Re:Anthropic hits $2T in Nasdaq debut
by rapind on Thursday November 19, 2026 @03:38PM
(#50566017)
Here's a different perspective. Some of these people who were writing boilerplate were more interested in solving a boring problem than trying to craft their way out of it. They understood the business domain, and were clever enough to fire up rails or w/e to build the custom thing they wanted. They never really cared much about the code, and now they get a massive speed boost from LLMs. These people still know their domain and they're still clever, so unless their market disappears (very possible!) they'll be fine.
[ Parent
| on HN ]
Re:Anthropic hits $2T in Nasdaq debut
by evantbyrne on Thursday November 19, 2026 @05:15PM
(#50567226)
A lot of teams have been choosing to build web apps the hard way for a while, which really skews the perceived impact of automation. It has always been possible for one dude to crank out an internal web app in a month. There are just more ways to get there in the same timeline now. Agentic coding hasn't eliminated the risks of shipping software either. If anything, we are living in a far more adversarial environment now. Some things are definitely low enough risk now for less-technical people to take over now, like converting designs to React, but even for that the writing has been on the wall since Figma exported CSS. This is all to say I don't see the need for expertise going away entirely, but rather this is a continuation of the existing trend of software engineering evolving as a profession.
What I'm seeing from customers are increased expectations too. If they have worked with agents, then they expect your software to be at least as easy to use as their agent. It's raising the bar. Interfaces falling short will kill a lot of business in the coming years. Embedding agents is an option too, but that needs to be scoped well with great context or it'll feel worse than using ChatGPT or Claude for your users, which again leaves the door open for competitors who do a better job. I don't think the current agent interface is optimal either, but it's evolving fast, which will drive other software UI to evolve fast too IMO.
It's going to be (already is) very disruptive. Which means a lot of opportunities and a lot of failures to be had.