Hacker Newsnew | past | comments | ask | show | jobs | submit | danieltanfh95's commentslogin

Please. Manus had a live demo in Google Expo 2025 in Singapore and they blew it. It was such bad taste.

Manus had 1 marketing gimmick with the agents. That is no longer anything novel.


A failed demo discounted a entire 100mm arr?

they are 8 months old. there is no proof of 100m arr.

> Cynics feel smart but optimists win.

survivorship bias.


context poisoning is a real problem that these memory providers only make worse.


IMO context poisoning is only fatal when you can't see what's going on (eg black box memory systems like ChatGPT memory). The memory system used in the OP is fully white box - you can see every raw LLM request (and see exactly how the memory influenced the final prompt payload).


That's significant, you can improve it in your own environment then.


Yeah exactly - it's all just tokens that you have full control over (you can run CRUD operations on). No hidden prompts / hidden memory.


The reason is quite straightforward. LLMs excel at mapping tasks but suck at first principles reasoning and validation.

When you are working on the AI map app, you are mapping your new idea to code.

When people are working with legacy code and fixing bugs, they are employing reasoning and validation.

The problem is management doesn't allow the engineers to discern which is which and just stuff it down their throats.


https://danieltan.weblog.lol/2025/08/function-colors-represe...

the core problem is that language/library authors need to provide some way to bridge between different execution contexts, like containing these different contexts (sync / async) under FSMs and then providing some sort of communication channel between both.


This would generally just discourage open software in general. Rebble is a non-profit and should not pretend to "own" any software or content. Eric didn't do things the polite way, but either way there's nothing to discuss here. Claiming that someone can steal something that is open source implies that they own said open source code / content. that's not how any of this works.

Reselling open source content is always going to be bad taste.


> young person complains about jobs because automation, outsourcing and immigration

> looks at resume

> garbage formatting that only AI would love, with little substantial content beyond the sea of candidates would offer.

All the talk about humans and yet producing a piece of paper that doesnt respect human time.


> garbage formatting that only AI would love

If AI would truly love it, then that seems like the best-case ATS-optimized resume to me. The right tool for the job. Imagine using real people to review applicants - what is this, the 1800s?


It's funny how Japanese contributors often go nuclear whenever things don't go their way instead of communicating etc. Bugs happen. this is tech. open communication is always the way to go before escalation. we see the same phenomenon in git, ruby etc etc etc.

Ironically, Japanese work culture encourages over-communication. It seems that open-source is considered a counter-culture that they want to escape japanese work culture from.


>In October 22, the sumobot was introduced to Japanese KB articles. I cannot accept its behavior and no words. [...] It has been working now without our acceptance, without controls, without communications.

How exactly did you manage to place the blame for no communication on Japanese contributors here given the actual complaint in question?


Because given the nature and scope of the complaint, they should have disclosed prior communication. It is clear here that they probably attempted, poorly, to communicate but ultimately chose to nuke the relationship instead, claiming that they have a monopoly over japanese translations.


Yes, lets dismiss Japanese culture..


This isn't dismissing, this is respecting that this is a risk factor in involving Japanese contributors.


Please reread what you wrote and try to see if you can see how it comes off as at least culturally insensitive if not outright racist.


I live in Japan so I have seen what danieltanfh95 describes many times in person. Japanese people at work seem to have infinite patience with whatever random bullshit their company will ask for, but outside of work they get easily frustrated when things don't go the way they expect.


It is really just BS. These are just basic DSA stuff. We deployed a real world solution by doing of all of that on our side. It's not magic. It's engineering.


This is honestly not better than the LLM-powered double booking system where i just record in plain text, and have the LLM convert it into journal entries. the entire thing is in plain text and backed by git+remote. Recording finances is the hardest part about tracking finances because you tend to forget/get lazy/need to track across multiple family members etc.

This is solved via LLMs that can transcribe, categorise, convert messy records into structured book keeping.

I remain unconvinced that anything more rigid is more useful.


Everyone has their own way of solving their problem. Still a lot of people like spreadsheets more I guess.


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

Search: