See which paragraphs read like a model wrote them.

Every paragraph gets marked human-ish, machine-ish or mixed, with the odds. When the evidence is thin it says it can't tell, and it says that a lot, on purpose. The model is 56 KB and runs in your browser, so nothing you paste gets uploaded.

0255075100HumanMachine% machine-shaped

Reading

anthropic/claude-haiku-4.5, generated 2026-09-18, prompt: "Write a LinkedIn post about why remote work makes teams more productive."

# Remote Work: The Productivity Game-Changer Teams Didn't See Coming

We've been told for decades that productivity happens in the office. Turns out, the data tells a different story.

After years of hybrid and remote transitions, here's what we're actually seeing:

**Fewer interruptions = deeper focus.** No surprise: people get more done when they're not interrupted 47 times a day by desk drop-bys. Remote work creates space for actual deep work.

**Asynchronous communication is underrated.** When not everyone is online simultaneously, teams document decisions better, respond more thoughtfully, and create searchable records that eliminate the "did we decide this?" confusion.

**Commute time becomes productive time.** Instead of losing 2 hours a day in traffic, people channel that energy into work, family, or just... not being exhausted.

**Flexibility builds retention.** Productive teams are hard to build. Keeping them together saves the company months of hiring and training costs.

**Fewer meetings about meetings.** Remote-first cultures tend to ask "is this meeting necessary?" more often. Spoiler: usually isn't.

That said—not every role works remote, and healthy teams still need connection. But the productivity boost when done right? That's real.

What's been your experience? Have you seen productivity shift with remote work?

#RemoteWork #Productivity #FutureOfWork #WorkCulture

Engine

Test WebGPU on this device

The sheet above scores each paragraph on the CPU, in well under a millisecond. WebGPU takes over for big batches, like a long page in the extension. This runs both here and checks they agree.

CPU
256 paragraphs, one at a time
WebGPU
the same paragraphs in one batch