@jasonfried Still lot more clicking in 5. Why not let user select all they want from the first + popover and Save?
English
Bharat Mandava
381 posts
First I critique something particularly embarrassing in Basecamp 4, and then show you how we completely redesigned the same flow in Basecamp 5. We grow!
English
Designers give me one Font you can use for the rest of your life
English
@BHolmesDev Quick/repeatable tasks: Claude
Heavy/reliable tasks: Codex
English
I’ve used Opus 4.6 and GPT 5.4 on a mix of projects since release, and want to break down where I think they uniquely excel. It’s more nuanced than you’d think!
Rigor of code - GPT 5.4. It goes the distance validating its work without asking. Opus needs explicit instruction to do this, and even then, it misses more edge cases.
Clarity of code - Opus 4.6. Claude is a better communicator, which carries into the code. Variable names are clearer and less mechanical, which improves reviewability. This is very important since code review is the bottleneck for most engineering teams. It also adds the right amount of doc comments. GPT simply never comments or explains its work; it’s like working with an obtuse engineer that wants the solution to speak for itself. Sometimes it does, other times not.
Similarly, rigor of plans goes to GPT 5.4, while clarity of plans goes to Opus 4.6. An interesting point though: GPT performs better talking through a strategy without a plan, while Opus needs planning mode to put in any rigor. I find myself forgetting plan mode altogether using GPT 5.4.
Quality of research - toss-up. Opus spends longer researching with web search, but GPT spends longer studying the existing codebase. You may think codebase research matters more, but researching how others solve the same problem can be just as important. Maybe more important for greenfield.
Quality of conversation - Opus 4.6. It’s just better to talk to, which matters using these things everyday. GPT 5.4 was clearly trained to challenge the user more, which results in a tendency to *always* say you are wrong. I’ve had bizarre interactions where GPT claims something is “not quite right,” the restates exactly what we’ve decided on in the last turn. On a personal level, it’s annoying. On a practical level, it makes iteration on a plan slower. THAT SAID, it takes sufficient pushing for Opus to challenge your thinking in this way. Simply say “I’m impartial” and ask questions to avoid that, as you would a person.
Overall winner - Opus to make it work, GPT to make it good. I don’t have a good system of when to switch tools, but on average, I prefer Opus early on and GPT for optimization and discussing architectural decisions. Opus is also better for any design related tasks (but state management in frontend apps is better handled by GPT).
English
@rezoundous Thats cause @thsottiaux resets the usage every day 👍🏼
English
@varunram Same as how Costco comes up with Kirkland version of its best selling products.
English
At this point, its pretty clear that if you are an app layer company using Claude Code SDK, it is inevitable Anthropic sees your usage and then develops that tool in house
English
@hey_yogini Deliverability - thats what matters the most.
Everything else is just noise.
English
Had a long debate in office today.
After noticing patterns from our users, we realised the biggest problem with email automations is decision fatigue.
What to send, when to send, will it land in spam? is anyone even reading this?
So we went through every click, every button, every screen. One question only: does this give the user more confidence or less?
Are you running email automations in your product? If yes, what's the biggest challenge?
English
@kylegalbraith @PlanetScale They have the best onboarding experience as well 💯
English
I can say this as a @PlanetScale customer: there is literally nothing better.
Every single person I've interacted with there knows their shit, unlike anyone I've met before them. Getting phenomenal support as we work through some scaling things we need to do for our next launch.
English
Zod isn't the only option.
Our team uses decoders, built by the wonderful @nvie. We find it's:
• Easier to read, as it looks like TypeScript.
• Quicker to debug, with its human-readable error messages.
• Lighter to ship, with tree-shaking and smaller bundles.
Get it ↓
English
@GregorySchier This is the right pricing model for Yaak. Mix of life time, personal and commercial licenses.
Experiment with pay what you want by keeping the current lifetime price as minimum and you’ll be surprised.
English
Turns out adding a lifetime license was a great idea. Selling like hotcakes!
I'll likely increase the price or discontinue lifetime in the future, but it's doing great at funding current development (what I was hoping)
English
@_skris Looks good. Still see the loading indicator. Consider using a "preload intent" so the transition feels more snappy.
