Saturday morning thoughts
The best part about joining a new company, especially a startup, is that it’s like hitting the reset button. You get a blank slate. You can be whoever you want to be. It’s not like getting promoted within your current company where people still see the old you.
Let’s say you’re a designer, good at your craft, but maybe not seen as a leader. Maybe your voice lacks confidence. Now you’re joining a new place as a design lead. You can just choose to be the most confident design leader there is. Use ChatGPT’s Advanced Voice, practice with the AI, and ask it to rate your delivery. Work on sounding confident.
This got me thinking about an AI product idea - something that helps you level up from an individual contributor to a manager. It could help you improve your presentation skills, teach you strategic thinking, and how to use the pyramid principle in your comms. I didn’t build it because I couldn’t figure out how it would be better than just using the latest GPT model with advanced voice features.
Cal AI went viral, thanks to TikTok, but I think its real success was due to simplifying calorie tracking. Snap a photo, and bam, structured data, daily calories logged. It cut out so many steps, so much friction. A lot of people start calorie tracking but never stick with it. Trust me, I know. I have tracked mine for two years now. A friend questioned if it can ever be truly accurate. Maybe not, there’s always a chance of hallucination. But, many people are happy with directionally accurate data rather than absolutely accurate data that requires a dozen steps. What are some other simple ideas like this?
There’s Pump.fun. It is making millions daily. Seems so obvious in hindsight. Binance got huge compared to Coinbase because they had a higher risk appetite for listing, let’s say, “interesting” coins. What if you had an even higher risk appetite?
Increasing intensity on one dimension is a cool framework for building new products. I saw a tweet about how people keep adding more caffeine to drinks, and it works. Coke -> Red Bull -> Panera Charged Lemonade.
If you’re building products with the same capital expenditure regardless of who you serve, you’re probably better off targeting high LTV users. Let’s say you want to build an AI astrology chatbot for Indian users and compete with Astrotalk. I’d say build a tarot card reader for white midwest Americans. Your costs are basically the same - fine-tuned Llama or prompting GPT. So why not target users who are more likely to pay what you need to cover your infrastructure costs?
Look at app revenue. The same app makes way more on the App Store than the Play Store. It’s not even close.
This is different from, say, 15-minute delivery. That could have probably only worked in India because of lower service and real estate costs compared to the US or Europe. I remember this 20VC podcast on quick commerce with all the big founders in the space. They debated whether to serve high LTV, high AOV customers who are okay paying for convenience, even with higher operating expenses (US/Europe), or if the model would succeed in India (lower AOV, lower service costs). It’s clear now it’s only really taken off in India. I think the other place it might work long-term is Dubai - people have the money, and there are enough service workers to keep costs lower than the US or Europe, though not as low as India.