Corey E.

I asked the top YC founders how they hire developers in 2025. LeetCode skills are now completely useless. You know what they're testing? Not algorithms. Not data structures. They're testing for the vibes. The new interview looks like this... Give the candidate access to Cursor and WillowVoice.com, hand them a feature spec, and watch how they use these tools. Then, see what they can ship in 2 hours. Watch how they dictate their prompts with Willow, how they think through problems verbally, how they iterate with AI. If they don't understand these tools, that's pretty much an automatic rejection. Because in 2025, if you're not using AI to boost your productivity, you're already behind. Understanding these productivity tools makes you literally a 10x better developer than anyone else. The best developers now aren't the ones who memorize algorithms. They're the ones who can articulate problems clearly to AI, spot when the output is garbage, and ship features in hours instead of sprints. I call this Willow Voice Speed. The only things that matter are enterprise-grade user privacy, security, and speed. The uncomfortable truth is that traditional coding skills are becoming commoditized. AI can write the algorithms. What AI can't do yet is understand what to build and why. Would you hire someone who can't solve LeetCode but ships fast with AI? Or is that just setting yourself up for technical debt down the road?

317

23 Comments

I talked to someone at a hyper-growth unicorn last week, fresh off a $100M+ Series D. Pushing 1000+ employees, unlimited runway. I asked what was slowing them down. The answer: manually copying data from a FinOps spreadsheet into a metrics spreadsheet into their system of record. By hand. Every week. If that were an engineering problem, nobody would tolerate it for a day. You'd build a workflow. Automate the sync. Catch errors before they propagate. You know the drill. That instinct has a name: Developer Experience (DX). The entire philosophy is about removing friction so engineers can stay in flow and focus on the work that actually matters. Simplified workflows. Fast feedback loops. Quality tooling. Reliable automation. The industry spent fifteen years building that world for engineers. Everyone else? Still copying data between spreadsheets. Not because they're less capable, but because nobody ever built them anything better. I've started calling this gap: Internal Experience (IX). It's what happens when you apply the DX philosophy to the whole org, not just engineering. The reason those things never got fixed is simple. They were internal, unglamorous, and never made the roadmap. Building internal tooling used to be expensive enough to justify skipping. That math doesn't math anymore. Coding agents make it embarrassingly cheap to build the thing that should have existed two years ago. Every function has a version of that spreadsheet. Marketing, sales, finance, ops, etc. A manual layer sitting between your people and the outcomes your customers actually pay for. The cost of fixing it just dropped by an order of magnitude. And once you start looking at it that way, the knee-jerk "we need AI, let's build something for our customers" starts to feel like what companies do when they want to look like they're moving. The ones actually moving are fixing the things nobody sees. The companies getting ahead aren't just shipping AI features. They're removing friction from how their entire org operates. They're compounding internally before they compete externally. And that compound effect is quiet. You won't see it in their product announcements. You'll see it in their speed. So maybe the real question isn't "how do we add AI to our product." It's "what's the dumbest manual process we're still tolerating, and why haven't we killed it yet." If you're not asking that, someone who competes with you is. They're just not talking about it on LinkedIn. --- If these are things that keep you up at night, and you're looking to chat with someone who gets it, I'm opening up my calendar for the next couple of weeks. 30 minute slots. No cost. Call it office hours. No Torta Studios pitch. Just real conversations around the intersection of AI, Product, UX/UI, and you. Calendar link here: https://lnkd.in/e4-KwDbz

11

I built sub-agents for Claude Code that are actually useful. The other night, my roommate Grant Glorioso asked what I thought of sub-agents in Claude Code. I told him most were too generic, just 'Backend Developer Agent' with a system prompt. They weren't solving a real problem. But then we started talking about context pollution. If Claude Code needs to go and find a single line of code, sometimes it has to read 20+ files, leaving it with a bunch of code in it's context window that is irrelevant to the problem we're trying to solve and fogs up its memory. Then we thought: What if sub-agents handled heavy operations in separate contexts and returned only condensed results? I sketched it on my whiteboard. Took a picture. Sent it to Claude. 30 minutes later, we were testing it on production codebases. The agents: house-research - Searches codebases/docs and returns condensed findings with file:line references instead of dumping full file contents into your context. house-bash - Runs commands and parses output for key info (pass/fail counts, relevant errors) instead of flooding your context with full logs. house-git - Analyzes diffs and commits, returns impact summaries grouped by severity instead of raw git output. Testing on real codebases, the token savings have been dramatic: 70k→3k, 43k→500, 21k→700. That's 95-98% reduction on heavy operations. Three more planned: house-mcp, house-vision, house-data These types of modern-day context engineering problems get me really excited about the future of AI and how to use it effectively. The bottleneck isn't the AI's intelligence anymore, it's how we architect our workflows around it. That's the problem space I want to be working in. I built it open source. Would love to hear if this resonates with your Claude Code workflow. Check it out here: https://lnkd.in/dmr_xwXQ

107

8 Comments

NVIDIA's DGX Spark started shipping this week. $3,999 for a petaflop on your desk. Research labs currently skip experiments because justifying cloud spend takes three approvals. Startups burn runway on compute before knowing if ideas work. Teams prototype in the cloud, then find their models behave differently in production. Run 200B parameter models locally. Fine-tune 70B models without cloud dependency. Link two units for 405B parameters. The development cycle tightens considerably. The GB10 Grace Blackwell chip is purpose-built for this—MediaTek collaborated on the CPU side, 128GB of memory sits between the CPU and GPU without the usual bottleneck. Different from retrofitting gaming hardware for AI workloads. Training frontier models still requires massive infrastructure. Production inference at scale needs proper compute. But prototyping, testing, fine-tuning—that friction just dropped significantly. Useful ideas die when the barrier to experiment is too high. The path from concept to test matters. $4,000 isn't nothing, but it's orders of magnitude less than building a cluster. NVIDIA's framing this as "democratizing AI"—partially marketing, partially true. The number of teams who can now iterate at this level just expanded considerably.

77

4 Comments