⏱️ 10 days.
That’s how long it took Anthropic to go from observing unexpected user behaviour → to shipping Claude Cowork. 🔥 Over the Christmas break, Claude Code went properly viral.
2 min read
Originally posted on LinkedIn · 21 reactions · 5 comments · View original →
⏱️ 10 days. Just 10 days.
That’s how long it took Anthropic to go from observing unexpected user behaviour → to shipping Claude Cowork.
- Over the Christmas break, Claude Code went properly viral. If you were anywhere near X or even LinkedIn, your timeline was probably flooded with people having their “aha” moments.
Even a Google Principal Engineer used Claude Code (how ironic) to build a working prototype of a system in about an hour — something her team had spent a year developing and debating.
One non-obvious behaviour: non-engineers started using it as a general-purpose orchestrator for whatever mattered to them: cleaning up years of messy Downloads, organising receipts and photos, hooking into home automation… even to monitor plant growth.
So Anthropic did the most AI-native thing possible: they treated the odd usage as a strong signal, then used an orchestrating agent fleet to compress the entire observe → decide → ship loop into ~10 days.
👀 If that’s not a wake-up call for CEOs and senior engineering leaders, I don’t know what it is.
That’s what 2026 will look like: shorter feedback loops, parallelised execution, lower coordination drag, smaller batches, higher throughput without drowning in rework.
I’ve already done some incredible things with Claude Cowork, and I can tell: the same acceleration curve we felt in 2025 in product development will hit every function in 2026: finance, legal, ops, HR, marketing, customer support, risk, compliance.
Anywhere work can be described as “here’s the context, here’s the outcome I’m looking for”.