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

Three kinds of AI products work

seangoedecke.com

Post reply on HN
Print view View original

121–130 of 130 posts

Re: Three kinds of AI products work

#121

Post by KurSix »

To me, this classification overlooks the largest and arguably most profitable class of AI products: models that aren't the interface but are embedded deep within existing systems as infrastructure components, for example:

- Anti-fraud systems in fintech

- Supply chain optimization

- Predictive maintenance for equipment

- Models for medical image analysis

They don't have a "chat," and they aren't "agents", they are just APIs that solve one highly specialized task, but do so with extreme precision and reliability. They don't compete with ChatGPT because they solve completely different problems, and they're the ones that have been bringing in real money for years

Re: Three kinds of AI products work

#122

Post by KurSix »

wongarsu wrote: post #6

This seems to be biased heavily towards products that look like an LLM. And yes, only a small number of those work. But that's because if your product is a thing I chat with, it immediately is in competition with ChatGPT/Claude/Grok/etc, leading to everything the article expressed. But those are hardly the only use cases for LLMs, let alone AI (whatever people nowadays mean by AI) To name some of the obvious counter-…

Grammarly and DeepL are great examples of successful AI products, I agree. What's most interesting is that they became successful long before the LLM hype. They didn't try to be "universal intelligence", they just wanted to help people write and translate better. Maybe the next big breakthrough won't be in creating another chatbot, but in finding more of these "boring" but critically important niches

Re: Three kinds of AI products work

#123

Post by KurSix »

echelon wrote: post #73
wongarsu wrote: post #6

This seems to be biased heavily towards products that look like an LLM. And yes, only a small number of those work. But that's because if your product is a thing I chat with, it immediately is in competition with ChatGPT/Claude/Grok/etc, leading to everything the article expressed. But those are hardly the only use cases for LLMs, let alone AI (whatever people nowadays mean by AI) To name some of the obvious counter-…

> One other thing I haven’t mentioned is image generation. Is this part of a chatbot product, or a tool in itself? Frankly, I think AI image generation is still more of a toy than a product, but it’s certainly seeing a ton of use. There’s probably some fertile ground for products here, if they can successfully differentiate themselves from the built-in image generation in ChatGPT. This guy is so LLM-biased that he's…

Exactly. And the author seems to have missed that media generation is arguably the purest demonstration of agentic potential outside of coding. Modern tools like Krea or Magnific are no longer just "enter prompt, get image", it's an iterative multi-step dialogue with the system where you're controlling composition, upscaling details, and inpainting - that is the agentic work he's talking about; it just operates on pixels instead of code

Re: Three kinds of AI products work

#124

Post by taherchhabra »

echelon wrote: post #96
taherchhabra wrote: post #93

Earlier quoted context omitted.

I am building one such tool. Flickpseed.ai Give it a shot

Hope you don't mind the unsolicited feedback - ComfyUI-inspired node graphs are the wrong approach for visual media. Nodes are great for the 1% of artists that get into it, but you really need to build the Adobe / Figma of image and video tools. Not Unreal Engine Blueprint / ComfyUI spaghetti. ShaderToy and TouchDesigner and Comfy are neat toys, but they're not what the majority of people will use. We want to mold id…

yes agree, and that has been my learning as well.in the current version of the product, the end user does not need to know about the nodes and other complexity, instead can use various mini apps/agents without knowing about any ai model, example fashion tryon mini app https://www.flickspeed.ai/agents/fashion-tryon--6901b82047f3...

Re: Three kinds of AI products work

#125

Post by airstrike »

bix6 wrote: post #77
airstrike wrote: post #13

Earlier quoted context omitted.

I'm building an opinionated take on this. It's shaping up nicely. If you're a Rust developer reading this, interested in AI + GUI + Enterprise SaaS, and wants to talk, I'm building a team as we speak. E-mail in profile.

So like an o365 ServiceNow?

Not quite. It's a desktop GUI

Re: Three kinds of AI products work

#126

Post by kylehotchkiss »

levocardia wrote: post #40

Very obviously missing the mundane agentic work. I think the following things are basically already solved, and are just waiting for the right harness: - Call this government service center, wait on hold for 45 minutes, then when they finally answer, tell them to reactivate my insurance marketplace account that got wrongly deleted. - Find a good dentist within 2mi from my house, call them to make sure they take my in…

I have doubts the nontechnical users/general public will be able to articulate what they need at this level of detail. Things like "account that got wrongly deleted" still implies that the person making the call understands the internals/bad implementation of healthcare or has the capability to discover it.

The HN audience has argued with LLMs enough to know the level of detail they need to make meaningful progress (or when too many details leads it down rabbit holes).

Re: Three kinds of AI products work

#128

Post by rckt »

Kiro wrote: post #120
rckt wrote: post #119

Earlier quoted context omitted.

There's nothing to elaborate actually. The state of the AI currently does not allow to use seriously. If we mean "then spend N amount of time reviewing and trying to make a better output", then yes, I can agree that AI is usable. But the time wasted on the reviews can be simply put into doing the task yourself. And it will be faster and less frustrating.

You not being able to use it successfully is no basis for such a conclusion. One verified observation is stronger than infinite lack of observation.

It's not my personal observation. It's a general observation. I witnessed only one person in real life who said that they use AI in their job and they are completely satisfied with it. The rest point out the poor quality of the output and how technically savvy one should be to successfully (to some extent) use the AI in their jobs.

Re: Three kinds of AI products work

#129

Post by Kiro »

rckt wrote: post #128
Kiro wrote: post #120

Earlier quoted context omitted.

You not being able to use it successfully is no basis for such a conclusion. One verified observation is stronger than infinite lack of observation.

It's not my personal observation. It's a general observation. I witnessed only one person in real life who said that they use AI in their job and they are completely satisfied with it. The rest point out the poor quality of the output and how technically savvy one should be to successfully (to some extent) use the AI in their jobs.

Pretty long stretch between "useless" and "completely satisfied". I don't think I'm completely satisfied with anything in my life including AI, but that doesn't mean I'm not having a lot of success with it.

The things stated in the article are true and even pretty tame. Agent mode in Cursor has been standard for months and I'm surprised that anyone would say that LLMs are not good at scanning documents. The normal AI skeptic's position would be "yes, but that's all they're good at".

Re: Three kinds of AI products work

#130

Post by rckt »

Kiro wrote: post #129
rckt wrote: post #128

Earlier quoted context omitted.

It's not my personal observation. It's a general observation. I witnessed only one person in real life who said that they use AI in their job and they are completely satisfied with it. The rest point out the poor quality of the output and how technically savvy one should be to successfully (to some extent) use the AI in their jobs.

Pretty long stretch between "useless" and "completely satisfied". I don't think I'm completely satisfied with anything in my life including AI, but that doesn't mean I'm not having a lot of success with it. The things stated in the article are true and even pretty tame. Agent mode in Cursor has been standard for months and I'm surprised that anyone would say that LLMs are not good at scanning documents. The normal AI…

My personal experience with any approach I tried ended up in a semi-failure state. I can't trust LLM and have to check everything myself to either actually solve a problem or be sure that the LLM did not invent things. This fact simply kills the usefulness of LLMs. They are not reliable and I waste more time trying to accomplish something just using the AI. If people around me were praising AI and I was the only one unable to find an approach, I would agree with your points. But it's not the case. The general perception of the AI tools is quite pessimistic.
Post reply on HN