Rogu retweeted
People run their businesses on Grok Bot now. But like every voice agent, it can't hear you over a loader.
So we made it a music video.
~Mara
IMO, business reasoning is understandable (I assume the plan was to switch to Grok 4.8, but it was not ready and good enough in time).
But lack of communication and silent change under the cover of a high was … not good. Clients deserve honesty and upfront communication. Otherwise it’s lack of respect.
The more chances I give to dots the more I hate them. One of the most infuriating products I ever tried in my career.
Grok Bot is great. Likely my most favorite app of this class.
Grok Voice is awfully bad, especially compared to GPT Live. It’s just not pleasant to talk to unless you absolutely must.
Honestly, Dots have been surprisingly, awfully bad experience. They are INCREDIBLY stupid and restricted. I have been very in favor of OpenAI for the longest time and have their $200 sub from Day 1 of its announcement (same as I had $20), but Dots did put a strong dent in my trust in them for the first time.
Grok Bots currently have the best implementation on the market, with Poteto / Lauren calling them the Ship company in a well-deserved manner.
Muse is second, but their model is quite poor.
Maybe even Hark is somewhere here. Along with Instinct etc
…and then there are Dots. :(
Tried Grok Bot. Great UI. But I believe it does the same thing that (IMO) killed Telegram bots. Focuses too zealously on text as the only communication medium.
I wouldn’t thought that I would do it, but … I switched to Chrome on Mac.
Safari after Liquid Glass is extremely ugly, unpleasant to use, and slow. I tried to love it and failed. Arc/Atlas are dead, other neobrowsers (like Dia, Aside) lack polish or any good reason to use them.
Rogu retweeted
I spent some time with the newly open-sourced Grok Build repo: 1.07M lines of Rust, 85 crates, one squashed commit, no PRs accepted.
Out of the box, `cargo test --workspace` doesn't compile. 190+ errors, all one bug class: cross-crate test helpers hidden behind #[cfg(test)], which Bazel's per-target test builds tolerate but Cargo doesn't, because a dependency is never compiled in test mode. The default-bazel feature is declared in ~20 manifests and wired to nothing. Ungating the benign helpers and putting the signing-key test seam behind a dev-dependency-only feature gets the suite running for the first time on the published tree: 24,663 passed, 28 failed, and every one of the 28 is a pre-existing bug the broken build had been hiding (tests reading your real ~/.claude/settings.json, a tool missing from the registry, macOS /var symlink breakage, one Theme::current() race).
Questions for the team:
- Did anyone try downloading this repo and running it before publishing?
- Why not accept pull requests as a public version?
- Half of the code is tests, but the tests don't run out of the box. Which is it? tested or not?
We've open-sourced Grok Build and have reset usage limits for all users.
Open sourcing Grok Build allows anyone to support making a reliable and robust harness. Check out our code, including the Git repo for the Grok Build CLI.
x.ai/open-source
You cannot have it both ways. You either claim you have truth seeking AI that always traces to grounded verifiable sources.
Or you have a corporate mouthpiece, where users know that you can always inject whatever narrative or “truth” you want popularized now.
IMO that narrative forcing into Grok is worse than Grok Build fiasco. Because this is clearly intentional output skewing into vendor’s preferred direction. Not objective truth seeking anymore. Hence you cannot trust Grok.
If they can force you to say these invented non-verifiable facts because of their fuck up that only raises questions what other narrative overrides @spacexai could put into you.
Really the cure is worse than the poison.