Teams & Enterprise
Teams Overview
When a team runs River, the whole system levels up. Shared Dataspaces become shared world models. Shared Agents become team infrastructure. Shared Workflows become team policy. The AI that knows your work knows your team's work, with the right boundaries.
This page covers how it all hangs together.
What changes when you go from solo to team#
| Solo | Team | |
|---|---|---|
| Dataspace | One personal Dataspace | One team Dataspace + your personal one |
| Agents | Yours | Team-shared infrastructure |
| Mail / Drive | Your accounts | Shared accounts also visible |
| Workflows | Yours | Team-defined processes |
| Search | Across your data | Across team data (with ACLs) |
| AI context | Your work | Your work + the team's |
Team Dataspace#
The team Dataspace is the shared workspace. Everyone with Member access sees the same entity tree, edits the same docs, runs the same Agents. The AI in the team Dataspace can reason across the entire team's work — that's what enables the patterns no single-tool stack can match.
When you're working "as the team," you're in the team Dataspace. When you're working "as you," you're in your personal Dataspace. The switcher is in the top-left chrome.
Team Agents and Workflows#
In a team Dataspace, Agents and Workflows are infrastructure. Some real ones we see teams build:
- Inbox Triage for a shared customer-success inbox — every team member sees the triage results
- PRD Reviewer that runs on every new entity in
/Engineering/PRDsand posts a review draft - Daily Engineering Digest at 5pm that synthesizes commits, PRs, and Linear issues
- Customer Health Monitor that watches Stripe + Slack + Mail for early warning signals
- Onboarding Briefing that runs when a new hire joins; assembles their first-week context
Anyone can see what the Agent did, when, and to which entities. The team's AI work becomes auditable.
Team rituals that work#
A few patterns we've seen change how teams operate:
- Standup automation — recordings auto-transcribe, an Agent extracts action items, a Doc updates, Slack gets a summary. The standup becomes a 10-minute discussion of the synthesis instead of round-robin updates.
- Customer brief auto-prep — every external meeting gets a one-pager generated the night before, with attendee history, recent emails, shared documents.
- Friday recap — an Agent synthesizes the week's activity into a recap doc; nobody has to write it from scratch.
- PR pre-review — an Agent reads every new PR and drafts review comments before a human looks.
Sensitive Dataspaces inside an Org#
When a subset of the team needs access to sensitive material (HR, M&A, legal), create a separate Dataspace inside the same Org with restricted membership. The data stays inside the Org but can't be seen by Members not on the Sensitive Dataspace.
Self-hosted team#
For teams with strict data requirements, run Self-Hosted River with a team Dataspace on your own infrastructure. Same features, your hardware.
Tips#
- Start with one team Dataspace. Add more only when you have a clear data partition need.
- Designate one person as the maintainer of critical Agents. Multiple owners + drift = quality decay.
- Encourage the team to share useful Agents broadly. The compounding value is real.