Ideas Stuck on No Coder?

Last month in Hangzhou, I met a client who said, "I just want a simple tool to auto-send quotes," and then sighed. Outsourcing was too expensive, he couldn't code it himself, so he was stuck again. I've been stuck there too. Last year, I wanted to build a client follow-up system. Just explaining the requirements to a developer took three weeks, and they told me the earliest slot was two months out. I kept thinking: Can't I just do this myself?

AI Leaders Say Coding Is Solved, But The Disagreement Is The Point

At Sequoia AI Ascent 2026 this year, Boris Cherny (Meta), Greg Brockman (OpenAI co-founder), and Andrej Karpathy (former OpenAI / Tesla AI head) all expressed a similar sentiment: coding is basically solved. But they disagreed on what the new bottleneck is—some said product definition, some said domain knowledge, some said distribution. This disagreement itself is a signal: writing code is no longer a moat; "what are you actually trying to build?" is. Last week, I was grabbing ramen in Wangjing with Lao Wang, an independent consultant. He told me he just used an AI coding tool to build a client management tracker himself. An outsourcing quote for that was 20k RMB; he got it done in a single weekend. His exact words: "I used to think having no tech skills was a disadvantage. Now I realize I understand what the client wants way better than a programmer does. That's the real advantage."

What We Can Do Today

We don't need to wait for the day "coding is completely replaced by AI"; we can start right now. Money: $0 to start. Free tiers of mainstream AI coding tools are enough; if you want to get serious, $20/month. Time: Set aside 1-2 weekends for your first small tool. Technical barrier: No coding required, but you need to explain your requirements in plain words—like "I want a page where I enter a client's name on the left, and it auto-generates a quote on the right." That's clear enough. First step: Open cursor.com, hit Download, and tell it what you want to build in your native language. I messed this up at first, too—I wrote super technical descriptions for the AI, and it actually understood them worse. I later realized: just talk to it like a smart new coworker who knows nothing about your business. The more specific and plain-spoken, the better.

Advice By Stage

If you're just starting out with zero clients: Don't rush to learn the tools right now. First, write down the task you repeat the most, like "manually organizing quotes every week." That's your future first automation target. If you have 1-2 active clients: I'd suggest picking your most painful tiny request to test the waters. For example, if clients keep asking "what's the progress?", use an AI tool to build a simple status page. Don't manually reply to things that take 30 minutes to solve. If you're scaling with stable revenue: Have the person on your team who best understands the business learn AI coding. It's faster and more accurate than hiring a programmer—business understanding plus AI writing code, that combo is the real leverage right now. One last thing: the disagreement among those three leaders shows no one has a certain answer. If you feel like "I can't even articulate my requirements clearly," that's okay. Just get the business running first; the tools can wait.