The more detailed you make your prompts, the better Claude Code performs. But to get it to work correctly, you need specifications so detailed that you’ve almost done the implementation.
You’re not necessarily saving time - you’re programming in Markdown instead of code. … So, to be effective, you need to behave just like Gabriel. You need to watch Claude plotting stuff in the console. As one moment it generates sensible code, then completely loses the plot the next.
Most organisations don’t even recognise what good engineering looks like. They treat software development as a commodity – a manufacturing production line – measured by how many features are shipped rather than whether the right outcomes are achieved. Few understand the value of investing in modern software engineering best practices and design – the things that make those outcomes sustainable.
I have never been the best engineer I’ve ever met.
Even if we may be in an AI bubble, it seems Altman is expecting OpenAI to survive the burst. “You should expect OpenAI to spend trillions of dollars on data center construction in the not very distant future,” Altman said. “You should expect a bunch of economists to wring their hands.”
I’m not sure whats more audacious. Declaring everyone ELSE is delusional or lighting billions on fire for the equivalent of a summary machine…
Our analysis suggests that tariff measures are already exerting measurable upward pressure on consumer prices. The rise in prices beginning in early 2025 coincides closely with tariff developments, and our model-based regressions confirm that these effects are statistically and economically significant.
At the same time, the pass-through remains partial; only a portion of the model-predicted effect has materialized so far. This could reflect delays in price adjustments, competitive pressure limiting firms’ ability to raise prices, or expectations that the tariffs may prove temporary.
“If you’re, like, farming, you’re doing something people really need,” Altman explained. “You’re making them food, you’re keeping them alive. This is real work.” But the farmer would see our modern jobs as “playing a game to fill your time,” and therefore not a “real job.”
“It’s very possible that if we could see those jobs of the future,” Altman said, we’d think “maybe our jobs were not as real as a farmer’s job, but it’s a lot more real than this game you’re playing to entertain yourself.
There are no weird files that shouldn’t belong there, no duplicate implementations, and no emojis all over the place. The comments still follow the style I want and, crucially, often aren’t there. I pay close attention to the fundamentals of system architecture, code layout, and database interaction. I’m incredibly opinionated. As a result, there are certain things I don’t let the AI do. I know it won’t reach the point where I could sign off on a commit.
Small time savings were observed across most use cases, with written tasks presenting the largest time savings. However, some tasks, like scheduling and generating images, incurred additional time to complete the task when participants used M365 Copilot. Additional time to complete tasks was primarily caused by either M365 Copilot being unable to produce high quality outputs or the task being additional workload only completed due to users having M365 Copilot.
In Claude Code, we’ve added checkpoints—one of our most requested features—that save your progress and allow you to roll back instantly to a previous state.
Dunno. recently I’ve been using OpenAI codex. I find I can work with much longer on the 20/month plan than I can sonnet. and I’m not nearly smart enough to detect differences in the two tools approach to solutions at this time.
Quote Citation: Anthropic, “Introducing Claude Sonnet 4.
More than three quarters (77%) of companies’ usage of Anthropic’s Claude AI software involved automation patterns, often including “full task delegation,” according to a research report the startup released on Monday. The finding was based on an analysis of traffic from Anthropic’s application programming interface, which is used by developers and businesses.
I think the critical thought here is maybe users are only leaning into AI where it makes sense now?
Do I program any faster? Not really. But it feels like I’ve gained 30% more time in my day because the machine is doing the work. I alternate between giving it instructions, reading a book, and reviewing the changes.
I think this is the way. and matches what I’ve experienced. Sure you can spool code from the CLI to your editor and review via git diffs. But are you accomplishing more throughput or just offloading syntaxual structure.
