Darren Shepherd

35.9K posts

Darren Shepherd

Darren Shepherd

@ibuildthecloud

I make AI do things I'm interested in. Co-Founder @Obots_ai Formerly @Rancher_Labs. k3s Creator. Member of The Church of Jesus Christ of Latter-Day Saints

Phoenix, AZ Katılım Temmuz 2013
307 Takip Edilen33.1K Takipçiler
Sabitlenmiş Tweet
Darren Shepherd
Darren Shepherd@ibuildthecloud·
Announcing: DISCOBOT🤖🎉 My personal coding agent session manager. I built this for myself so I can vibe code faster. Watch the video, link to the project thread below. I'll be live streaming today at 10AM PT to fully demo it and talk about why I built it.
English
10
31
98
20.1K
Darren Shepherd retweetledi
Darren Shepherd
Darren Shepherd@ibuildthecloud·
That's basically it. But I can buy a subscription from OpenAI and use it with [not codex cli]. That's why I dumped Anthropic. The value I get from Anthropic and/or OpenAI is the model. That's it. The layers on top are complete commodity. So yes, if I'm paying $200/mo for an amount of tokens, then I want to use the tokens as I please. But Anthropic has made it clear, you are paying $200/mo to use their product. That's not worth it to me at all. I have no particular interest in buying claude code. I just need a model. The harness is a commodity. I wouldn't say I'm mad at them, but I would say I think their mindset is silly/stupid. Subscription/Plan based make sense for token usage. I don't think pure utility billing is really the future of tokens. Not when the tokens are being consumed directly for individual use. Things are different if you are creating a service for other users. I think OpenAI has a better understanding of their value and what buying ChatGPT subscription means. Frankly I like the model a lot better, and that's why I'm their customer.
English
0
0
0
60
David Cramer
David Cramer@zeeg·
@ptremblay "i bought the claude code subscription and am mad i cant use it with [not claude code]"
English
8
0
64
2.8K
Darren Shepherd
Darren Shepherd@ibuildthecloud·
@Xrayez Response? Do you also not have a definition of intelligence? Is asking for a definition illogical?
English
1
0
0
13
Darren Shepherd
Darren Shepherd@ibuildthecloud·
@Xrayez I have said I don't have a good definition for intelligence, not that it can be defined. Can you please offer a definition? Given your definition, I might just agree with you.
English
1
0
0
18
Darren Shepherd retweetledi
📸🔭Brandon Berkoff🚀✨
I’m not asking you, I’m telling you. Take a moment and listen to this 81 second response from Victor Glover after being asked if he had any thoughts leading up to Easter. I don’t quite think it can be overstated how perfect this crew is for the job.
English
533
8.3K
45.9K
1.7M
Darren Shepherd
Darren Shepherd@ibuildthecloud·
Ok. I dunno what justification you'd want. If you want to move out in space, make sense to be able to manufacture things in space. I think it's a bit silly to ask for practical justifications when it come to space exploration. It's always been questionable, but the continued efforts and research benefits society as a side effect. I don't see any downside in trying.
English
0
0
2
119
patagucci perf papi
patagucci perf papi@kenwheeler·
you mfs are about to get some artisanally handcrafted maritime international waters ass tweets
patagucci perf papi tweet media
English
15
0
82
6.1K
Darren Shepherd
Darren Shepherd@ibuildthecloud·
@jonasferoz Why not just tell the LLM you'd like to plan or design. That's what I do. It works fine.
English
1
0
0
55
Jonas Feroz
Jonas Feroz@jonasferoz·
@ibuildthecloud I love plain prompts for brownfield but with a new project I think plan mode has value, particularly for ambitious noobs like me. Also plan mode produces fairly consistent output and even if it's slop I know I can iterate down. I don't actually enjoy using it but it works.
English
2
0
0
31
Darren Shepherd
Darren Shepherd@ibuildthecloud·
@mjackson I feel there's a close correlation with people that use gpt5.4 and those that don't find plan mode useful. Codex is basically trained to not need it. Most people want plan mode so that the LLM doesn't just start doing things. But that's the default behavior of codex.
English
0
0
0
100
MJ
MJ@mjackson·
I’ve been skipping plan mode lately. Just talk to the agent. Hash out the details. “What questions do you have for me?” works quite well. If it’s small enough, and you have enough context left in the current thread, go for it. If not, write a design doc and start a new thread.
David Cramer@zeeg

