Sachin Jain’s Post

One more post about AI for the road… Until there’s something else I need to say. Five days ago, just after my “Gen-AI developed” app released on the Apple App Store, I was ready to move on to my next project. Or at the very least take a little break. Something was nagging at me though. My card game was built under a (self impossed) deadline and was largely thrown together with a lot of: “Yo Chat-GPT, this thing is broke dude. Fix it.” I’m exaggerating a little bit, but only a little. Ultimately, when the app went on sale, it wasn’t fully baked. It worked and did what it was supposed to, but as a game, it didn’t have much going for it. This is where my well earned break was thrown in the bin. I started to think about how I could extend the game (with some help from friends and my partner) and I slowly built out the app. But it wasn’t that straightforward. The code Chat-GPT had generated didn’t really give much room to extend. I wanted to add new game modes, new characters and new screens. All of the original code wouldn’t allow for that. Gen-AI had written precisely what it needed and nothing more - often sacrificing future expansion in favour of efficiency. Some might argue that this also happens with humans and I’m certainly faulty of that. The lesson I’ve learned with this experience is to always think of the bigger picture, because Gen-AI won’t do that for you. If you’re fitting a kitchen and you ask it to size up a fridge, it’s not going to understand the knock on effects of everything else in the room unless you tell it. The same goes for my app - I didn’t tell GPT that the game needed to be a springboard for future growth. In the end, I actually rewrote the 95% of the codebase, keeping some of the more complex snippets that were generated with the previous version. I still used Gen-AI for unfamiliar parts, but this time round I kept things modular, thought of the bigger picture and made sure I wasn’t backed into a corner I couldn’t get out of.

You’ve got to love it when a feature update is so good that it gets you to resubscribe! SerpApi releasing AIO scraping + a new ChatGPT model that is better at writing code = My favourite ChatGPT use case. Writing Google Apps Script. 😄 This script requires the list of queries to be in column A and will prompt what column for the output (a standard GSC output using the "queries" tab works well, you can enter "f" for the column output in this case). When sorting the output from GSC, I filtered for the US, the last seven days, and only the results on the first page (average position less than 10). I just tested on an enterprise-level client's home page and the output is very helpful! Steal this script at: https://lnkd.in/gzcSy_Hg Thanks to Greg Bernhardt for bringing this to my attention! #WebScraping #GoogleSearchConsole #GoogleAppsScript

Short 15-Minute Project: how LLMs changed the websites robots.txt ? I discovered some interesting insights about how the top websites handle web crawlers. I compiled a list of the leading websites, combined with the top 50 media/news sites from various sources, and deduplicated them. I wrote a short script to download their robots.txt files to see which crawler bots they block. Here are the results, from the top 113 websites I analyzed: • 46 block GPTBot - OpenAI Bot • 31 block Google-Extended - Google's LLM Bot • 26 block CCBot (Common Crawl), used to train LLMs • 16 block Anthropic’s bot • 14 block PerplexityBot Websites are desperately trying to combat this new user paradigm, where people turn to ChatGPT for immediate answers instead of visiting websites. The current state of web browsing is plagued by pop-ups, intrusive call-to-actions, cookie consent pop-ups, paywalled content and overall a frustrating user experience. Users are increasingly turning to ChatGPT-like systems, bypassing these annoyances. With chatGPT alone receiving around 1.8 billion (and growing) page views a month, not counting API calls and other applications, we are witnessing a significant shift in web traffic. This trend suggests we might soon experience a completely different way to interact with information. This new paradigm might include intelligent assistants that compile information, fact-checked from multiple sources, providing users with reliable data. Websites that rely on fake information will face tough times ahead.

Just finished the course “Building Apps with AI Tools: ChatGPT, Semantic Kernel, and Langchain” by Denys Linkov! Check it out: https://lnkd.in/e-WuD2JG #langchain #openaiproducts #semantickernel #applicationdevelopment.

Just finished the course “Building Apps with AI Tools: ChatGPT, Semantic Kernel, and Langchain” by Denys Linkov! Check it out: https://lnkd.in/gB6FAx6j #langchain #openaiproducts #semantickernel #applicationdevelopment.

Here's an apparently leaked changelog from "Google Gemini Advanced" release allegedly coming Weds. I think it opens some potential industry shifts. Here's what you need to know. 1. The huge thing is the rebrand from Bard to Gemini. I never liked Google Bard. The name was too high brow and made the tool seem like a writing assistant, not taking into account the multi-modal nature of modern apps. 2. Is this going to be better than GPT 4? Who knows. I'm surprised at how subdued the promises are here. Complex tasks like coding, logical reasoning? Yawn. However, it does mention the ability to upload file, documents, data and more. Regardless of how close to GPT 4 turbo this is, it's clearly meant to compete with Chat GPT. 3. Gemini Advanced. You're going to have to pay for the Ultra version of Gemini. The ChatBot industry is coalescing about a specific pricing model. $20/month for advanced capabilities which match what Chat GPT is currently doing. The differentiation will possibly be Google's integration into android and their other tools which are quite mighty. 4. Standalone Android app. These services are looking very similar with individual mobile apps and services. Google has a clear advantage being able to integrate with Android phone services like no one else. We already saw what they did with the Galaxy S24 (I want that phone). Unless OpenAI does a deal with Apple (which I wouldn't be shocked if they do), the integrations are to Google's advantage. Buckle up! It should be an interesting next few months. OpenAI is ahead in terms of market dominance, but Google's going all in.

Just finished the course “Building Apps with AI Tools: ChatGPT, Semantic Kernel, and Langchain” by Denys Linkov! Check it out: https://lnkd.in/dNFQ_d-E #langchain #openaiproducts #semantickernel #applicationdevelopment.

Just finished the course “Building Apps with AI Tools: ChatGPT, Semantic Kernel, and Langchain” by Denys Linkov! Check it out: https://lnkd.in/gGwq-3Ts #langchain #openaiproducts #semantickernel #applicationdevelopment.

Just finished the course “Building Apps with AI Tools: ChatGPT, Semantic Kernel, and Langchain” by Denys Linkov! Check it out: https://lnkd.in/eMqi2NC2 #langchain #openaiproducts #semantickernel #applicationdevelopment.

Just finished the course “Building Apps with AI Tools: ChatGPT, Semantic Kernel, and Langchain” by Denys Linkov! Check it out: https://lnkd.in/eMqi2NC2 #langchain #openaiproducts #semantickernel #applicationdevelopment.