My genuine question--outside of programming-related use-cases (which are remarkable), what has come out of the generative AI "boom" that's a profitable product, or has a viable path to profitability?
What "business transformation" has happened in the private sector as a result of "agents" or whatever? What companies look promising (don't say another VSC fork)?
Even if there are use cases, all of them are operating at extreme losses. As someone else said, maybe throw away video content? But I suspect that the providers are burning piles of cash to maintain users which is true for all AI driven products.
I mean, I'm a software developer and we have AI tools that we use, maybe not customer facing, but we have internal tools that provide lots of value.
Random example #1: we regularly scrape our competitors for their prices and offers, but it's insane to maintain all of that because our competitors update their pages/dom frequently so this doesn't scale over dozens of competitors and thousands of products. We replaced thousands of Playwright scraping code with few hundreds of lines of instructions for an LLM to find the information on the pages itself (it can figure out where/what is).
Random example #2: we sell some products that require lots of customer care and instructions as we're in the home maintenance business and sometimes there are questions about products and their installations that are extremely specific and we do have a chat and call center for our customers to assist them. What our customer care reps do is that they need to find and combine information often from multiple hundreds-page long PDFs. Thus we built an OCR+Rag pipeline to assist our customer care reps. This has provided tremendous value and cut our costs by a lot. It also has the potential to replace them completely and move them to other tasks, but it's gonna be a long time before we can really trust an LLM for that. The risks of giving incorrect or wrong information is too great right now. But internally customer care reps can double check the information as the LLM points out the original pdf sources (up to the specific line).
On top of that I'd say that there's also other tools that both make money and help us, such as AI in creative software like Adobe tools. Some of our graphic experts said that this speeds them up a lot, we need to produce or edit tons of images, illustrations, etc for our products.
And those are just few random examples of how AI definitely can speed up and improve margins of our company, while making cloud and saas vendors money too.
Seems like the above scenarios save a lot of time, aka money.
Maybe not as much as the AI leaders hyped up.
The other example also improved a lot customer satisfaction when dealing with our reps as they are much faster now, can often answer questions without putting them on hold.
I guess my examples weren't exciting (billion dollar) enough.
I'm not into this space, but I also assume that translation work is moving from a writing to an editing job, since an LLM can do that pretty well. I just had an LLM correcting my German mistakes for an email I had to send today.
I'm also not sure that profitability is relevant just yet, as inference costs keep going down. Programming-related use-cases do encompass a lot of why GenAI is useful, simply because they make anyone able to write their mothertongue a programmer. It's fine if LLM-based products bootstrap with high-margin industries like tech, and move into thinner margin ones as costs go down. Of course I'm also not saying the current valuations and hype are justified (I don't know).