Skip to content

Comment on AI Does $58k of Engineering Work for $15, but Enterprises Can't Use It

Comments

Meanwhile, McDonald's just killed their IBM Watson drive-thru because it kept adding bacon to ice cream orders. The AI worked fine. The architecture was broken.

It is killing me that this statement isn't followed-up on. I can't find any more technical details about this online.

Come to think of it, the whole article is like this. Broad statements without any details to support its claims:

A team I know needed a code review tool last week. Instead of evaluating vendors, they described what they wanted to Claude. Four hours later they had a custom application integrated with their workflow. Cost: $3.40.

That's cool, I'm glad they could vibecode something in a week. What does it actually do? How well will it work next month, next year?

In March 2025, Claude 3.5 completed what amounted to $58,000 of freelance SWE tasks for about $15 in API credits. Real engineering work, benchmarked against OpenAI's SWE-Lancer.

Is there a link to a source on this? It'd be cool to read more.

This entire blog post, like the one before it, feels like a big ad for ... Prava, I think? I have no idea what they do. (And the fact that they continually cite DOGE isn't inspiring confidence.)

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.