Ramp’s AI Coworker Turns Employee Workflows Into Reusable Skills

Ramp, the fintech giant, has developed and deployed an internal AI productivity suite, Glass, for all its employees. 

The company built the proprietary software to bypass the complex configuration processes previously required to use AI models. 

“99% of Ramp uses AI daily. But we noticed most people were stuck. Not because the models weren't good enough, but because the setup was too painful and unintuitive,” Eric Glyman, Co-founder and CEO of Ramp, posted on X. 

Glass connects to Ramp’s existing enterprise systems via a single sign-on process, securely linking with workplace applications such as Slack and Notion. The platform functions as a persistent digital workspace featuring split-pane windows, allowing users to view data files, inline code, and text generation simultaneously. 

Glyman stated that there are 350+ reusable skills built by employees that can be used in Glass. “When one person on a team figures out a better workflow, everyone on that team gets it and gets more productive,” he said. 

Glass updates its contextual memory daily by scanning authenticated applications, enabling the software to automatically track ongoing projects, team structures, and internal documents.

The company stated in a blog post that Glass was designed to prevent individual discoveries from remaining siloed. “The biggest failure mode wasn’t that people couldn’t figure things out. It was that everyone had to figure things out alone,” stated Seb Goddijn, an engineer from the company. 

To manage the volume of available workflows, the platform uses an algorithmic guide called Sensei. This feature analyses a user's role and active projects to recommend the most relevant automation skills upon login. 

Employees can also schedule these tasks to run in the background using custom, chronological triggers, with the system delivering the final reports directly into team communication channels.

Goddijin added that instead of buying a SaaS tool with similar capabilities, Ramp chose to build it in-house because internal AI capability is considered a ‘moat’, owning the system enables faster iteration and fixes, and the learnings directly inform its customer-facing products. 

“We have a Slack channel where users report issues, and our team triages them into tickets automatically, with most resolved in hours. You cannot do that while waiting on a vendor’s roadmap,” he said.