@davepl1968@Dell Careful there Dave, you might have people believing you? Since Kevin is 72, he would have been teenager during Apollo Years. Since Michael Dell was born in 1965. I guess the years overlap. Dell could have thought of the precision device before kindergarten. I know it is a joke.
@bendee983 Speeding up code generation improves a non bottleneck in most organizations. Are we really not delivering product because we can’t code fast enough. I would suggest reading Eli Goldratt’s The Goal. Code Generation is not the Herbie in the process.
@heygurisingh@subhajit_pal@AnthropicAI Curious how you are defining down. Looking at the charts, every service listed is above 99%. Are you expecting 5 9’s or similar. Doe’s Anthropic offer that SLA? Does any AI provider offer that SLA?
@CtrlAltDwayne Welcome to the modern SAAS world, especially one with companies dealing with rapid AI adoption. In my 29 plus years of Internet, I have experienced plenty of this, even from major companies, like Microsoft. OpenAI has even bigger issue with court case. Nadella's testimony hurt.
@dotmariusz@CtrlAltDwayne Your Claude part is factually false. I hope you are making a hyperbole. If Claude edits two lines, your statement is false. Claude $20 plan will edit 100’s of lines of code in a session. I have been doing this day after day for past month.
@devXritesh By your post, not enough information is provided like what level are you currently working at? We have passed the tipping point with AI related to coding. The big change is not necessarily immediate, but over next two years the standard Scrum and giflow development will evolve.
@crsanchezx I would say my undergraduate degree in Industrial Engineering is still a good choice. It focuses on systems thinking, total quality improvement, and already (for decades) addressed automation in the factory and offices. I had combined with Computer Science emphasis.
@ritu_twts The real question is how many can you program solely from memory without AI, Internet, or even code hints/helpers. Since I switch between languages (mainly Javascript, Java, Typescript, I would say almost none. Especially if one throws in type hints.
@Layton_Gott With any new technology there is always change. I am sure the Word Processing pool had mixed emotions when everyone got computers on there desk and there was no more WP pool. When the Internet was at the tipping point management did not want users having browsers.
@ritu_twts Store in secret store. Permission granted to said store limited to who needs the secret. Typically that is only team level need. Teams are 2 pizzas teams at largest.
@thisiswhyibuilt @BacLeodiv Sonnet or Opus? I have been using Sonnet 4.6. I tried image generation a few days ago (logo). That is where Claude started burning through the tokens. I don’t remember if that was in opus. Never got an acceptable result. Gemini base (no sub) succeeded in about 15 .
@thisiswhyibuilt @BacLeodiv What model are you using? What language are you developing in? I am on $20 plan and get 4-5 hours deep effort in past several days with tokens left. Using Sonnet, monorepo, TS, Python. Refactored entire project. First described changes to Claude to build Task.md
@housecor So far, I have had to prompt AI, typically before a commit to update the README and other documents relating to the changes. Even with individual tasks I take them one step at a time. Even with those checks I can be 5x to 10x faster.
@AdamTornhill@simplescuriosos Completely agree. I do worry however how the entry level generation learns the craft of breaking problems into tasks to hand to an agent.
@asaio87 There are plenty and most don’t work for startups nor for software companies. They work in FinTech, Healthcare, and other Brick and mortar industries. Many supporting systems 10!plus years old.