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

91–100 of 130 posts

Re: Three kinds of AI products work

#91

Post by Animats »

> So you can never give a support chatbot real support powers like “refund this customer”, because the moment you do, thousands of people will immediately find the right way to jailbreak your chatbot into giving them money.

And that's the elephant in the room. AI "agents" can't do much until someone solves that problem. Most AI "agents" work for and favor the business operating the agent, but impose the costs of their errors on the customer. Errors are an externality, like pollution. This is no good.

Re: Three kinds of AI products work

#92

Post by gervwyk »

ohyoutravel wrote: post #47
torlok wrote: post #8

So the only AI products that work is a chat bot you can talk to, or a chat bot that can perform tasks for you. Next thing you'll tell me is that the only businesses that work are ones where you can ask somebody to do something for you in exchange for money.

Realistically there are only four types of businesses writ large: tourism, food service, railroads, and sales. People building AI-based products should focus on those verticals.

lol. would love an episode on how Micheal and Dwight responds to Jims Ai slop.

Re: Three kinds of AI products work

#93

Post by taherchhabra »

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…

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

Re: Three kinds of AI products work

#95

Post by gherkinnn »

Aldipower wrote: post #39
kevin_thibedeau wrote: post #30

Earlier quoted context omitted.

They're idiot savants. Use them for their strengths. Know their weaknesses.

So, what are their strengths then? I've fed it with a detailed, very well documented and typed API description. Asking to construct me some not too hard code snippets based on that. GPT-5 then pretend to do the right thing, but actually is creating meaningless nonsense out of it. Even after I tried to reiterate and refine my tasks. Every junior dev is waaay better.

I recently had something no longer compile. I got bored sniffing around after maybe an hour, set Claude in Zed on to it, got a snack, and by the time I was back it had found the problem.

When I am unsure how to implement something, I give an LMM a rough description and then tell it to ask me five questions it needs to get a good solution. More often than not, that uncovers a blind spot.

LLMs remain unhelpful at writing code beyond trivial tasks though.

Re: Three kinds of AI products work

#96

Post by echelon »

taherchhabra wrote: post #93
echelon wrote: post #73

Earlier quoted context omitted.

> 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…

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 ideas like clay.

Watch the demos Adobe just gave from their conference two weeks ago. That's what you should build. Something artists and designers and creatives intuit as an extension of themselves. Not a mathematical abacus.

Re: Three kinds of AI products work

#98

Post by qayxc »

carsoon wrote: post #59
Shebanator wrote: post #5

Author forgot about image, video, and music creation. These have all been quite successfully commercially, though maybe not as much artistically.

Recent articles seem only to mean LLMs when they reference AI. There are tons of commercial usecases for other models. Image Classification models, Image Generation models (traditionally difusion models, although some do use llm for image now), TTS models, Speach Transcription, translation models, AI driving models(autopilot), AI risk assessment for fraud, 3D structural engineering enhancement models. With many of th…

> With many of the good usecases of AI the end user doesn't know that ai exists and so it doesn't feel like there is AI present.

This! The best technology is the one that you don't notice and that doesn't get in the way. A prominent example is the failure of the first generation of smart phones: they only took off once someone (Apple) managed to the hide OS and its details properly from the user. We need the same for AI - chat is simply not a good interface for every use case.

Re: Three kinds of AI products work

#99

Post by qayxc »

irq-1 wrote: post #70
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…

> - Find a good dentist within 2mi from my house, call them to make sure they take my insurance, and book an appointment sometime in the next two weeks no earlier than 11am The web caused dentists to make websites, but they don't post their appointment calendar; they don't have to. Will AI looking for appointments cause businesses to post live, structured data (like calendars)? The complexity of scheduling and multip…

The problem is that we're reverting back to the stone age by throwing unnecessary resources at problems that have a simple and effective solution: open, standardised, and accessible APIs.

We wouldn't need to use an expensive (compute-wise) AI agent to do things like making appointments. Especially if in the end you'd end up with bots talking to bots anyway. The digital equivalent of always up-to-date yellow pages would solve many of these issues. Super simple and "dumb" but reliable programs could perform such tasks.

Scheduling multiple calendars doesn't require "AI" - it's a comparatively simple optimisation problem that can be solved using computationally cheap existing algorithms. It seems more and more to me that AI - and LLMs in particular - are the hammer and now literally everything looks like a nail...

Re: Three kinds of AI products work

#100

Post by rckt »

> The third real AI product is the coding agent. People have been talking about this for years, but it was only really in 2025 that the technology behind coding agents became feasible

No, they didn’t.

> LLMs are particularly good at tasks like “skim through ten pages of search results” or “keyword search this giant dataset for any information on a particular topic”.

No. They are not.

Amazing article without any support of the statements made. Just, “because I think so”. Cool.

Post reply on HN