Tom Sharp (@t_sharp) header
Tom Sharp (@t_sharp) profile picture

building brightplace: https://nitter.gq/t.co/o57MI7li9B AI | DevOps | Infra

NYC
Joined July 2017
Photo by Tom Sharp (@t_sharp)
Tom Sharp retweeted
Sorry to say but: Vibe coding didn't kill software engineering, it just made bad engineering easier to hide.
38
3
47
2,143
I keep seeing people on X declare that software engineering is dead, usually alongside some vibe-coded app. I don’t think prompting an agent to build an app in a weekend means it’s the end of software engineering.
2
7
I expect AI to change the job substantially, including how much code engineers write themselves. But if implementation becomes cheaper, being able to decide what’s worth implementing and evaluate the result seems pretty valuable to me.
1
1
Serious question. Are we going to need engineering docs in the future? If the code is the source of truth and we can understand the code much faster with agents, what purpose do the docs serve?
7
The great thing about identifying features that aren't in the critical path of MVP is you get to build them and rebuild everything you already built right before launching the MVP.
10
I have a feeling that there is going to be a shift to more localized apps as people get tired of sharing their data. This will be especially true for AI chat and agents - as tiny models get more powerful we will see an uptick in local agents
1
18
Similarly I could see more companies being built that just offer a service in exchange for a fee, no “hidden” business model. People are tired of the real business being hidden and/or their data being sold.
13
Tom Sharp retweeted
New: A map of the most important skills in AI Engineering.
606
4,143
485
24,721
7,067,661
Tom Sharp retweeted
I wouldn't be surprised if there are several more back and forths. I have no idea who ends on top. If AI tops out somehow, open weights win, and there are a thousand model companies whose moats are proprietary training data. Or four after a new JP Morgan combines them.
13
4
2
189
30,632
AI terms you should know, defined in plain english.
2
1
1
75
Guardrails: rules that catch bad outputs before they reach the user. Content filters, schema validation, fallback responses.
1
3
Eval: measuring whether your AI actually works. Not vibes. Metrics, test cases, regression checks.
3