There’s AWS Bedrock Knowledge Base (Amazon proprietary RAG solution) which can digest PDFs and, as far as I tested it on real world documents, it works pretty well and is cost effective.
Do you have an idea what would be my external IP address? On my phone connected to a mobile network, I get assigned mobile IP address which is my external IP address. It's not attached to the SIM card because it changes when I reconnected. Is it handles by the BTS software? Do I get assigned an IP address and BTS communicates on my behalf using that address which comes from the mobile network operators pool?
Can I use this to run my own mobile network? Is there something like a blank SIM card which I could use for it? I don't need global coverage but is it possible to create my own BTS on a PC (with some antenna connected to it) and then have my own SIM card which I can insert into regular phone/device and have it connected to my BTS and connect to the Internet?
I don’t really know. I don’t write posts to optimize for SEO (include FAQ at the end or something like that) and hope it’s just good content people will share.
There are also SEO pages which do not have any useful content. I think I should have more of them because my competitors have only SEO pages but I don’t have time for it as I have to focus on the product and customer support. Probably a good mix between useful content blog posts (maybe with SEO filling) and strictly SEO pages is best to bring traffic.
I think that LLM costs, even GPT-4o, are probably lower compared to proxy costs usually required for web scraping at scale. The cost of residential/mobile proxies is a few $ per GB. If I were to process cleaned data obtained using 1GB of residential/mobile proxy transfer, I wouldn't pay more for LLM.
I agree that web scraping is a shady business in many cases but there is definitely a difference between setting up a few mobile proxies for yourself and using devices and networks which belong to other people without them even knowing this until they cannot access some websites because there was a bot detected in their network.
I would also include deceptive credits systems used by SaaS which have usage-based like subscriptions. It’s a bait and switch variant. First, you think one call to the API is one credit but it always turns out that you need calls which consume 20 or 50 credits instead and you have to move to a more expensive plan and buy millions of credits every month. Second, unused credits do not roll over to the next month so your effective cost per call is orders of magnitude larger compared to what you expected.
I tried not buying food with added sugar but it’s surprisingly difficult. Here is an interesting analysis I did some time ago which shows that for half of the food items, sugar is the main ingredient: https://scrapingfish.com/blog/scraping-walmart
At https://scrapingfish.com/ we have both options, usage based https://scrapingfish.com/buy and subscriptions (monthly unlimited requests plan) https://scrapingfish.com/unlimited. Despite subscriptions being cheaper option per request, usage based is way more popular. Only less than 10% of our users have subscribed to unlimited monthly plan. I guess usage based plans give users more control over how much they spend or maybe they simply don't want to subscribe to another service.
I’m still working on a web scraping API (https://scrapingfish.com/). For some people it’s evil bot but for others it’s enabler for public data access. I think it’s useful.