Ramp’s Glass gives employees and AI models a shared company knowledge layerMachine translation

Every company I’m aware of is either building or seeking to buy something that places their data, people, and agents on the same page.

Ask a model how many U.K. customers your company has added, and it will probably count U.S. companies that have U.K. offices.

Ramp’s Glass got that question right because Finance had corrected the definition months earlier. The correction sits in a shared knowledge layer that every employee’s session reads from.

It puts a shared, editable layer of company knowledge in front of whichever model you run: definitions, saved Skills, and connections to Linear, Notion, Google, Slack and Gong.

The model is swappable by design. Glass runs open or closed models, and Ramp’s position is that you switch the day a better one ships.