Hadar BismutIf you don't know Ramp, you're in good company. I didn't either. But lately I've been hearing a lot about it as the company that reached the milestone of 99% AI adoption, so I went to look into it.
Ramp is an American fintech company valued at $32 billion, with about 2,000 employees and over $1 billion in annual revenue, and everyone there is called a Builder. No FE, no data, no analysts. Everyone is a builder.
When a company reports 99% AI adoption, the first things that come to mind are culture, motivation, maybe high-quality learning programs. But the problem Ramp discovered was simpler: the work environment required terminal windows, npm installs and manual MCP configuration. Non-technical employees simply couldn't get started because the entry point was too technical.
Their insight: what limits AI adoption is the employee's ability to experiment and work with the models on their own, more than the power of the model.
The first thing they built was a work environment called Glass. Employees log in through the company's regular SSO, and from the moment they're logged in, everything is already connected: Gong, Salesforce and all the internal systems. Zero setup, zero IT tickets.
What's interesting is what they chose not to do: they didn't dumb down the environment. They removed friction and kept the capability. Power users still get multi-window workflows and deep integrations. The entry is easy, and the ceiling stays high. Salespeople can ask Glass to pull context from a Gong call, enrich it with Salesforce data and write a follow-up, all without leaving the environment.
The second thing is a recommendation engine called Sensei. To understand why it's needed, you first need to understand what's in Dojo.
Dojo is an internal marketplace with over 350 skills. It sounds impressive, and in practice that can be exactly the problem: a library of 350 items where nobody finds anything. Sensei is what prevents that. It learns your role, the tools you have connected and what you've been working on recently, and it keeps matching you with relevant skills, every day and well past day one. A new account manager doesn't look at 350 skills. They see the five that are relevant to their first week.
But the third thing is what really caught my attention.
Dojo itself, the internal marketplace they built. The idea is simple: when someone on the sales team discovers a great workflow for analyzing customer calls and generating battlecards, she packages it as a skill and publishes it. From that moment, every other employee in the company can use it, and of course publish skills of their own. Every skill that goes live goes through code review and versioning like any other code in the company, and from then on it becomes the professional standard for its domain.
Finance, design, operations, sales: each domain builds its own shared foundation.
What they actually created is a complete layer of professional standards, built from the bottom up, by domain, by the people who do the work.
Glass lowered the barrier to entry.
Sensei turned a huge library into a tool you can work with.
Dojo turned private knowledge into shared infrastructure.
This is a long process, and it isn't simple. And it's the difference between an organization that talks AI and an organization that builds the infrastructure that lets employees really use AI.