I'm trying to focus currently on the process aroud it, not model itself. If you have clearly defined phaces with acceptance criterias and check lists on each, like solution design, planning, dev, testing etc. you can even use different models for different tasks to minimize cost
Any AI is is non-deterministic by nature, this is more about statistics, but not dedicated test cases. So, even if you swich to the model which is better in metric results you still cannot guarantee it will be better in your specific use case or do the same. What you can probably do is to automate your important test scenarious and run each every time you change model or system prompt
UI is one thing, but there is also UX, otherwise you may get perfect look, but very inconvinient to the users. If you are using AI, especially Claude - try Claude Design. I found it really usefull for myself, but kind of expensive
Many people now can build simple products and I expect we will see a lot of small new apps in the nearest future. Quality is a different question, but for small apps AI can build it pretty well. But if you will try to build something bigger - AI which we have currently still cannot do that. Who knows what we will get in a year, bat barrier definnetely will become less and less. So what becomes really important (it was always important, but know becomes kind of critical to survive) is how to shape your idea, position it, find clients, advertise, etc. This is what we as progrummers typically didn'd do, we were thinking that our nice products will find a way to clients by themselvs.
I had the same error that I'm violating ToS after payment but not for the new account, for account I used for half a year, but it seems it was some kind of error from Anthropik side, I just tested my cloude cli, it was working fine and I did nothing. Not sure your case the same but they have definetely buggy system
I work with AI a lot, and at this point of time this is very skilled junior, maybe middle, even Opus 4.7. It will not build really good product for you without proper guidance. At the same time - there no restrictions on project size, AI can help you to work on any project. But key point here help you, but not do that for you. You need to build your product developemnt process and educate AI how to use it. Follow all the best practices, including propoer planning, architecture, UI/UX preparation, and review/approval gates. After you have all in place you will see this developemnt becomes a bit slower then just hey AI, build calculator for me, but you will be able to work on bigger projects and actually scale. There some open source plagins available, for example you can look at Supperpowers plugin which help with process organization in development circle
I kind of agree with that, but this is because I started my carier when it was Windows 95 and I remember that contract between me and Windows. But if you will try to compare it version by version you will see that it was changing constantly, but maybe was not so visible before win 11. Core principles were untauched.
- Would you like to do that to encrease quality - build architecture manually and then ask AI to refine and code it, add more automation testing with help of AI.
Tell why you would like to do that, the answers can be different.