Daniel Imfeld

4.2K posts

Daniel Imfeld banner
Daniel Imfeld

Daniel Imfeld

@dimfeld

Serial medical data startup founder. Posts on Rust, Svelte, GeoData, DB/Analytics, ML/AI

Lihue, HI شامل ہوئے Nisan 2007
759 فالونگ804 فالوورز
Daniel Imfeld
Daniel Imfeld@dimfeld·
If you acquire a company and there are bonuses based on revenue targets, the terms of the acquisition usually require you to make a good faith effort to allow the company to achieve those targets. It's not illegal but it is breach of contract. As for immoral? If I hire you or buy your company, and give you targets for a bonus, and then take action designed to prevent you from hitting those targets, that seems quite obviously immoral.
English
0
0
106
1.3K
Alex Wiggin
Alex Wiggin@realalexwiggin·
@dimfeld @_NathanCalvin It's unclear to me what that would be illegal or even immoral. Why (and under what conditions) should one party be compelled to harm their shareholders' interests in favor of some third parties?
English
4
0
0
1.3K
Nathan Calvin
Nathan Calvin@_NathanCalvin·
The CEO of Krafton (creator of PUBG) asked ChatGPT to create a "corporate takeover strategy" to prevent a company they acquired from hitting a revenue target within a certain time window (which would trigger an additional payout). ChatGPT (against his lawyer's advice) suggested locking down the acquired companies Steam account to prevent them from publishing Subnautica 2 in the time window, which the CEO of Krafton followed. ChatGPT's advice did not hold up at trial and the judge was not happy. The opinion is a wild read and includes several direct quotes from the Krafton CEO's ChatGPT conversation. I feel like it's gonna take a few more high profile examples like this until executives start realizing that conversations with ChatGPT are not privileged and you probably shouldn't describe your questionably legal schemes to them in detail!
Nathan Calvin tweet media
English
132
2K
18.2K
2M
Diptanu Choudhury
Diptanu Choudhury@diptanu·
Didn’t they buy windsurf? Some folks were saying the windsurf release in December was solid over the Christmas break.
Steve Yegge@Steve_Yegge

I was chatting with my buddy at Google, who's been a tech director there for about 20 years, about their AI adoption. Craziest convo I've had all year. The TL;DR is that Google engineering appears to have the same AI adoption footprint as John Deere, the tractor company. Most of the industry has the same internal adoption curve: 20% agentic power users, 20% outright refusers, 60% still using Cursor or equivalent chat tool. It turns out Google has this curve too. But why is Google so... average? How is it that a handful of companies are taking off like a spaceship, and the rest, including Google, are mired in inaction? My buddy's observation was key here: There has been an industry-wide hiring freeze for 18+ months, during which time nobody has been moving jobs. So there are no clued-in people coming in from the outside to tell Google how far behind they are, how utterly mediocre they have become as an eng org. He says the problem is that they can't use Claude Code because it's the enemy, and Gemini has never been good enough to capture people's workflows like Claude has, so basically agentic coding just never really took off inside Google. They're all just plodding along, completely oblivious to what's happening out there right now. Not only is Google not able to do anything about it, they don't seem to be aware of the problem at all. I'm having major flashbacks to fifty years ago as a kid at the La Brea Tar Pits, asking, "why can't they just climb out?" My Google friend and I had this conversation over a month ago. I didn't share it because I wanted to look around a bit, and see if it's really as bad as all that. I've been talking to people from dozens of companies since then. And yeah. It's as bad as all that. Google is about average. Some companies at the bottom have near-zero AI adoption and can't even get budget for AI. They may have moats and high walls, but the horde is coming for them all the same. And then there are a few companies I've met recently who are *amazingly* leaned in to AI adoption. One category-leader company just cancelled IntelliJ for a thousand engineers. That's an incredibly bold move, one of many they're making towards agentic adoption. In my opinion, that company is setting themselves up for a _huge_ W. As for the rest, well, it's the Great Siloing. Everyone's flying blind. With nobody moving companies, no company knows where they stand on the AI adoption curve. Nobody knows how they're doing compared to everyone else. Half of them just check a box: "We enabled {Copilot/Cursor} for everyone!" Cue smug celebrations. They think this is like getting SOC2 compliance, just a thing they turn on and now it's "solved." And they don't realize that they've done effectively nothing at all. All because of a hiring freeze.

