Back to Blog
Product

One Setup, Zero Friction: Introducing Lumina Team Workspaces

Lumina Team Mar 11, 2026 6 min read
One Setup, Zero Friction: Introducing Lumina Team Workspaces

Getting a team onto a new tool is hard. Getting a team onto an AI analytics tool, where every member needs the right agent, the right data, and the right credentials, is normally a project in itself. We didn't want Lumina to work that way.

Today we're shipping Team Workspaces: a structured collaboration model that separates who sets things up from who uses them. One admin, one configuration, zero setup for the rest of the team.

The Problem with "Share Everything" Collaboration

Most collaborative tools default to a flat permission model. Everyone can see and change everything. That works fine for documents. It doesn't work well for AI workspaces, where the configuration is the product. If every team member can change the agent, edit the data connection, or reconfigure the project, you end up with a fragile, inconsistent environment that nobody trusts.

We've also seen the opposite failure: tools that are so locked down that members spend their first hour just trying to connect to a data source, hunting for credentials, or figuring out which agent they're supposed to be using.

"The best onboarding experience is no onboarding experience. A member should join a team and immediately see exactly what they need."

Three Roles, One Clear Flow

Team Workspaces are built around three distinct roles, each with a specific scope:

  • L

    Lumina Admin

    Creates teams at the platform level. This is the Lumina account admin, the person who manages the subscription and initial setup.

  • A

    Team Admin

    Owns the team's configuration. Invites members, connects cloud data sources, adds agents, builds project bundles, and controls who sees what. Has full access to the Team Dashboard.

  • M

    Team Member

    Sees their assigned project cards. Clicks a data source row to browse a cloud bucket and load a file. Optionally shares their own custom agents with the team from Agent Studio.

What the Admin Sets Up (Once)

The Team Dashboard is the admin's command centre. Everything happens here, and members never need to touch it.

1. Connect data sources. The admin adds Google Cloud Storage connections: bucket name plus a service account JSON key. Credentials are encrypted server-side. Members browse and load files without ever seeing a credential.

2. Add agents. The admin opens the Agents tab and clicks Add Agent. A modal shows all agents from Agent Studio: name, role, description, and whether they're published or still private. Click Add and the agent is available for project assignment.

3. Bundle into projects. A Team Project ties agents and data sources together under a single name. The admin picks a name, selects agents, selects data sources, and decides whether to assign it to all members or a specific subset. That's it.

Changes take effect immediately. Edit a project and every assigned member sees the updated card on their next page load.

What the Member Sees

When a team member opens Connect Data, they see a Your Team Projects section. Each card shows the project name, how many agents and data sources it contains, and a clickable row for each data source. No menus. No configuration. No Team Dashboard to navigate.

Clicking a data source opens a file browser directly into that GCS bucket. Select a file, and it loads into the analysis workspace, with the project's assigned agent ready to use.

We deliberately removed the redundant buttons that used to appear on these cards. The "Use Project" button just opened the team page, already accessible via other routes. The "View Team" link went to an admin-only dashboard that members had no business being in. The data source rows are the action. Everything else was noise.

Members Can Contribute Too

The flow doesn't only go top-down. Team members who build agents in Agent Studio can share them upward. On any agent card in the studio, the menu (⋯) now offers Share with Team. A project picker lets the member choose which project to attach the agent to. The backend verifies they actually authored the agent before allowing the share. Members can contribute their own work, not co-opt someone else's.

Admins see contributed agents appear in the Agents tab with a note showing who shared them and when. They can remove any agent from the team pool at any time.

Access Discipline

A few things we're opinionated about:

  • The Team Dashboard is strictly admin-only. If a team member somehow navigates there, they're redirected to the Connect Data page.
  • The "My Team" link in the user menu only appears for team admins. Members have no reason to see it.
  • The Team Dashboard has no "Use" button. It's a configuration interface. Execution happens in the main workspace after loading data.
  • Members cannot modify project configurations. That's the admin's job.

These aren't arbitrary restrictions. They're the boundaries that make the experience coherent. When everyone knows their lane, no one spends time figuring out what they're supposed to be doing.

Who This Is For

Team Workspaces are built for organizations that have a clear distinction between the people who set up analytical environments and the people who use them. Think: a data science team lead deploying a custom agent to field engineers, a consulting firm giving clients access to a pre-configured analysis workspace, or an operations manager sharing production data with a maintenance crew.

The common thread: configuration complexity is real, and it shouldn't be everyone's problem. The admin eats the complexity once. The members get clarity.

Getting Started

Team Workspaces are available on the Team and Enterprise plans. To start:

  1. Open the user menu and select My Team.
  2. Go to Members → Invite to bring in your first member.
  3. Add a GCS connection in the Data Sources tab.
  4. Add agents in the Agents tab.
  5. Create a project in the Projects tab and assign it.

The in-app Guide button on the Team Dashboard walks through each step with a five-screen admin handbook. Members get their own three-screen handbook the first time they're guided there.

Questions? Reach us at support@lumina.express.

See it on your own data

Upload a dataset and ask in plain English. No setup, no pipeline, runs in your browser.

Get new insights in your inbox

No spam. Just new articles on industrial AI, data sovereignty, and the reasoning layer, sent when we publish.