Workspace guide
One AI foundation, three ways to put it to work.
Relay separates individual exploration, collaborative production, and embedded automation so teams can adopt the right operating model without buying into a false hierarchy.
Compare an option
Every option is listed below. The selector condenses the view when scripts are available.
Notebook
Focused drafting for an individual analyst.
- Working context What the assistant can use while completing a task.
- Personal files Manual uploads per workspace
- Review How work is checked before it reaches another system.
- Author review
- Reusable methods How teams preserve instructions and quality standards.
- Saved prompts
- Observability What operators can inspect after a run.
- Conversation history
- Best fit
- Research and first drafts
Studio
Shared AI work with review and reusable standards.
Recommended
- Working context What the assistant can use while completing a task.
- Shared knowledge Sources inherit team access
- Review How work is checked before it reaches another system.
- Named approvers Comments and version history
- Reusable methods How teams preserve instructions and quality standards.
- Team playbooks Variables, examples, and owners
- Observability What operators can inspect after a run.
- Usage and feedback Filterable by team and playbook
- Best fit
- Repeatable team deliverables
Runtime
Governed automations embedded in production systems.
- Working context What the assistant can use while completing a task.
- Live systems Scoped retrieval through APIs
- Review How work is checked before it reaches another system.
- Policy gates Deterministic checks before action
- Reusable methods How teams preserve instructions and quality standards.
- Versioned workflows Tested and deployed through CI
- Observability What operators can inspect after a run.
- Traces and evaluations Inputs, tools, latency, and cost
- Best fit
- Customer-facing automation
Product capabilities compared across Notebook, Studio, Runtime| Capability | Notebook Focused drafting for an individual analyst. | Shared AI work with review and reusable standards. | Runtime Governed automations embedded in production systems. |
|---|
| Working context What the assistant can use while completing a task. | Personal files Manual uploads per workspace | Shared knowledge Sources inherit team access | Live systems Scoped retrieval through APIs |
|---|
| Review How work is checked before it reaches another system. | Author review | Named approvers Comments and version history | Policy gates Deterministic checks before action |
|---|
| Reusable methods How teams preserve instructions and quality standards. | Saved prompts | Team playbooks Variables, examples, and owners | Versioned workflows Tested and deployed through CI |
|---|
| Observability What operators can inspect after a run. | Conversation history | Usage and feedback Filterable by team and playbook | Traces and evaluations Inputs, tools, latency, and cost |
|---|
| Best fit | Research and first drafts | Repeatable team deliverables | Customer-facing automation |
|---|
Illustrative product configuration for the Relay demo. Availability and governance requirements should be validated against your organization’s policies.