Johnny

Founder (exited). Builder across product, dev, and health & fitness. Now deep in the AI-native shift, looking for the next thing worth going all-in on.

[Johnny @johnnycon]

2nd time @clerk has a major outage in the last couple months. Maybe we should be self-hosting auth after all.

[Johnny @johnnycon]

@ritu_twts They’ve done the best job with marketing (in the vibe era) and were very early to the MPC game. Anyone who isn’t a traditional developer will likely end up on supabase if they let the model choose. Experienced devs usually don’t choose supabase.

[Johnny @johnnycon]

@conductor_build Found it. It was an expired github PAT (token). Started working after I generated an updated one.

UI could definitely add some UX around this scenario.

[Johnny @johnnycon]

@conductor_build any idea what would cause this? Conductor is stuck on this message. I'm unable to create PR requests on any projects. I've closed/reopened conductor, and removed/re-added projects. This is happening on all projects, not just one.

[Johnny @johnnycon]

@conductor_build On the latest version btw.

[Johnny @johnnycon]

@thsottiaux Came a little late. I was tokenMAXXing yesterday per your suggestion (and evening reset message), and hit my rate limit at midnight. So had to upgrade from the 5x to 20x plan 😭 . It's fine though, it was gonna happen with /goal anyway.

[Johnny @johnnycon]

@DanielSmidstrup what kind of front-end design? web? ios? android?

i've not had luck with GPT 5.5 on design. I definitely prefer Opus there.

I prefer GPT 5.5 for complex code task, and opus for everyday code on popular frameworks (NextJS, TypeScript). GPT 5.5 for things like Go Lang.

[Johnny @johnnycon]

running my first long ( > 1 hour) codex /goal.

it's now 13+ hours in and still seemingly working towards the correct goal. Hasn't gone astray yet. We'll see if the result meets at least 70% of my expectation.

[Johnny @johnnycon]

Woke up to one of my Anthropic API keys actively being used at about $30 every 10 minutes. Turns out it was from the @braintrust hack 😭 . Stopped it about an hour in (when I woke up), $189 in damages. Any recourse, or just par for the course?

[Johnny @johnnycon]

@MendyOK @theo @conductor_build Isn’t that what big terminal mode is?

[Johnny @johnnycon]

@Shpigford I’m also a big time conductor user. I recently went from $200 /mo Claude / $20 /mo OpenAI, to $100 / $100 split. I’ve never hit my limit on any plan, so we’ll see if Claude is still usable after the switch. But yeah, could drive away devs in droves.

[Johnny @johnnycon]

@charlieholtz I haven’t tried big terminal mode yet. Going to give it a go now.

[Johnny @johnnycon]

@mattyp The timing is perfectly accurate.

[Johnny @johnnycon]

@charlieholtz My workflow recently has been to open up a GPT 5.5 and Opus 4.6 and give them the same prompts and work through them simultaneously. For my code (mostly swift, go, and infra), they seem pretty neck and neck.

[Johnny @johnnycon]

@charlieholtz @ycombinator @aaron_epstein Love that!

[Johnny @johnnycon]

@charlieholtz Nice. Will give the html method a try once that next release drops.

[Johnny @johnnycon]

@lucasbastianik I use conductor as my daily driver. Haven’t noticed anything broken thus far. Has been working really well for me since the update.

[Johnny @johnnycon]

First time this year that Opus is not my primary coding model. GPT-5.5 is now my preference for 70% of tasks.

Still using Opus every day (4.6 though), but if i'm working on something complex, it's GPT 5.5 all day.

[Johnny @johnnycon]

@AdamTornhill The speed of decision making is definitely much faster now (with Agents coding). A lot of slow thinking used to happen while writing the code, now that doesn't exist, so it's 10x the decision making compared to the old model. I love it, but it is a big shift.

[Johnny @johnnycon]

@zuess05 It's less about the code and more about the system. There is a good balance to be struck. Vibe coding without proper steering, guidance and systems thinking will eventually bite you with slow velocity, slow iteration cycles and wasted tokens later.