Workspace context
Workspace context is a block of free-form instructions for the Streya agent that is included in every prompt, in every conversation, for everyone in the workspace. It’s where workspace-wide knowledge and preferences live — the things you’d tell a new analyst once and expect them to apply to every answer thereafter.
You edit it from your workspace settings.
What to put in it
Section titled “What to put in it”Use workspace context for anything that should shape every answer:
- Rules and directives. Hard constraints the agent must follow. “Never report figures for the current, incomplete month.”
- Best practices specific to this workspace. How analysis should be approached here. “When comparing periods, default to year-over-year, not month-over-month.”
- Answer preferences. How you want results presented. “Report revenue in thousands. Prefer charts over tables for trends.”
- Company description. Who you are and what you do, so the agent reasons in your context. “We’re a B2B wholesaler; ‘customer’ means a retail account, not an end consumer.”
- Details unique to your business. Fiscal calendar, key segments, naming quirks, anything an outsider wouldn’t know.
Context vs. descriptions
Section titled “Context vs. descriptions”Workspace context and semantic-model descriptions both teach the agent, at different scopes:
| Workspace context | Descriptions | |
|---|---|---|
| Scope | The whole workspace | One view or one field |
| Use for | Company background, global rules, answer preferences | What a specific dataset or field means, its conventions and gotchas |
Rule of thumb: if it applies to every question, it’s context; if it explains one field or dataset, it’s a description. Keep each piece of knowledge in one place — restating a field’s caveat in the workspace context just creates drift.