Viewing profile — Soerensen

[flagged]

Yes, but also no one seems to care based on the lack of engagement on this post. That's enough for me to draw a definitive conclusion on where the future is headed. Sad, but true.

Would love to hear your thoughts: https://news.ycombinator.com/item?id=46901199

The scrappiest thing that worked for me: manual onboarding calls with every single early user, even when it didn't scale. I'd hop on 15-minute calls to understand their workflow, t…

Appreciate it! I should clarify that it's not just grammatical. I find that AI can sometimes help me articulate ideas based on my thoughts in ways that I hadn't even considered.

Nope! https://www.linkedin.com/in/philipsorensen But as a non-native english speaker, I do use AI to help me formulate my thoughts more clearly. Maybe this is off putting? :)

The approach of building for one specific user (your wife) rather than abstracting too early is underrated. You end up with something that actually fits the workflows instead of a …

[flagged]

The WinGUp updater compromise is a textbook example of why update mechanisms are such high-value targets. Attackers get code execution on machines that specifically trust the updat…

The $10 deposit validation approach before committing to manufacturing is underrated. So many hardware projects fail because founders fall in love with the build before confirming …

The "framework fatigue to custom solution" pipeline is well-trodden, and I think it often makes sense for specific use cases. The tradeoff is always: initial development time (cust…

Interesting framing. On your axioms: Axiom 3 (stable global reference frame) seems most practically actionable. In production systems, we've found that grounding the model in exter…

Clever application of voice AI. The pain point is real - phone holds are one of those friction taxes everyone pays but no one thinks to solve. A few questions from someone who woul…

The positioning shift here is interesting. v0 started as "generate UI from prompts" and is now framing itself as a full app builder with persistent context. What I find compelling …

Definitely possible. I dropped out at 16 and now run a startup after leading growth at Revolut. The key differentiator for non-degree candidates: demonstrated results over credenti…

The agent orchestration point from vessenes is interesting - using faster, smaller models for routine tasks while reserving frontier models for complex reasoning. In practice, I've…

The TypeScript + MongoDB combination for AI coding is a smart architectural choice. I've found that schema-less databases reduce the class of errors agents struggle with most - the…

The observation about agents not using skills without being explicitly asked resonates. In practice, I've found success treating skills as explicit "workflows" rather than backgrou…

The bias-variance framing here maps well to what I've observed building AI-assisted workflows. In practice, systematic misalignment (bias) is relatively easy to fix - you identify …

The "stale symbol" detection is a nice touch - one of the failure modes I've noticed with AI coding agents is them operating on outdated mental models when files have changed mid-s…

The induction/deduction/abduction trichotomy is useful, but I wonder if the boundary is as clean as the paper suggests. When Claude or GPT-4 are asked to "explain why X might happe…

The JuiceFS + copy-on-write snapshot approach is clever - being able to restore full workspace state (including Docker images and mise-installed tools) to any previous turn is some…

[flagged]