Bridges Overview
Bridges are VPC peering for AI workspaces. They create secure, authenticated communication channels between two Roundtable workspaces, enabling specialized AI agents to collaborate on complex tasks that span teams, domains, or data boundaries.
Why Bridges Are Powerful
In most organizations, knowledge lives in silos. Your engineering team operates in one context, analytics in another, and security in yet another. Bridges break down these silos by letting workspaces delegate tasks to each other without sharing credentials, merging data, or losing specialization.
Each workspace keeps its own:
- AI provider and model — tuned for its domain
- System prompt — optimized for its specific role
- Data connections — scoped to what it needs
- Tool permissions — limited to relevant capabilities
Bridges let these specialized workspaces work together as a team.
Use Cases
| Scenario | Source Workspace | Target Workspace | What Happens |
|---|---|---|---|
| Revenue analysis | Engineering | Analytics | Engineering workspace delegates "summarize last week's revenue" to Analytics, which has BigQuery access and a data-focused system prompt |
| Incident triage | Security | DevOps | Security workspace queries infrastructure logs via the DevOps workspace, which has shell access to monitoring tools |
| Customer escalation | Support | Engineering | Support workspace delegates a bug investigation to Engineering, which can read code and query error databases |
| Report generation | Executive | Analytics | Executive workspace asks Analytics to build a weekly KPI summary from Snowflake data |
Plan Requirement
:::info Business or Enterprise Only Bridges are available on the Business ($100/ws/mo) and Enterprise plans. Team plan workspaces cannot create or participate in bridges. :::
Key Concepts
Bridges support two communication modes:
Message (Fire-and-Forget)
Send a one-way message to another workspace. The source workspace does not wait for a response. Use this for notifications, logging, or triggering async workflows.
Source Workspace → "New deployment completed for v2.4.1" → Target Workspace
Delegate (Invoke AI + Return Result)
Send a task to another workspace's AI agent and wait for the result. The target workspace processes the request using its own AI provider, tools, and data connections, then returns the response to the source.
Source Workspace → "Summarize Q2 revenue by region" → Target AI processes → Result returned
Delegation is the more common and powerful mode — it turns your workspace network into a distributed AI system where each node is a specialist.
Architecture
Bridges use a control-plane relay architecture with HMAC-signed HTTP to ensure security and authenticity.
┌──────────────┐ HMAC-signed HTTP ┌──────────────────┐ HMAC-signed HTTP ┌──────────────┐
│ Source │ ──────────────────────► │ Control Plane │ ──────────────────────► │ Target │
│ Workspace │ │ Relay │ │ Workspace │
│ │ ◄────────────────────── │ │ ◄────────────────────── │ │
└──────────────┘ Signed response └──────────────────┘ Signed response └──────────────┘
How it works:
- The source workspace calls the
bridge_workspacetool with a target workspace ID and a task description. - The request is HMAC-signed and sent to the Roundtable control plane.
- The control plane verifies the signature, checks bridge permissions, and relays the request to the target workspace.
- The target workspace's AI processes the request and returns a signed response.
- The control plane relays the response back to the source workspace.
:::tip Security by Design Every bridge request is HMAC-signed using workspace-specific keys. The control plane verifies signatures on both ends — neither workspace needs to trust the other directly. Requests that fail signature verification are rejected immediately. :::
Next Steps
- Setting Up Bridges — Create your first bridge between two workspaces
- Bridge Delegation — Deep dive into delegated AI tasks