Live data from Hacker News

Show HN: SiteGPT – Create ChatGPT-like chatbots trained on your website content

sitegpt.ai

31–40 of 103 posts

Re: Show HN: SiteGPT – Create ChatGPT-like chatbots trained on your website content

#31
post #17

This is really simple to build yourself. Just use SentenanceTransformers in python, pineconeDB, and whatever screen scrape library you line. Chunk your site content before storing in pineconeDB and then search and whatever you get back use as context for the GPT Chat API.

This comment has the same energy as the infamous Dropbox comment on HN. Do you think no-coders and others in general are really going to build out their own implementation? That said, companies like Intercom, Zendesk and other customer service companies are already starting to do this.

What’s the infamous Dropbox comment? Sounds hilarious.

Re: Show HN: SiteGPT – Create ChatGPT-like chatbots trained on your website content

#32

Earlier quoted context omitted.

This comment has the same energy as the infamous Dropbox comment on HN. Do you think no-coders and others in general are really going to build out their own implementation? That said, companies like Intercom, Zendesk and other customer service companies are already starting to do this.

What’s the infamous Dropbox comment? Sounds hilarious.

> I have a few qualms with this app:

> 1. For a Linux user, you can already build such a system yourself quite trivially by getting an FTP account, mounting it locally with curlftpfs, and then using SVN or CVS on the mounted filesystem. From Windows or Mac, this FTP account could be accessed through built-in software.

> 2. It doesn't actually replace a USB drive. Most people I know e-mail files to themselves or host them somewhere online to be able to perform presentations, but they still carry a USB drive in case there are connectivity problems. This does not solve the connectivity issue.

> 3. It does not seem very "viral" or income-generating. I know this is premature at this point, but without charging users for the service, is it reasonable to expect to make money off of this?

https://news.ycombinator.com/item?id=8863#9224

Re: Show HN: SiteGPT – Create ChatGPT-like chatbots trained on your website content

#33

Earlier quoted context omitted.

Not exactly. Given one of my plans has 5000 pages, it is not possible to just paste the entire content in a prompt. Open AI API has a max tokens limit. I will first do some pre-processing on the content and fetch the relevant pieces of content before giving it as a prompt to the API.

how do you decide what content on the page to index, and how to split them to fit the context window? Amazing concept btw - would love to see more examples (like a chatbot for a more well-known site).

It's pretty straightforward forward with LangChain and GPT-Index. There are lot of tutorials on the Internet for the same like this one https://youtu.be/9TxEQQyv9cE

Re: Show HN: SiteGPT – Create ChatGPT-like chatbots trained on your website content

#35

I'll pay attention to these things when they are open source instead of a service.

Not to go full "Dropbox in a weekend", but if you're technical enough to self-host, this is something you can build for yourself

Everyone is going straight to embeddings, but it'd be easy enough to use old school NLP summarization from NLTK (https://www.nltk.org/)

Hook that up a web scraping library like https://scrapy.org/ and get a summary of each page.

Then embed a site map in your system prompt and use langchain (https://github.com/hwchase17/langchain) to allow GPT to query for a specific page's summary.

-

The point of this isn't to say that's how OP did it, but there might be people seeing stuff like this and wondering how on earth to get into it: This is something you could build in a weekend with pretty much no understanding of AI

Re: Show HN: SiteGPT – Create ChatGPT-like chatbots trained on your website content

#37

I'll pay attention to these things when they are open source instead of a service.

You can get pretty far with this https://github.com/marqo-ai/marqo. Choose your LLM of choice to pair with it. Examples https://github.com/marqo-ai/marqo/blob/mainline/examples/GPT...

Re: Show HN: SiteGPT – Create ChatGPT-like chatbots trained on your website content

#38

I'll pay attention to these things when they are open source instead of a service.

Sweet spot is both. Open source. Pay an expert $19/m to handle this for you knowing you can fallback if they shutdown

Once there are enough of these it may not matter? Just like aws isn’t open source but we use it.

Re: Show HN: SiteGPT – Create ChatGPT-like chatbots trained on your website content

#39
post #17

This is really simple to build yourself. Just use SentenanceTransformers in python, pineconeDB, and whatever screen scrape library you line. Chunk your site content before storing in pineconeDB and then search and whatever you get back use as context for the GPT Chat API.

Yes, my target audience are no-coders or anyone who just wants something ready to integrate on their website without handling any infra on their own. Some of the customers I got are non-technical people. So it was perfect for them. They get a very good chatbot even if they don't know how to code.

Any business/website owner who deals with customer support will like it (assuming it works well).

My guess is the people on here poo pooing this idea are programmers who don't deal with customer support. Don't let their negative response deter you. Let them be fools.

Re: Show HN: SiteGPT – Create ChatGPT-like chatbots trained on your website content

#40

I'll pay attention to these things when they are open source instead of a service.

Not to go full "Dropbox in a weekend", but if you're technical enough to self-host, this is something you can build for yourself Everyone is going straight to embeddings, but it'd be easy enough to use old school NLP summarization from NLTK ( https://www.nltk.org/ ) Hook that up a web scraping library like https://scrapy.org/ and get a summary of each page. Then embed a site map in your system prompt and use langchai…

Is that (i.e. GPT) not still a service?

What people want is something they can run on their own hardware without sending their queries to some third party service which is doing who knows what with them.

This is already possible if you want to mess around with green code that isn't in system repositories yet and buy expensive hardware to make it fast, but you can imagine why some people don't have the time or money for that.

I'm waiting for Intel or AMD to realize there would be a line out the door if they'd make a CPU with an iGPU that could use system memory and run these models at even a quarter of the speed of typical discrete GPUs.

Post reply on HN