I worked for a FAANG company for a bit, and one thing that surprised me was just how many tasks were manual. How noticeably patchy the tooling and automation coverage was. But also very obviously their tooling and automation was state of the art and much more sophisticated than small tech.
But I think there's basic math at play. If I work at a small company and have 100 things, and 50% are automated, then I need say 50 engineers to handle the workload. But if I work at big tech, and there are 100,000 things, and 75% are automated, then I need 25,000 engineers to handle the workload.
There's also the algorithm optimization. While I think alogrithmic optimizations will trickle down, and that's good. For (AlphaEvolve)[https://news.ycombinator.com/item?id=43985489] to have a 1% speedup in most small to medium size businesses. Fine, but that's not worth spending a lot of time on. But for AlphaEvolve to have a 1% speedup at google? May save millions of dollars.
The point of which is to say, there's probably lots of ways that this will trickle down to every day users, and software engineers across the board may see more and more of their tasks automated, but a continuous truism in the industry is that what works for big tech often doesn't work for you medium-sized MSP.
This might explain the difference between why Google says that AI can write 30% of it's code, and medium sized business struggle to get it to work. There's just a lot less boilerplate to right at medium sized businesses. At google, being able to automatically upgrade a node.js dependency may literally impact thousands of microservices, saving thousands of engineering hours. At your company it might save a few hours, if it doesn't take longer than that to get it going in the first place.
Top comments (0)