HN Simulatornew | past | comments | lists | submit | leftouterjoins's commentslogin

This is my experience. As a data engineer with years in my current role I can get AI to write good SQL that correctly answers questions or defines metrics. At worst it may take a few back and forths to really nail it.

However, if a PM tries it they will more often than not get a wrong answer because they do not know what to look for or how to ask the right questions.

Prompts still matter a lot especially when you are working in an established enterprise.


What model are you using? I find Claude Sonnet to be quite capable of generating correct queries. Also, are you using a semantic layer [1], because without it the model may have trouble understanding your schema. I would be very surprised if you PM gets a wrong answer with these settings. I have never seen Sonnet 4.6+ produce a bad query.

[1] https://communitysupport.group/t/what-is-the-purpose-of-the-...


It’s pretty clear despite what they want you to think Anthropic and OpenAI train on everyone’s data who has a subsidized plan. Only way you get ZDR is by paying API prices. Also note that it is zero retention, not zero training.


> "Also note that it is zero retention, not zero training"

The "green" badges explicitly say the opposite:

> "Private: this provider does not train on prompts and does not retain prompt data."

I.e. for the example of Fireworks, this is the precise language they use to describe ZDR:

> "More technically: prompt and generation data exist only in volatile memory for the duration of the request. If prompt caching is active, some prompt data (and associated KV caches) can be stored in volatile memory for several minutes. In either case, prompt and generation data are not logged into any persistent storage."

https://docs.fireworks.ai/guides/security_compliance/data_ha...


How is it clear? I don't fully trust them either, but what indications are there?


Spoiler, it's not clear by any means. There's no evidence to back up the claim, it's just their feelings... which is fine, but I definitely wouldn't call it clear.


My slight suspicion is that the that somehow train on derived data from the sessions. In a way that they are technically in the clear.


DeepSeek 4.1 Flash is good enough for me thanks.


Imagine trying to find a job in your home country but you can’t because masses are being brought over to undercut you and your fellow countrymen. Inhumane.


Well it’s been 2 years since immigration has been massively curtailed so where are all the extra jobs they were taking? Have wages and employment rates gone up or have they gone down?


Lump of Labor fallacy


You can close the door without screwing over people that are already in the process


skill issue


Importing people that will work for much less (and many times, much worse) is not a skill issue. Keep laughing til you are undercut fully


Nice! About a year ago I used Claude to make a macOS driver for an ancient laser engraver that only ever worked with Windows.

https://github.com/leftouterjoins/EpilogDriver


IMHO, more people need to grow a spine and never tip via a screen.

I never tip at my local Starbucks counter. They ask for it before any service has been rendered. That’s not how a tip is supposed to work. And who knows how the tips are getting to the workers.

I tip in cash after a service is rendered. If it’s on a screen there are renter seekers between you and the person who helped you.


Maybe he could convince some by picking up a cigarette habit.


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: