Zapier stopped work for a week and hit 97% AI adoption
The operational pause most leaders are too busy to try is the only move that actually works
Hey Adopter,
In early 2023, a company whose entire product is built on automation had 10% internal AI adoption. Its own staff. The people who sell workflow automation for a living. Using it for roughly one in ten tasks.
That number should make you uncomfortable if you’re sitting on a team where AI adoption looks good on a slide but flat in practice.
By early 2026, that same company hit 97% adoption across its entire global workforce. Not opt-in usage. Not “logged in at least once.” Active, daily integration across every department, technical and non-technical alike. The outcome was a workforce of 800 people supported by over 800 specialised AI agents.
Where does AI adoption actually stall in your organisation?
Here’s what makes this worth your time. The company didn’t get there by rolling out better tools, running more workshops, or issuing a memo from the top. What they did was structurally different from what almost every enterprise AI rollout looks like. And the gap between their result and the industry average is not small. Nearly 30% of generative AI projects across the market are abandoned after proof of concept. Zapier went the other direction entirely.
The mechanisms behind this are specific, replicable, and uncomfortable for anyone who thinks a software licence and a lunch-and-learn constitute an adoption strategy.
The full breakdown, including the exact four-step intervention that drove the shift, how Klarna ran a parallel AI rollout with headline numbers and ended up with a $152 million net loss, and what the right success metrics look like when your CFO starts asking hard questions, is below.
This piece is for subscribers. And if you want to go deeper, the full AI Adopters case study on Zapier’s enterprise transformation is linked at the end.