English
3
0
1
862
Daniel Imfeld
Daniel Imfeld@dimfeld·
Tell Claude to generate a walkthrough of a PR and then followup prompt to find issues, and the issues are mostly nitpicks. Fresh session that just finds issues, and it does a proper review. It's really interesting how previous context can anchor the model to what's normal.
English
0
0
1
93
Daniel Imfeld
Daniel Imfeld@dimfeld·
@jdnoc oh sorry this is for the Codex CLI. Just realized you might have meant the MacOS GUI app, and I don't know the answer there
English
0
0
1
46
Daniel Imfeld
Daniel Imfeld@dimfeld·
@jdnoc I just build it myself: 1. Install Rust 2. git clone the repo 3. cd codex-rs/cli 4. cargo install --locked --path .
English
1
0
1
96
Jordan O'Connor
Jordan O'Connor@jdnoc·
How do I get on the codex app dev / nightly build?
English
2
0
4
599
Daniel Imfeld
Daniel Imfeld@dimfeld·
@diptanu They bought some of Windsurf's team, which then went on to build Google Antigravity. The actual Windsurf product was bought by @cognition
English
1
0
1
85
Daniel Imfeld
Daniel Imfeld@dimfeld·
@AlecStapp Wow thank you for this. Got everything done including rental property, multiple K1s, and charitable donations in 2 hours. With TurboTax I had been spending two evenings to enter all that info.
English
0
0
4
457
Alec Stapp
Alec Stapp@AlecStapp·
This may sound like a sponsored post but I promise it's not (I just hate TurboTax and their rent-seeking lobbyists with the passion of a thousand suns): Use FreeTaxUSA instead. It's >90% cheaper than TurboTax and easier to use/higher quality software.
Alec Stapp tweet media
English
192
467
6K
637.2K
Daniel Imfeld
Daniel Imfeld@dimfeld·
Codex in sandbox mode refuses to write .git and it's hardcoded. If you don't mind building your own you can disable this with a quick patch. github.com/dimfeld/openai…
English
0
0
0
95
Daniel Imfeld
Daniel Imfeld@dimfeld·
So many SaaS products now selling cost tracking for your agents. I should really write an blog post about how we did automatic LLM call cost tracking with an AI SDK middleware, which is linked to each agent step. It's not that hard to do and you retain full control of your data.
English
1
0
0
64
Daniel Imfeld
Daniel Imfeld@dimfeld·
As much as I like Bun, every single new release breaks their bundler in a different way and I have to add some other workaround. Getting pretty tired of this
English
1
0
0
116
Jonathan Chang
Jonathan Chang@ChangJonathanC·
@jxnlco So the short answer is: 0.117.0 got slower because it flipped the default from legacy TUI startup to app-server-backed TUI startup. If you want, I can turn this into a concise upstream issue with the exact repro and timings.
English
2
0
5
345
Jonathan Chang
Jonathan Chang@ChangJonathanC·
codex takes ~3 seconds to start, claude code takes < 1 second. @jxnlco please fix
English
2
0
5
4.8K
Daniel Imfeld
Daniel Imfeld@dimfeld·
svelte-bench v2 should be for things like "does it not tru to just use $effect for everything" and "does it know that $derived is writable" Harder to test but these are by far the most frustrating things. Claude doesn't even believe me the first time I tell it that derived is writable
English
0
0
0
75
Daniel Imfeld
Daniel Imfeld@dimfeld·
@rsms I’ve been experimenting with pretty much the same thing. Also too early for me to say definitively but I like it so far.
English
0
0
0
224
Rasmus Andersson
Rasmus Andersson@rsms·
A workflow I've found useful lately for agent-driven programming is to have an extra level of git-repo indirection locally: a bare git repo + many local clones that pull/push to that one bare repo. The bare repo push/pull to a central repo like github. I.e. ~/src/foo-bare # git remote on different computer ~/src/foo-a # git remote is ~/src/foo-bare ~/src/foo-b # git remote is ~/src/foo-bare ~/src/foo-... The -a -b etc clones are like tables in a workshop. Pick an empty table when you start a new project for foo, then keep working on that table. When you're done, clean the table. Reuse the -a -b ... clones. What this enables is both offline work as well as merging/sharing changes across projects very quickly and without the need to push them to something like github. Maybe there's a better way and I'm only starting to try this approach out, but it's interesting enough to share I think!
English
15
1
80
13.6K
Daniel Imfeld
Daniel Imfeld@dimfeld·
@rsdgpt Upvote github.com/anthropics/cla… and hopefully they’ll make a way to do it For now I just did alias claude=“claude --model opus” and similar alias for sonnet to work around it
English
0
0
0
34
Ryan D’Onofrio
Ryan D’Onofrio@rsdgpt·
why does switching my model in one claude code session change it for all of them. do sandboxes solve this?
English
1
0
0
220
David Crawshaw
David Crawshaw@davidcrawshaw·
@steipete @dpetrou agreed! And if you start a checkout with ‘cp -c’ (macOS) or ‘cp —redline=auto’ (linux xfs/zfs/btrfs) you don’t even pay for storing the extra bytes.
English
2
0
12
1.6K
David Petrou
David Petrou@dpetrou·
ok, @steipete , curious about two things: 1. why no worktrees? how do you avoid multiple agents stepping on each other in the same repo otherwise? (for me, conflict resolution on pulls works.) 2. assuming the reports of you using pi were correct, why did you switch to codex cli?
English
5
1
69
45.7K
Daniel Imfeld
Daniel Imfeld@dimfeld·
Feels a bit silly but the biggest win for me with gpt 5.3 over 5.2 is that it actually prefers to run tests now. Haven't seen "Tests not run (not requested)" since it came out
English
0
0
1
150
Daniel Imfeld
Daniel Imfeld@dimfeld·
To combat the deluge of AI-assisted PRs at work, I have a new PR review workflow: - Forked diffview.nvim to support Github-style review mode - Claude command that reviews the PR and generates a guide for me - Skill that reads my comments and creates a PR review Been just starting to try this out but so far it's saving me a ton of time.
English
1
0
0
98