English
this video is not sped up
Sai Krishna ⚡️ Superblog.ai@_skris
moved superblog dashboard to aws just couldn't deal with @railway going down every now and then, almost lost a precious customer after they faced downtime multiple times. they just said - sai, superblog is big now (narrator: it is not). you can't keep experimenting with infra. thankfully they did not churn! the dx and quality (when it works) of railway are fire. nothing comes close. but help-tickets unanswered, strange errors and this one time, I had to deploy a minor fix but had to wait for 4 hours due to their build system failure. absolutely no communication in the dashboard. the status page is weirdly built (shows 'all systems functional' even during the downtime of one of their systems. flagged by multiple people) i actually considered going back to @vercel. seemed no brainer. pay 20/mo and forget everything. in case of ddos, ceo actually responds. except the cold starts. fkn cold starts man. then considered to use my fav @Hetzner_Online (hey vps gang). made up my mind to setup cf zero trust + caprover + waf. then realised rarely we need to reboot the vps after security patches (automated with unattended-upgrades but still). didn't want to be offline for 30-60s (i'm locked-in to grow). other option: use an lb and maintain two vps to alternate during patch+reboots. nah, not worth it. just looked at aws lightsail container vs app runner vs fargate. app runner seemed no brainer as a product. but god damn. the dx suuuuuuuucks. fkn sucks. connected a git repo but build fails because the provisioned instance config is not enough to run the build (but it is 2x the run time requirements). then had to setup github actions to build and push the docker image to ecr. then some stupid health check issue prevented the app from getting deployed successfully. for every change, damnnnn 20 mins wasted. finally fixed it. but then faaaak, wtf aws apprunner team. its been 4+ years since i suggested to allow adding env vars in bulk. still not implemented? literally every other platform has this. you dont even need claude/codex for this. i had to manually add 40 env vars one by one. who tf is running the product? @awscloud annnyyways, finally connected the domain to app runner. the dashboard is flynnnnn. db and app are both in eu-central-1. maybe even in the same fkn rack, who knows? seriosuly should've taken the hit long back and spent 2-3 hours to deploy to aws. wanna give superblog a try? superblog.ai (this is cf pages btw. login to dashboard actually)
English
@DorianDevelops yes and no.
Quick/repeatable tasks: Claude 4.6
Heavy/reliable tasks: Codex 5.3
Have to use both to get the maximum benefit.
Also, pass around the plan between them until both agree to it. Better outcome most of the time.
English
@yashhq_22 Both.
Quick/repeatable tasks: Claude
Heavy/reliable tasks: Codex
English
@arvidkahl If you are not using both Codex and Claude, you are leaving a lot on the table. Different models for different sets of tasks.
Quick/repeatable - Claude
Heavy/reliable - Codex
English
I don't really understand devs who jump back and forth between LLM providers. Claude, Codex, Claude, Gemini, back to Codex, then Claude.
What marginal improvement can be worth the massive context shift every single time?
Pick one and build. You won't have time to switch.
English
@jarredsumner The universe conspired to send the right person to the right place. Thank you, @jarredsumner, for sharing all this information. I’ve learned a lot from you over the years.
English
In the next version of Claude Code
Claude Code’s Bash tool runs faster and uses less memory. Claude writing 1 GB to stdout 10 times:
Before: 115 seconds
After: 15 seconds (7x faster)
English
@ajambrosino The Spark is fast. Keeps running into "stream disconnected before completion:" too often.
English
New in the Codex app:
- 5.3-codex-spark
- Forking
- Pop-out window
- Mark unread
- More perf and quality stuff
- (a secret fun thing)
- Tomorrow: first Windows alpha invites
English
@thelifeofrishi yes. Lazygit. or better use skills like code-review-excellence to review PRs from Codex/Claude.
English
For the past few months, a small crew got together to overhaul @NotionHQ sidebar / navigation. We called the project "Slippery Slope" :)
For anyone who's worked on a similar project knows how hard it is to make any changes to navigation, because it directly impacts how millions of people use your product every single day.
To my surprise and delight, the team was able to ship something remarkably better in a matter of weeks. We've been using it internally for some time now, and I cannot imagine going back.
We start ramping to users tomorrow, but feel free to reply here, and I'll try my best to turn on early access!
English