Being a SWE , I have seen people including me use a variety of tools for various development stages such as claude code , lovable , figma , anti gravity, gemini , gpt.
Claude code is by far the best out of them and is really effective for teams where product and engineering are tightly integrated - enabling non-technical people to directly interact with the codebase.
But, the main reality is that people thought AI will make work easier but it has actually increased the expectations and stress levels of people at a multifold level. Currently on the ground level AI does help a lot but the real impact on productivity is not that much, especially for senior engineers as they spend more time on complex architecture and modify codes generated by AI for complex tasks.
Ness and Zinnov conducted a detailed analysis of more than 100 software engineers . The study observed that high code complexity environments saw around a 10 per cent reduction in task completion time, suggesting that skilled engineers will continue to play a crucial role in complex software development.
To state the obvious - the future of AI and the transformative potential that it contains is massive but currently its overall impact is more towards junior roles and is less disruptive for highly complex code and senior roles.
The rate at which new tools are coming is happening almost overnight and as models get better at reasoning the ājunior vs seniorā impact will disappear - not in the near future though.
I am really excited to see the next revolution of AI even though it feels a bit scary and intimidating
Thoughts are two. I"ll never buy it and Iāll never sell it short.
What I think is that itās a big player game. Winner takes all.
āCongratulations Anthropic, you win. Everything.ā - Not Ricky Gervais
Jokes aside, the gap that you mentioned, between the real productivity gains and the investment expenditure happening now is precisely the reason why I am neither buying AI or shorting it.
Despite all the heavy expenditure incurred by the top tech companies, expenditure whose benefits would accrue in future period but costs are being incurred now, the top tech companies have All, without exception, reported improvement in profit margins.
Your morning post inspired me to use Claude as 1st time user. Despite having no professional software development experience, I was able to develop a personal finance learning tool tailored to the Indian context by the end of the day. The tool covers five life stages and one universal stage, featuring 31 topics and 310 quiz questions. Achieving this in the pre-AI era would have required immense time and capital. AI, and Claude in particular, is truly revolutionary.
Rehashed and repetitive from previous presentations, and largely borrowed from existing canon & literature. To be honest, not much value and nothing new. The only difference is change in semantics & language. For newcomers, Iām sure it will be useful.
My intro to the topic was Joel Greenblatt (the terribly titled book āYou can be a stock market geniusā). This presentation was a nice relook at special situations in the Indian context. In recent times, Iām seeing special situations as a reasonable way to entering certain picks which would never come close to the valuations I like otherwise. So this presentation gave me a confirmation to that approach.
But Iām yet to dig into Prof Bakshiās other materials (long pending TODO). So I wouldnāt know if the content of this presentation is fresh or not.
