Skip to main content
Available on the Enterprise plan.Only workspace admins can link and manage workspace context docs.

Overview

Workspace context docs let admins link existing Docs as persistent knowledge for Dovetail AI. Once linked, that content is injected into every Chat, Agent, and Ask Dovetail conversation across the workspace — so you don’t have to explain who you are, what you build, or how your team talks about customers every time you ask a question. If a linked doc is updated, the context refreshes automatically. There’s nothing to re-link or re-save.

How workspace context docs work

Where they apply

Workspace context docs apply everywhere Dovetail AI shows up — Chat, Agents, and Ask Dovetail — for every user in the workspace. Applies to Slack and Teams too, if you’ve connected Chat in Slack and Teams. Access is still governed by permissions. See Permissions and sharing below.

How content is processed

  • Short docs are used as-is.
  • Longer docs are automatically summarized so they fit cleanly into the AI’s context window without crowding out other context or degrading response quality.
  • Edits to a linked doc are picked up automatically on the next interaction — you never need to re-link.

How context docs fit with other context

Chat draws on more than just workspace context docs. Here’s how the layers relate: Context docs teach the AI who you are. Your projects and channels teach it what customers said. For the full breakdown of how all context layers combine, see Chat: Technical overview.

What to include (and what to leave out)

Good candidates for a context doc:
  • Company and product context
  • Strategy and priorities — OKRs, focus areas, what’s out of scope
  • Customer and market context — ICP, segments, personas
  • Product domain knowledge — features, workflows, known limitations
  • Pricing and packaging
  • Brand and tone guidelines
  • Research frameworks and report templates
  • Glossaries and preferred terminology
Not a good fit:
  • Raw research data (interviews, transcripts, tickets) → belongs in a Project or Channel, not a context doc
  • Findings from a specific study → use a Project Doc or that project’s overview instead
  • A one-off answer to a single question → just ask Chat at the right scope
  • Highly sensitive content you don’t want broadly accessible → see Permissions and sharing

Set up workspace context docs

  1. Go to Workspace settings → AI → Custom context.
  2. Search for and select existing Docs to link.
  3. Save your selection.
  4. Confirm sharing — linked docs need to be shared with the workspace at Can view access or higher (see Permissions and sharing).
  5. Test it — open Chat at the workspace level and ask something that depends on background knowledge, like “What’s our ICP?” or “How should I refer to our product tiers?”
Start with 3–5 focused docs rather than one long reference file. Focused docs are easier to keep up to date, and summarize more accurately when they’re long.

Best practices for writing context docs

Keep each doc focused. One purpose per doc. Link several focused docs instead of a single catch-all. Use clear structure. Headings, bullet points, and labeled sections help the AI parse a doc correctly — especially once it gets summarized. Be explicit. Spell out preferred terminology (“customers,” not “users”), product and plan names, what’s in scope vs. out of scope, and any compliance or privacy rules. Separate facts from behavior. Company info, product specs, and glossaries are facts — they belong in a context doc. Tone, formatting, and citation preferences are behavior — they’re often better set through your workspace’s custom instructions, with a context doc as backup for anything that needs more detail. Keep docs maintained. Treat context docs as living reference material. Outdated strategy or product info leads directly to outdated AI answers. Don’t duplicate project-level content. Project-specific research context belongs in project overviews and project Docs, not workspace context docs.

Permissions and sharing

Linked docs don’t have to be shared workspace-wide. Chat still respects every existing Dovetail permission — the AI can’t surface content a user can’t already see.
  • If a linked doc has restricted permissions, only users with at least Can view access to that doc will have it applied to their chat context.
  • For the context to benefit the whole workspace, share each linked doc with the entire workspace at Can view access or higher.
For more on access levels, see Access and permissions.
A focused starting set covers most workspaces well. Create each as its own Doc, fill in the details, and link all five in Workspace settings → AI → Custom context.

1. Company & product overview

Give the AI foundational knowledge about who you are and what you build.

2. Terminology & glossary

Keep language consistent across every AI-generated response.

3. Brand voice & AI response guidelines

Shape how the AI communicates — tone, format, and behavioral rules.
If your workspace already sets tone and formatting through custom instructions, you may not need this as a separate doc. Use whichever is easier for your team to maintain.

4. Customer & market context

Help the AI understand who your customers are and how you talk about them.

5. Research & reporting standards

Keep AI-generated Docs and summaries aligned with your team’s methodology.
Quick-start checklist: fill in the placeholders above, share each doc with the workspace at Can view access or higher, link all five in Workspace settings → AI → Custom context, then test with a few questions from your starter set. Assign an owner to review and update the docs quarterly.

Test your setup

Once your docs are linked, try a few questions that only a well-informed teammate could answer well:

FAQs

No. Context docs provide background knowledge about your company and product. Chat still searches your workspace data for every question and cites its sources — context docs just make those answers better informed.
Workspace context docs carry company-wide reference knowledge that applies everywhere in Dovetail AI. A project overview carries context specific to a single project.

Related articles