ProductOSDocs
Sign in

Tools

Knowledge

Everything ProductOS knows about your workspace, and the per-source toggles that control what its AI features read.

The Knowledge page is everything ProductOS knows about your workspace, and what it feeds to AI features: ProductSense, Copilot, and inline AI field generation. It's the grounding block behind every AI answer, with a toggle per source.

What you'll see

  • Workspace identity: company name, industry / space, product description, target audience, additional context. It belongs to the workspace, so every member sees the same values and every member's AI calls are grounded on them. Owners and admins edit it; everyone else sees it read-only, with when it was last updated
  • Connected sources: one row per source (workspace identity; vision, mission & pillars; competitors; market trends; personas; objectives & key results; outcomes & learnings; connected repositories; positioning; messaging frameworks; SWOT analysis; pricing tiers; battle cards; deal blockers; win/loss records), each with its live row count, when it was last updated, a link to the page it comes from, and an include / exclude toggle
  • Used by: the AI features that read this block before they answer or suggest

Common actions

  • Fill in the identity fields and save (⌘Enter / Ctrl+Enter from a multi-line field); they lead every AI prompt. Company name and industry take up to 200 characters, target audience 1,000, product description and additional context 2,000; a counter appears near the limit, and Save stays off while a field is over it
  • Toggle a source off to keep it out of the AI prompts you run (the toggles are yours, not the workspace's)
  • ⌘. / Ctrl+.: open the AI overlay from anywhere; both its auto-fired insight and its Copilot chat are grounded in this block

How grounding works

Each source is read live from the page where it lives, no duplication. When an AI feature runs, the enabled sources are loaded and formatted into the prompt as grounding context, alongside whatever entity or page you triggered it from. Empty sources contribute nothing, so an empty vision → vague AI output.

What it uses

Your own Anthropic API key (configured under Settings → AI). Per-message cost varies; the usage panel on the same settings page breaks it down by feature. See BYO Anthropic key for cost expectations.

Related

Last reviewed 2026-09-23