How do I invite my team in Future AGI and set up RBAC (Growth/Enterprise)?

Most teams hit the same wall once their first agentic workflow works: “How do we safely give more people access without breaking anything?” On Future AGI Growth and Enterprise plans, that’s where team invites and RBAC (role-based access control) come in. You can onboard your org, assign the right roles, and keep evaluations, datasets, and production traces locked down to the right people.

Quick Answer: On Growth and Enterprise, you invite teammates from the Workspace or Organization settings, then assign them roles (Viewer, Editor, Admin, etc.) that control what they can see and change across Datasets, Experiments, Evaluations, Improve workflows, and Monitor & Protect. RBAC lets you scale collaboration without compromising evaluation integrity or safety.

---

The Quick Overview

• What It Is: A team and permissions system for Future AGI that lets you invite colleagues, group them by workspace/org, and control access using RBAC.
• Who It Is For: AI teams on Growth or Enterprise plans who need multiple collaborators across DS/ML/Eng/Safety, with clear boundaries between dev, staging, and production.
• Core Problem Solved: Prevents “everyone is an admin” chaos—protecting prompts, datasets, eval configs, and production guardrails while still enabling fast collaboration.

---

How It Works

Future AGI structures work into stages—Datasets → Experiment → Evaluate → Improve → Monitor & Protect. RBAC sits across all of these, defining who can:

• Create or edit synthetic datasets and scenarios
• Launch Experiments and change model/tool configs
• Edit evaluation metrics or scoring logic
• Approve prompt/workflow changes pushed from Improve
• View or modify production traces, alerts, and safety rules in Monitor & Protect

At a high level:

1. Set Up Your Organization & Workspace(s): Confirm you’re on Growth/Enterprise, define one or more workspaces (e.g., “RAG Chatbot Prod,” “Voice Agent Dev”), and ensure the right admin(s) are in place.
2. Invite Teammates: Add users via email or SSO, assign them to the right workspace(s), and pick default roles.
3. Configure RBAC: Map roles to responsibilities—e.g., who can change evaluation configs vs. who can just view results—and periodically review access as your team and surface area grow.

The exact UI labels may evolve, but the principles below will help you implement a clean, auditable access pattern.

---

1. Confirm Your Plan and Access Level

RBAC and org-level team management are available on Growth and Enterprise plans.

1. Check your plan

   • Go to Settings → Billing / Plan.
   • Confirm you’re on Growth or Enterprise.
   • If you’re on a lower tier and need RBAC, use the in-product upgrade flow or contact us.

2. Verify you have admin privileges

   • Only Org Admins (or equivalent “Owner” role) can:
     • Invite/remove members at the org level.
     • Manage roles and RBAC policies.
     • Configure SSO/SCIM (Enterprise).
   • If you don’t see team management options, ask your current admin to grant you admin access.

---

2. Invite Your Team to Future AGI

Once you’re an admin on Growth or Enterprise, you can start inviting teammates.

2.1 Open Team Management

• Navigate to Settings → Organization / Team (label may appear as “Team,” “Members,” or “Organization” depending on your plan).
• You’ll see:
  • Current members
  • Roles
  • Invite status (pending/active)
  • Workspaces each user can access

2.2 Send Invites

1. Click Invite Member or Invite Teammates.
2. Enter one or more email addresses (comma- or line-separated).
3. Choose:
   • Workspace(s): Which workspace(s) they should join by default.
   • Role: Assign a role like Viewer, Editor, or Admin (details in RBAC section).
4. Click Send Invites.

Each invitee will receive an email with a link to:

• Create or sign in to their Future AGI account.
• Accept the workspace/org invitation.
• Land directly in the assigned workspace.

2.3 Manage Pending Invites

• In the Team / Members view, you can:
  • Resend an invite if it expired or wasn’t found.
  • Revoke an invite if it was sent to the wrong user.
• Once accepted, the user appears as Active, and you can modify their roles and workspace access.

