building things, ex founder @seedlysg @usemitohealth, yc alum, tatler gen.t leader and prestige 40u40

🇸🇬/🇺🇸
🌉 After 9 years, I’m thrilled to be back in San Francisco for @ycombinator S24. Yesterday marked the start of the batch retreat in Sonoma, and I've already met incredible companies, batch mates, and YC partners. I'm excited and looking forward to lots of learning during this retreat - Unconferences, group sessions, and more!
8
8
195
26,812
Opus 5.5 made me realise something. I'm not really comparing models anymore. I'm comparing the whole setup around them. Model quality, remote control, subagents, limits, context, how much babysitting it needs. The harness matters a lot now.
2
36
If you could only pay for ONE right now: Codex or Claude Code? After Opus 5.5, this is a much harder question for me than it was a few weeks ago.
1
3
70
Subagents are great for independent work: research, review, separate investigations. For small edits or tightly coupled steps, keeping one agent in context is usually simpler.
4
70
For people using both Codex and Claude Code: What would actually make you switch fully to one? Model quality? Usage limits? Remote control? Speed? Better subagents? Less babysitting? For me, the workflow is starting to matter almost as much as the model.
1
3
107
I realised my AGENTS.md / CLAUDE.md keeps getting longer. A useful check: “Would I still want the agent to follow this for a one-line fix?” If not, it probably shouldn’t be global.
1
4
95
Did Anthropic just take the lead back with Opus 5.5? Been pretty impressed with it so far, especially for coding. Feels like the Codex vs Claude race just got a lot more interesting again.
1
2
63
If you’re a paid Claude subscriber, claim your free credit here.
Replying to @ClaudeDevs
Follow the link below to claim the credit or run /claim-credit in the CLI. You’ll need GitHub connected to start a session. Claim by Oct 7. Terms apply. claude.ai/code/claim-credit/…
2
138
One thing I like more in the Codex app than Claude Code is remote control. Being able to kick off work, leave, then check in or steer it from my phone is really useful. If Claude Code nails this too, Opus 5.5 + proper remote control might be enough to pull me back fully.
4
5
159
Lol I resubscribed to Claude Max and Opus 5.5 is making me feel better about it already. Been using it for coding and it barely made a dent in my usage. Feels way more efficient so far. Anyone else seeing the same?
1
53
Great day to be building with AI. > Opus 5.5, GPT-6 Astra, Sol, and Luna just dropped > Fresh usage limits just reset Productivity is about to go through the roof. Time to cook.
1
102
OpenAI's new Agents API exposes the Codex harness, including context management and subagents. I'm curious how much of my current agent setup I could replace with this instead of maintaining the orchestration myself.
2
94
Before asking a coding agent to fix a bug, try this: "Reproduce it. Write a test that fails for this reason. Make the smallest fix. Show it passing. Leave unrelated code alone." Much easier to review than a surprise refactor.
3
3
96
Been dropping this prompt into my repos to see where Jev could actually be useful. Try it:
2
1
77
We’re evaluating whether Jev could meaningfully improve this product. Research Jev in depth as of September 2026. Start with: typesafe.ai typesafe.ai/blog/introducing… Then search the web for anything else relevant. Next, inspect this codebase and identify where Jev could improve existing features, make workflows faster, cheaper or more reliable, unlock new customer-facing features, enable new agent capabilities, or automate backend and internal workflows. Do not focus only on replacing an existing model with Jev. The more important question is: What can we build now that we probably would not have considered building before Jev existed? For each strong idea, explain what it is, why Jev is a good fit, how it would integrate into this codebase, the expected impact, implementation effort, risks or limitations, and whether our current stack could achieve the same result without Jev. Rank the ideas by value versus effort and recommend the top three to prototype first. Do not modify production code yet. Research and recommend first.
61
I posted about using ~$25K worth of tokens through Claude Max + ChatGPT Pro. The replies: X: “NGMI” Threads: “WTF are you doing with $25K worth of tokens?” Same post. Completely different vibes. 😂
1
73
If you’re interested in the solo founder journey, check out @Buildingitmyway’s video. I met Maxime earlier this year and have been impressed by everything he’s been building since. This guy has no shortage of ideas and always seems to have something interesting going on. Worth a watch 👇
aaaand we're live on Youtube! have been documenting my journey as a solofounder journey in Singapore: projects, lifestyle, investor meetings, strategy, wins, failures... hope you'll enjoy it 🍿 piped.video/watch?v=sDxEkMn4…
1
1
1
224
Goodbye to my trusty co-founder after 9 months of building together 👋
It’s been a good run, Claude. 🫡
116
It’s been a good run, Claude. 🫡
Claude’s limits are getting ridiculous. I’m seriously thinking of switching away. Already on Codex Pro 20x, so if I’m moving another $200/month elsewhere… Where would you put it?
1
1
8
7,270
Started building this app a few months ago. Today it’s at DR 74. Mostly from experimenting with SEO, backlinks, and pages that compound over time. Maybe I know a thing or two about SEO after all 🤔
1
4
159