It might not always be possible to directly translate it to revenue, but in my company, we started using an RAG system to help the Customer Support team answer inquiries. It replaced the old Knowledge Base articles search system, and it improved the number of processed inquiries by about 30%. This decreased customer support times and probably helped make customers happier, which is a pretty nice win, even if it would be pretty much impossible to attribute any revenue numbers directly to it.

My experience says, soon there will be a reduction in the company of the customer support jobs, bringing the customer satisfaction on par to before aka lower, but with slightly more money in the shareholders' pockets.

So a really knowledgeable, specific search engine without ads or a customer base.

I hope you don’t pay for that service, you could train and run it yourself in… a week?

is this custom built or are you using a third party system that integrates with Zendesk, Intercom etc?

It's a custom-built Slack bot that uses our KB articles, and previous customer tickets, to formulate responses and find links to relevant data. Our Customer Service staff uses those generated responses to write actual replies to customers. We currently don't have plans to let the system respond to customer queries directly.

As categories:

- Language translation: is, in general, much better with the ML models than prior automated attempts.

- Internal search tools (GPT + all your internal docs, private)

- Voice-to-text transcription in lots of medical and medical adjacent fields. HN tends to be skeptical of this ("It's going to hallucinate diagnoses!"), but it has a lower error rate than traditional speech rec and human transcription. I met someone who built their own no-code speech to text for their partner's veterinary practice and saved them an hour a day of notetaking.

There's a bunch of businesses in all these areas.

> - Internal search tools (GPT + all your internal docs, private)

the actual hard part of this which is searching for relevant stuff to feed the LLM, which it just formulates into a readable stuff has been around for years. vector databases have been around forever.

Can we please get voice-to-text/action using LLM for cars and home automation. Voice activated anything in a car is such a hassle, because you have to say exactly, "Hey car, turn the temperature to x degrees". You can't say, "Hey car, turn up the heat" Or "Hey car, more AC please". Or "Hey car, max cold".

Alexa in the home is much better now than it used to be, but still way worse than voice chat with ChatGPT.

There are quite a bit of AI for academic researching tools out there that are generating quite a bit of revenue or being bought up by academic publishing companies. You also have dedicated copy writing and article writing tools that basically automate (read plagiarize) the entire academic writing process.

Given what law firms are paying for CoCounsel, Thomson Reuters sure as heck are.

https://faraday.ai - conventional ML not GenAI

At GE Healtchare, "AI", i.e. Neural Network simulation, is used as an optional PET scan image enhancement transformation (glossing over important details of what that actually means). Now, the revenue is from the scanner + software, but it's a contributor to revenue, and arguably sustainable.

Not a foundational model nor a coding helper.

I was speaking more to GenAI solutions, this is more traditional ML