What Ramp's Glass Teaches Us About Enterprise AI Adoption

When Ramp revealed Glass, their internal AI platform, the numbers were staggering: 350+ AI skills, 99.5% employee adoption, and measurable productivity gains across every department. For anyone paying attention to enterprise AI, the takeaway was clear.

The question isn't whether AI can transform how teams work. It's whether your organization can build the infrastructure to make it happen.

The Glass Playbook

Ramp didn't roll out ChatGPT access and call it a day. They built a structured system with three critical components:

1. A skill marketplace (Dojo) where proven workflows get packaged and shared across the company
2. A proficiency framework (L0-L3) that gives employees a clear path from awareness to mastery
3. Real workflow integration that embeds AI into the tools people already use

The result? Not a handful of power users carrying the team, but near-universal adoption. That 99.5% number isn't a vanity metric. It represents a fundamental shift in how an entire organization works.

Why Most Companies Can't Replicate This

Here's the uncomfortable truth: Ramp is an AI-native fintech company with world-class engineering talent. They had the resources, technical depth, and cultural DNA to build Glass from scratch.

Most enterprises don't.

According to McKinsey's 2025 AI survey, 72% of organizations have adopted AI in at least one function, but only 21% report meaningful productivity improvements. The gap between "we have AI tools" and "AI has changed how we work" is enormous.

The typical enterprise approach looks like this:

• Buy licenses. Give everyone access to an AI assistant.
• Run training. A few workshops, maybe a certification program.
• Hope for adoption. Cross fingers that people figure it out.
• Measure nothing. No visibility into whether behavior actually changed.

This is why most AI rollouts plateau at 15-25% active usage. The tools are available, but the workflows, scaffolding, and measurement systems don't exist.

The Three Ingredients That Actually Matter

What Ramp got right, and what any organization needs, comes down to three things:

1. Capture What Your Best People Do

Ramp's Dojo marketplace works because it codifies how top performers use AI. The senior engineer who built a debugging workflow, the finance analyst who automated reconciliation, the marketer who nailed audience segmentation: their knowledge becomes reusable.

Most organizations let this institutional knowledge live in individual heads. When those people leave, the knowledge walks out the door with them.

2. Give People a Path, Not Just a Tool

The L0-L3 proficiency framework matters because it turns AI adoption from a binary (using it or not) into a journey. People can see where they are, understand what's next, and build skills progressively.

Without a clear progression model, you get a bimodal distribution: a small group of enthusiasts at one end and a large group of non-adopters at the other. A framework gives the middle majority a reason to engage.

3. Measure Behavior, Not Completion

Ramp didn't track how many people completed an AI training course. They tracked how many people actually used AI skills in their daily work. That's the difference between measuring compliance and measuring impact.

Productizing the Glass Approach

The lesson from Ramp isn't "build your own Glass." It's that the Glass approach, capturing senior workflows, providing structured learning paths, and measuring real behavior change, is the right model for enterprise AI adoption.

That's exactly what Nova does. Instead of requiring a dedicated engineering team to build internal tooling, Nova provides the platform:

• Workflow authoring lets your senior employees capture how they use AI, turning tribal knowledge into structured, repeatable sessions
• Guided learning walks every team member through those workflows while producing real deliverables, not hypothetical exercises
• Proficiency tracking maps each person's progress across an L0-L3 framework, giving managers visibility and employees a clear growth path
• Behavior metrics measure whether people actually change how they work, not just whether they sat through a course

The companies that win the AI adoption race won't be the ones with the most tools. They'll be the ones that build systems to capture, transfer, and measure how their people use those tools.

Ramp proved the model works. Nova makes it accessible to everyone else.

Ready to Build Your AI Workforce?

Stop guessing whether your AI investments are paying off. Nova gives you the same structured approach that drove 99.5% adoption at Ramp, without building it from scratch.

Talk to our team to see how Nova can transform AI adoption at your organization.