Ramp is offering early access to Glass, an AI workspace built around teams sharing context, sessions and reusable workflows. The pitch is a shift from personal chatbot histories to AI work that colleagues can pick up and improve together, with the usual question of who gets to see what close behind.
Ramp Watch analysis
What happened
In a post published on 30 October, Ramp describes “multiplayer AI” as people working with an AI assistant in shared sessions and projects, rather than each person keeping useful prompts and work in a private chat. Its new product, Glass, is presented as a workspace where teams can connect existing tools, use shared company knowledge, bring colleagues into sessions and turn useful processes into reusable workflows. Ramp says teams can choose from multiple model providers. The company is inviting people to request early access.
Ramp argues that shared context and workflows matter more than simply sharing chats. Its examples include a support agent and engineer working in the same session, and colleagues reusing a process instead of rebuilding it from memory. It also distinguishes this approach from multi-agent AI: the emphasis here is people collaborating with AI, not several AI agents coordinating with one another.
Why it matters
If useful work stays in one person’s chat history, colleagues may have to start again when they take over. Shared sessions and workflows could make handovers less wasteful and help teams reuse what has already worked. That is a more concrete workplace promise than putting another chatbot icon in the toolbar.
The trade-off is that team AI needs sensible boundaries. Ramp says personal chats can stay private, while projects and workflows are shared with the team that owns them. It also advises auditing access to connected files and assigning an owner to keep shared workflows current. Those details matter: a company brain is only helpful if it knows the right things and not everything.
Our read
Glass is Ramp’s attempt to make AI collaboration part of the workspace rather than a series of isolated chats. The useful test will be whether people actually reuse each other’s work, not how many shared projects appear in a launch post. If you request early access, start with a workflow that benefits from a handover and make its sharing rules clear before connecting company information.
What to watch
- When Glass moves beyond early access and which teams can use it.
- Whether shared workflows are reused and improved by colleagues other than their creators.
- How the product handles permissions for connected tools and private work.
Discussion spark: Would you want your team’s best AI workflows shared by default, or should each person decide what work leaves their private chat?
Sources and evidence
- Multiplayer AI: How Teams Collaborate on AI Work (30 October 2026, 20:08 UTC)
not affiliated with or endorsed by Ramp