---

3. Understand RBAC Roles on Growth/Enterprise

RBAC in Future AGI is designed around the lifecycle: Datasets, Experiment, Evaluate, Improve, and Monitor & Protect. While exact names may differ in your instance, Growth/Enterprise typically follow this shape:

Note: If your org uses custom roles, your Future AGI admin may have renamed or extended these defaults.

3.1 Common Role Types

• Org Admin / Owner

  • Full org-wide control.
  • Can invite/remove users, manage billing, set up SSO/SCIM, and configure global policies.
  • Can manage all workspaces and RBAC settings.

• Workspace Admin

  • Full control within specific workspace(s).
  • Can:
    • Create/edit Datasets, Experiments, Evaluations, Improve workflows.
    • Configure Monitor & Protect (alerts, guardrails) in that workspace.
    • Invite/remove workspace members (within org rules) and assign workspace roles.
  • Cannot change org billing/plan.

• Editor / Contributor

  • Day-to-day builder role.
  • Can:
    • Create and edit Datasets (incl. synthetic & edge cases).
    • Configure and run Experiments (model, tools, hyperparameters).
    • Adjust Evaluate settings and metrics (where allowed).
    • Use Improve to refine prompts/workflows.
    • View production traces and evaluation results; often limited ability to change Monitor & Protect policies.

• Viewer / Read-Only

  • Analysis and stakeholder role.
  • Can:
    • View Datasets, Experiments, Evaluate results, Improve suggestions, and Monitor & Protect dashboards.
    • Inspect traces and error localization.
  • Cannot make changes, run new experiments, or modify prompts, guardrails, or configs.

• Safety / Compliance Admin (Enterprise, where configured)

  • Focused on Monitor & Protect.
  • Can:
    • Create and manage safety policies (toxicity, sexism, privacy, prompt injection, etc.).
    • Configure blocking rules on inputs/outputs.
    • Review violations and audit logs.
  • May have limited access to edit prompts/workflows directly, depending on your internal policies.

---

4. Map Roles to Your AI Lifecycle

The key to effective RBAC is mapping responsibilities to Future AGI’s stages.

4.1 Datasets

• Who should have edit access:
  • DS/ML/Applied Scientists.
  • Evaluations engineers.
• Why: This is where you define synthetic datasets, edge cases, and regression suites. You want high iteration speed but also control to avoid accidental drift in benchmark data.

Suggested:

• Editors: DS/ML/Applied Scientists.
• Viewers: Product, stakeholders, safety.

4.2 Experiment

• Who should have edit access:
  • ML Engineers, Agent Engineers, Applied Scientists.
• Why: Changing model configs, tools, or agent routing logic directly affects performance and cost.

Suggested:

• Workspace Admin / Editors: Core builders.
• Viewers: Analysts, product managers, leadership.

4.3 Evaluate

• Who should have edit access:
  • Evaluation owners, research leads.
• Why: Metrics define what “good” means. You want tight ownership to ensure comparability over time.

Suggested:

• Editors: Eval owners; others view only.
• View-only: Most builders and stakeholders.

4.4 Improve

• Who should have edit access:
  • Prompt authors, workflow designers, agent engineers.
• Why: Improve is where you close the loop and “automatically refine your prompt” using evaluation feedback. Changes here can “upgrade” live configurations.

Suggested:

• Editors: A small group of owners responsible for production behavior.
• Viewers: Everyone else.

4.5 Monitor & Protect

• Who should have edit access:
  • Production owners, SRE/ML Ops, Safety/Compliance.
• Why: Here you trace production workflows, set alerts, and apply safety guardrails with minimal latency. Misconfigurations can either block too much or let harmful content through.

Suggested:

• Admins: Production owners and safety leads.
• Viewers: Builders and analysts.

---

5. Configure RBAC in Practice

5.1 Set Workspace-Level Roles

1. Go to Settings → Workspace → Members.
2. For each user:
   • Assign a Workspace Role (Admin, Editor, Viewer).
   • Optionally restrict them to specific projects or applications if your workspace is large.

Use workspace boundaries to separate:

• Dev / Staging / Prod for the same agentic workflow.
• Different products (e.g., “Support Chatbot,” “Sales Email Generator,” “Voice Agent”).

