Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

We rewrote JSONata with AI in a day, saved $500k/year

reco.ai

Post reply on HN
Print view View original

271–278 of 278 posts

Re: We rewrote JSONata with AI in a day, saved $500k/year

#271

Post by hu3 »

chrismorgan wrote: post #257
kjksf wrote: post #247

Earlier quoted context omitted.

> port it in less than one day There's confidence and there's barking mad delusion. Here's the reality. I once ported 50k loc from Java to Go. Here are details: https://blog.kowalczyk.info/article/19f2fe97f06a47c3b1f118fd... Java => Go is easier than JavaScript => Go because languages are more similar. That was a very line-by-line port. Because I was paid by hour I took detailed notes. I spent 601 hours to port it. 5…

I stand by my estimate, also having done interlanguage ports. I’m not saying any project of such size could be ported in one day, but this very much looks to be one of the more straightforward ones. Being a small project also accelerates things, as there are far fewer moving parts, concepts, &c. to keep in order. I wouldn’t say that Java → Go is inherently easier than JavaScript → Go. There are more features in JavaS…

I'm with the other commenter. There's no way to port 5k lines in a day with confidence unless using LLMs + strong unit tests.

I won't even ask for an example of otherwise, but feel free to provide a repo where a human did that.

Re: We rewrote JSONata with AI in a day, saved $500k/year

#272

Post by sarchertech »

spiderfarmer wrote: post #266
sarchertech wrote: post #190

Earlier quoted context omitted.

If you’re taking the time to review the code, none of the the above applies to you.

Yeah, but I'm still using AI to do the things you seem to think it's not able to do.

Your human crafted go fast and break things code is not equivalent to what I was talking about.

Look at OpenClaw for an example of what I’m talking about. That’s what happens when you let AI go wild without review or understanding. It’s half a million lines of code and it’s 2 months old. When AI adds a new feature it has no choice but to add and change tens of thousands of lines of code because it’s a spaghetti nightmare.

AI cannot refactor OpenClaw without making massive functionality changes. It might be possible for a human to spend a very significant amount of time guiding it to clean it up. But if you aren’t reading the code in the first place you ain’t gonna do that.

Without human guidance and review AI will absolutely create mountains of slop that are beyond its ability to maintain.

Re: We rewrote JSONata with AI in a day, saved $500k/year

#273

Post by collinmanderson »

macNchz wrote: post #216
xnorswap wrote: post #171

Earlier quoted context omitted.

Perhaps it's specific because it's Opus 4.6, released February 5th. https://www.anthropic.com/news/claude-opus-4-6

Opus 4.5 to 4.6 was pretty incremental, I didn't see much of a difference. The big coding model moments in recent recollection, IMO, were something like: - Sonnet 3.5 update in October 2024: ability to generate actually-working code using context from a codebase became genuinely feasible. - Claude 4 release in May 2025: big tool calling improvements meant that agentic editors like Claude Code could operate on a notic…

Yes, for me I think it was around Nov/Dec 2025, along with harness improvements, and hearing about lots of successes with agenic programming. Having the agent managing its own context and doing the full software engineering loop with writing code, running it, and seeing if it works. That was already there before February 9th.

Re: We rewrote JSONata with AI in a day, saved $500k/year

#274

Post by simultsop »

pullshark91 wrote: post #217
simultsop wrote: post #110

Earlier quoted context omitted.

but they saved $500k. Before some humans knew about constraints in it. Now nobody knows. Jokes aside, we will probably see everyone doing this, trying to remove human hands off of code, because they corrupt and AI does not. Joke jokes aside why did we even code until AI?

I don't understand if you're joking or not. I hope you are...

It is not me, it is the whole industry doing it

Re: We rewrote JSONata with AI in a day, saved $500k/year

#275

Post by mbrumlow »

ezst wrote: post #115
Aurornis wrote: post #19

The key point for me was not the rewrite in Go or even the use of AI, it was that they started with this architecture: > The reference implementation is JavaScript, whereas our pipeline is in Go. So for years we’ve been running a fleet of jsonata-js pods on Kubernetes - Node.js processes that our Go services call over RPC. That meant that for every event (and expression) we had to serialize, send over the network, ev…

>> This was costing us ~$300K/year in compute, and the number kept growing as more customers and detection rules were added. > For something so core to the business, I'm baffled that they let it get to the point where it was costing $300K per year. And this, this is the core/true/insightful story the executives will never hear about.

I am always having these arguments. We are paying this other company x a year for something we should build if we really need it.

The rebuttals I always get are “I want you working on something that I can’t pay another company for”. I think it sounds good, but in the long run we always end up a budget conversations and head count limits because we spend so much money on external services and software we should just build.

Every company ever has this problem.

But now with AI. The cost of showing the company “yes we can” is so cheap. I worry for companies who have promotable replacements.

Re: We rewrote JSONata with AI in a day, saved $500k/year

#276

Post by jlarocco »

dpark wrote: post #238
jlarocco wrote: post #215

Earlier quoted context omitted.

> But most of the time the people responsible for the codebase / architecture do not want those questions raised. I don't know if that matches my experience. I've seen plenty of places where the dev teams complain about tech debt and other kludges costing too much, slowing them down and causing other problems, but management don't want to "waste time re-writing working code". But now that management read on linkedin…

Eliminating manual toil seems like a huge win for LLMs. There are a ton of straightforward-but-tedious projects that no one wants to fund because they take 2 dev weeks to implement and the result is a hard to quantify quality of codebase improvement. Some of these can now be handled by an LLM in a day and so they suddenly become extremely tractable. You don’t have to embrace vibe coding to benefit from cheap debt pay…

That's pretty optimistic. First of all, the people who were manually toiling are getting laid off - LLMs aren't exactly making their lives better.

And I'm not talking about cases where an AI can do things faster. We have a few tech debt tickets at work right now where using an AI will take the same amount of time, because the "hard part" isn't writing the code but working with other teams to organize or roll out the changes. But since we can use AI, management is suddenly interested.

It's silly, and I can't wait for the AI bubble to burst.

Re: We rewrote JSONata with AI in a day, saved $500k/year

#277

Post by dpark »

jlarocco wrote: post #276
dpark wrote: post #238

Earlier quoted context omitted.

Eliminating manual toil seems like a huge win for LLMs. There are a ton of straightforward-but-tedious projects that no one wants to fund because they take 2 dev weeks to implement and the result is a hard to quantify quality of codebase improvement. Some of these can now be handled by an LLM in a day and so they suddenly become extremely tractable. You don’t have to embrace vibe coding to benefit from cheap debt pay…

That's pretty optimistic. First of all, the people who were manually toiling are getting laid off - LLMs aren't exactly making their lives better. And I'm not talking about cases where an AI can do things faster. We have a few tech debt tickets at work right now where using an AI will take the same amount of time, because the "hard part" isn't writing the code but working with other teams to organize or roll out the…

> First of all, the people who were manually toiling are getting laid off

I was referring to the sort of work that just never gets funded. Cleanup, refactoring.

If you have business critical toil being done by people who now get laid off, that is obviously a cause for concern.

> the "hard part" isn't writing the code but working with other teams to organize or roll out the changes. But since we can use AI, management is suddenly interested.

So AI has convinced your management to let you pay down tech debt? Seems like a win.

Post reply on HN