85% of the workforce does not have a valuable AI use case.

That’s what we see in almost every enterprise we work with. Companies deploy ChatGPT or Copilot, hold a few training sessions, and leave the workforce to figure it out for themselves. But almost no one does.

Klaviyo, a 2,400-person marketing automation company, took a different approach from the start. 

When they kicked off their AI transformation in 2025, they didn't just roll out tools and hope for adoption - they built a deliberate strategy to change how their entire workforce operates.

By the end of 2025, 1,800 employees (three-quarters of the company) had demoed a value-driving AI use case. Here’s how they made it happen.

1. Make experimentation safe, approved, and fun

In August 2025, Klaviyo's CEO sent a company-wide memo with a clear message: AI is here. We need to jump in or get left behind. (At Section, we call this the “Why of AI” - and every company needs one). 

But instead of just saying “AI is here - use it,” Klaviyo gave people permission to play.

They hosted an Enablement Month in September, bringing in OpenAI, Google, Atlassian, Zoom, and leaders from local tech companies to demo AI automations. They gave every employee a learning stipend to spend on AI - not just work tools, but anything they wanted to experiment with.

“You want people to play and experiment,” says Blake Schuller, Klaviyo's AI Transformation Lead. “If you want people to engage, make it cool, make it fun, make it exciting.”

Klaviyo understood something most companies miss: people need to build confidence with AI in low-stakes environments before they'll trust it with work that matters. 

2. Give people three months to build something real

In September, after giving employees a month to experiment, Klaviyo added a new requirement.

Every employee needed to identify an AI goal that would drive value in their specific role. 

They asked employees to commit to building out a use case over the next three months, and be ready to demo by December. This wasn’t a suggestion - it was an expectation tied to end-of-year performance reviews. 

And 1,800 employees completed it in that time frame. Some built automations that saved them hours every week; some failed and pivoted to a different approach. Some built things that were genuinely impressive - like the recent college grad in finance who taught herself HTML and CSS to create a custom site that pulls in stock data with smart tables.

There was no formal grading system - Klaviyo just asked: Did you build something that made your work better?

“You know your role best,” Blake explains. “I can't come in and say that's not valuable when it saves you 14 hours a week.”

3. Managers (not HR) decide if AI goals are meaningful

Instead of HR dictating what everyone’s AI goals should be, Klaviyo had individuals identify their own goals, with manager support. 

The manager's job wasn't to tell people what to build. It was to make sure the goal was actually meaningful before the employee spent three months on it. 

“The people leader was the driver who helped the team succeed," Blake says. "They didn’t say, ‘You need to build XYZ’. They helped the individual identify what would drive value in the context of the business.” 

Klaviyo also added an AI Impact tag to performance reviews, which ranged from “no impact” to “shifted how the business operates.” It wasn't part of the formal performance rating, but it sent a signal that the company cared about how much value the employee was creating with AI.  

The 2026 vision: Embed AI into every part of daily work

Blake's vision for this year is radical: Stop setting AI goals altogether.

“I don't picture us doing an AI goal again,” she says. “AI should be embedded in any goal you have. The more we can embed it into our everyday workflow, the more successful we'll be.”

To embed AI in daily work, Klaviyo is adding AI use cases into regular goal-setting conversations. They're building “Powered by Practice” segments into company-wide meetings where real employees share how their day-to-day workflow has changed with AI.

“I'd love to see Klaviyo get to the point where AI is just literally how we work,” Blake says.