5.2 Configure Org-Level Policies (Enterprise)

Enterprise plans typically support:

• Org-wide role presets (e.g., “Prod Safety Admin,” “Eval Owner”).
• SSO groups → roles mapping.
• Possibly SCIM for automated provisioning/deprovisioning.

Work with your IT/security team to:

1. Integrate SSO (Okta, Azure AD, etc.).
2. Map identity groups (e.g., “LLM-engineering”) to Future AGI roles.
3. Define least-privilege defaults:
   • New users default to Viewer in most workspaces.
   • Editors/Admins must be explicitly granted.

5.3 Periodic Access Reviews

At least quarterly (monthly for high-risk workloads), have an org or workspace admin:

• Export or review the member list.
• Remove inactive users or those who have left the company.
• Downgrade roles for users who no longer need edit/production access.
• Confirm safety policies in Monitor & Protect are still aligned with current risk posture.

---

6. Features & Benefits Breakdown

Core Feature	What It Does	Primary Benefit
Workspace-Based Access	Segments Datasets, Experiments, Evaluate, Improve, Monitor & Protect by workspace	Clean separation of dev/staging/prod and different products
Role-Based Permissions	Assigns capabilities (view, edit, admin) per user or group	Least-privilege access; prevents accidental changes
Org-Level Admin Control	Centralizes user management, billing, and security policies	Easier to scale teams and stay compliant

---

7. Ideal Use Cases

• Best for multi-team evaluation workflows: Because you can give DS/ML teams full control over Datasets/Experiments while keeping Evaluate metrics and production guardrails owned by a smaller core group.
• Best for regulated or safety-critical deployments: Because RBAC lets you restrict who can change safety policies, prompts, and Monitor & Protect rules, while still letting stakeholders inspect traces and evaluation outcomes.

---

8. Limitations & Considerations

• Exact role names may differ: Your instance (especially on Enterprise) might use custom role labels. Always check your in-product role descriptions or internal docs.
• RBAC doesn’t replace internal process: You still need code review, change management, and approval workflows around high-impact changes (e.g., production guardrails, critical metrics). Use Future AGI’s traces and evaluation history as part of that process.

---

9. Pricing & Plans

Team invites and RBAC are fully supported on:

• Growth: Best for scaling AI teams that need multiple workspaces, structured roles, and collaboration across builders, PMs, and stakeholders.
• Enterprise: Best for larger organizations needing SSO/SCIM, custom roles, org-level policies, advanced security controls, and tighter integration into existing compliance workflows.

If you’re not sure which plan you’re on, or whether your current tier supports the RBAC setup you need, reach out and we’ll walk you through the options.

---

Frequently Asked Questions

Can I restrict someone to view-only access for Experiments and Evaluations?

Short Answer: Yes. Assign them a Viewer or read-only role in the relevant workspace.

Details: Viewers can inspect Datasets, Experiments, Evaluate results, Improve suggestions, and Monitor & Protect dashboards but cannot modify configs, run new experiments, or change prompts/guardrails. This is ideal for product, leadership, and analytics stakeholders who need visibility without the risk of accidental changes.

---

Can I separate dev, staging, and production access in Future AGI?

Short Answer: Yes—use separate workspaces and roles for each environment.

Details: A common pattern is:

• Dev workspace: Most builders are Editors; fast iteration on datasets, experiments, and prompts.
• Staging workspace: Limited Editors; used for pre-production evaluation against curated datasets.
• Production workspace: Very few Admins/Editors; most users are Viewers. Monitor & Protect and safety guardrails are tightly controlled. RBAC ensures that only designated owners can modify production evaluation configs, prompts, and safety rules, while everyone else can still see traces and performance metrics.

---

Summary

Inviting your team into Future AGI on Growth or Enterprise is straightforward: set up your org and workspaces, invite users via email/SSO, and assign roles that mirror how you build and ship agentic workflows. RBAC gives you the control to keep Datasets, Experiments, Evaluate configs, Improve workflows, and Monitor & Protect guardrails in the right hands—so you can scale collaboration without sacrificing reliability or safety.

Next Step

Get Started