I use plan mode btw, and you should too. It is extraordinarily effective at helping you bring clarity, and in most harnesses it also helps improve the active context window. Can you ship lines of programming that compile without it? Yes. Except your job is not just shipping "code that compiles".

English
12
0
33
10.3K
Darren Shepherd
Darren Shepherd@ibuildthecloud·
Never write your own crypto, but you didnt say anything about the LLM writing it.
English
0
0
3
590
Ryan Florence
Ryan Florence@ryanflorence·
@thdxr Well, they *are* the first generation to have lower cognitive abilities than their parents.
English
3
0
39
3.7K
dax
dax@thdxr·
so my gen-z coworkers i noticed they say words wrong all the time or they'll mix up 2 similar sounding words is this a thing
English
156
3
726
122.6K
Darren Shepherd
Darren Shepherd@ibuildthecloud·
@gabemonroy Also Gemini has shockingly bad integration with gsuite. Major L. They focus only on looking up information and not "doing things"
English
1
0
2
364
Gabe Monroy
Gabe Monroy@gabemonroy·
I’ve been using Claude more for “personal assistant” stuff and it’s shocking how bad Gemini has become. Gemini is embarrassingly sycophantic, overly structured with formatting, and is just generally inauthentic. I can’t tell if this is an Anthropic W or a Google L.
English
2
0
6
2.1K
Darren Shepherd
Darren Shepherd@ibuildthecloud·
We will achieve AGI before Anthropic can get this to be clear. This is like round 3 over a couple months.
Boris Cherny@bcherny

@EricBuess Yep, working on improving clarity here to make it more explicit

English
0
0
3
832
Darren Shepherd
Darren Shepherd@ibuildthecloud·
@mattpocockuk It's because they aren't creating a developer tool, they are creating a product. But the product just happens to be for developers that are using it as a tool.🙃
English
0
0
4
659
Matt Pocock
Matt Pocock@mattpocockuk·
I don't know what the fuss is about. Anthropic's rules on using subscriptions are very simple: Claude Code = OK Claude's online platform = OK Agent SDK running in personal software = OK... ish? Agent SDK running in commercial software = NOT OK Claude Code running in CI = ?? Oh, maybe it's not so simple... Agent SDK running in CI = ?? claude -p running in CI = ?? claude -p running in personal software = OK claude -p running on open source software, but run on my personal computer = ?? claude -p running on distributed sandboxes, kicked off by me = ?? Distributing open source software which relies on claude -p, and documenting how to use your subscription with it = ?? A thousand other edge cases = ?? Let me be clear. I have never before experienced, from any developer tool, such a frustrating lack of clarity over the basic terms of usage. I personally asked, 3 weeks ago, and have received nothing but delays. The recent @bcherny announcement did absolutely nothing to clarify things. I say this as someone who just released a Claude Code course - my incentives all align with supporting Anthropic.
Boris Cherny@bcherny

@EricBuess Yep, working on improving clarity here to make it more explicit

English
149
116
2.3K
284.5K
Darren Shepherd
Darren Shepherd@ibuildthecloud·
I removed the EnterPlanMode tool and peace has been restored. No more stupid useless token wasting plan mode with gpt5.4
English
1
0
4
659
Darren Shepherd
Darren Shepherd@ibuildthecloud·
wow, ask gpt5.4 for a design and it write a novel.
English
2
0
4
663