Last night, I wanted to do some basic research of a local company. I placed the following prompt into ChatGPT Codex and Claude Cowork: "I need you to do a deep dive analysis of this website. I'm thinking about making a business to compete with this. <web address> I need to understand all products, services, programs and camps that they offer in detail with the pricing and frequency. Let me know if you have any questions" ChatGPT didn't ask any questions, it just got to work, and spit out a low quality, and highly incorrect assessment. It didn't really give the information I asked for, and then it lied about whether the information was available. Claude crushed the task. Right away, it asked me some clarifying questions about the task to ensure we were on the same page, and then it delivered a competitive analysis in Word and a financial summary in Excel. Claude gave me SWOT analysis, pricing summaries, and it also went the extra mile to research local competitors as well. The output was above and beyond what I initially asked for, and this gave me several different areas to explore in subsequent questions over the following hour. I will always have a fondness in my heart for ChatGPT, as it was my first real heavily used LLM, but it just continues to disappoint me over the last couple of months. Every time I touch ChatGPT recently, I feel like I'm logging onto MySpace while my friends have already navigated to Facebook....

I can't make it tonight unfortunately, see you all next week. Curious if people would be open to moving to a morning coffee/catch up format with a tight 30 minute limit. We can always leave our longer form shoot the shish sessions in place too...

Hackathon? Group project? Comedy roast of @Oleg H ?

but I'm still happy with the output https://www.youtube.com/watch?v=mi_ccLy_JXw

The world is finally waking up to the death of the complexity tax. SAAS companies are getting wrecked