Coding Agents Have Passed the Singularity — What Should Humans Do?

Singularity in coding, reportedly

It has been reported by Huxiu (虎嗅) that agent-driven coding has passed a practical tipping point: what used to take teams weeks can now be produced in minutes. In one anecdote, a developer friend named Gui Cang reportedly has an agent that writes features and ships multiple packages a day — yet an Apple (苹果) code-signing step still takes an hour. Fast code; slow operations. The claim points to a broader shift: coding is ceasing to be the scarce bottleneck it once was.

Unbundle first, rebundle later

History offers a pattern. When a previously scarce capability becomes cheap, supply explodes — printing made books ubiquitous; cloud made startups prolific. Why would software be different? The immediate effect is fragmentation: features, micro-products and vertical tools proliferate. But users’ attention, trust and distribution channels do not expand at the same pace. So after a period of unbundling, new aggregation layers re-emerge. The key question is not whether more software will be made — it will — but what will be worth using?

The organizational reckoning

Companies that merely bolt agents onto old processes are kidding themselves. R&D used to set the cadence for product, design, testing and growth; now code can outrun every other function. It has been reported that many firms still insist “AI must fit existing rules,” preserving legacy workflows that are optimized for scarcity. The real work is reorganizing around new production systems: curation, orchestration, go-to-market and trust become the scarce assets, not lines of code.

So what should humans do?

Short answer: move up the stack. Focus on what machines cannot cheaply replicate at scale — judgment about what deserves attention, durable relationships, distribution channels and novel workflows that rebundle fragmented capabilities into higher-value systems. Amid US‑China tech tensions and export controls that shape access to compute and models, this strategic shift matters geopolitically as well as commercially. Will humans go back to being gatekeepers, or will they design the gates differently? The next decade will answer that.