Steven Pu

I spent grades 5-8 in Salt Lake City, nicest people I have ever met in my entire life.

@dunkhippo33 💯 agree. For me personally, I don't like the idea of my future & fate being controlled by others (managers). In business, if you have a problematic customer, you can choose not to do business with them. In a job, you can't choose your bosses.

A lot of people fixate on money as the reason to start a business. But a bigger reason should be about controlling your time. As a founder, you can take off in the middle of the day to coach your kid soccer or do something fun when most of the rest of the world cannot. And then you can come back to it at 10pm or 2am and there's no hassle.

@ericbahn 🦞 coded

Clamp Hustle 2026 is a wrap. My heart is full. I'm totally exhausted. This was the very best one we've ever done. Thank you so much to all the campers who brought your authentic selves and such great energy. I look forward to seeing you all in 2027!

One thing to note, always consult multiple AI models when it comes to open-source tooling recommendations. A single model can often have blind spots, often from over-indexing on some guidance in its context.

. @claudeai: manually compacting context is the closest thing to manually compacting context, hmm... you don't say. 🤣

@bearlyai These meters give me PTSD.

Someone made a small hardware device for the desk that monitors Claude token usage. Calls it Clawdmeter. Kinda genius.

wow Claude Design is IMPRESSIVE.

Claude code doesn't know about Claude design 🥲

Very large chunk of problems generated by @claudeai code is due to insufficiently granular planning. Don't skimp on ultra-detailed upfront planning. List out every step, assumption, exception, use flowcharts, whatever you can. Debugging is 100x more grueling than planning.

Pinning package versions suddenly seem like a really good idea.

Trying to upgrade my @claudeai code plan. Claude won't let me, gave me a strange error "address_changed" and asked me to contact human support. Trying to talk to a human rn, 30 min waiting and counting...

@ericbahn @AndrewYang How do we realistically handle the transition period before the AI-driven era of plenty? Modern societies won't be able to handle such rapid structural economic shifts.

What should I ask our future President @AndrewYang today?

@garrytan Wow just did this last night. After asking AI to create skills on how to write specific scientific docs by ingesting samples, I realized I could ask it to further create a skill based on our session on how to ingest docs & create skills. Created a skill to create skills.

My viral meta-meta-prompting post about personal AI is directly a consequence of the thinking about and building with metaprompting in January and February of this year 👇

How to write a bot reply on @X. - Summary - Positive assessment to show it cares - Personal connection for relatability - Counterfactual strawman for intellectual elevation - Emoji to show it really is a human

@chrysb I actually always ask AI to create a mermaid.js ingestible flow diagram before it even writes any code, and then critique every flow first. In fact, all the code written are annotated with the node / flow numbers from the diagram. Otherwise it's just chaos.

i spoke to a founder yesterday - their CTO finally read their agent-made codebase after months and panicked when he realized it was impossible to understand wtf was going on my rule of thumb is: if your codebase starts written by agents, don’t try to understand it instead, align at the architectural level before any building happens, and ask the agent to maintain a living architecture diagram of how the system works there are three altitudes that matter: - Top-level: architecture - Mid-level: patterns & abstractions - Low-level: file-level code in today’s world, a CTO should be deeply concerned with #1. #2 matters too, but not as critical as #1. if #1 and #2 are dialed in, #3 is where most of the high leverage agentic gains live. as long as you understand the architecture and critical interfaces, it becomes much easier to reason about ground truth and meaningfully iterate understanding and informing the architecture / patterns / abstractions give your codebase maximum longevity and agent maintainability

Haven't tried @opencode yet but the team is hilariously unhinged. Worth trying the product just for this. 🤣