Last Thursday afternoon, I froze for 3 minutes

Last Thursday afternoon I was at Xixi Coffee shop writing a proposal when I saw a piece of news — the CEO of some SaaS company had cut his dev team by 70% and replaced everyone with AI. I figured it was a joke at first, but clicked through and — it was real.

Even crazier, the next day: a bunch of the laid-off folks uploaded a project called OpenExecutive on GitHub, having AI do the CEO's job directly. Open source. Free.

My first thought wasn't "AI is amazing." It was: how many years away are we small bosses from this?

What's actually going on, and who's using it?

The short version: a CEO named Sam (alias) ran a mid-size SaaS company, originally 30 people. At the start of the year he made a bet — cut backend, frontend, QA; keep 8 product and ops folks; let AI write code, talk to customers, send weekly reports for the rest.

Three months in, the business didn't collapse. It grew 12%. Sam posted on Twitter: "What a CEO does every week — 80% is meetings and emails."

Among the 22 laid-off folks was someone called Xiao Wang (former Alibaba P7 engineer, anonymous), who led the OpenExecutive project: break a CEO's weekly work — reading data, running standups, writing OKRs, talking to investors — into 200+ small tasks, hand them off to multiple AIs to collaborate. GitHub stars are already at 12,000.

I'd always assumed "AI replacing CEOs" was marketing copy. Looking at those star counts now, I have to admit: this is actually happening — and the barrier is way lower than I thought.

If I want to mess with it myself, how high is the barrier?

I spent two hours over the weekend trying it out. Let me break it open for you:

  • Money: Running it costs about 30–80 RMB/month using cheap models like GPT-4o-mini or Claude Haiku. If you're handling sensitive customer data, you'd need your own server — separate cost.
  • Time: 2–4 hours to get started. You need to feed it your business info (what you sell, who buys, what you do weekly).
  • Technical bar: The GitHub project page has a one-click deploy button, but you need to know what an "API key" is — think of it as the key to AI. Without it, you can't get in. If you really don't get it, search "OpenExecutive deployment tutorial" on Bilibili — tons of step-by-step videos.
  • First step: Don't rush to deploy. Open the project page github.com/SenteLabsAI/OpenExecutive and read the README for 10 minutes. Then decide if you want to go further.

Honest take: it's fine not to try it now. Not everyone needs this tool — especially if you haven't hit 10K RMB/month revenue yet. AI-as-CEO is a burden at that stage, not a help.

Advice for three types of friends

Just starting out (0 customers / side project testing): I'd say skip it for now. AI-as-CEO is a "save time" tool, but your most valuable thing right now is figuring out whether customers actually want to buy. Get through 10 real conversations yourself first.

Have 1–2 stable customers: Spend a weekend playing with the OpenExecutive demo — but only let it do one thing: auto-organize customer feedback every week. No meetings, no decisions, no sending emails. Treat it as an AI assistant first.

Scaling up (3+ people, stable revenue): Worth a serious look, but I wouldn't hand over the CEO job wholesale. First, list out what you do every week. Mark which tasks are "info organizing" (AI's strong suit) vs. "decision-making" (AI's weak spot). Let AI do the first 30%; you handle the last 70%.

Real talk: I used to think "AI replacing CEO" was a big-company problem. Looking now, that line is already at our feet. Panic? No need. Pretend you don't see it? Also no.