Confluence is usually where the answer lives, and almost never where you can find it. A space accumulates hundreds of pages over a few years: runbooks, decision records, onboarding notes, the policy someone wrote in 2023 and never deprecated. Confluence's own search ranks Deploy runbook (NEW) and Deploy v1 (DEPRECATED) about equally, and neither result tells you which one your team actually follows.
Opening pages one at a time and asking about each is fine when you know where to look. When you don't, what you want is to search the entire space at once: one question, one answer, drawn from every page, with a citation you can click.
Why whole-space search is normally hard
Two things get in the way.
Permissions. Most AI tools that promise to search your wiki do it from a server, through the Confluence API. That needs an administrator to approve broad scopes, and even then the tool sees what the integration account can see - which is rarely what you can see. Restricted spaces, draft pages and personal spaces stay invisible.
Trust. A tool that reads a few hundred pages and produces a confident paragraph is worse than useless if you can't tell which page it came from. Wikis are full of contradictions: a current runbook and its deprecated twin, two policies with different retention numbers. An answer without a citation is a rumor with better grammar.
The approach: index the space from your own session
Sidenote is a browser extension, so it sidesteps both problems. It reads Confluence through the session you are already signed into, on Cloud, Server or Data Center. Nothing is crawled on a server, and no administrator has to approve anything.
That is also what makes whole-space indexing safe. When Sidenote walks a space, it walks it as you: every page it reads is a page your account could have opened by hand. Pages you cannot see are skipped, not fetched with elevated permissions.
Step 1 - Open any page in the space
Sign in to Confluence as usual and open a single page in the space you want to search. That is enough for the side panel to identify the space.
Step 2 - Choose "Index this space"
Sidenote walks the space and adds every page you can open to one collection. The crawl paces itself so it never hammers your wiki, you can stop it mid-run, and it resumes from where it stopped rather than starting again.
Step 3 - Ask once, across everything
Ask your question in plain language. The answer draws on the whole indexed space rather than the tab in front of you:
What is our current hotfix deploy process, and who has to approve it?
Step 4 - Click the citation
Every claim quotes the passage it came from and names the page. Click a citation and Sidenote opens that page at the exact paragraph, so you can confirm you are reading the live runbook and not the deprecated copy someone forked into a personal space.
What it costs, and what it does not
Whole-space indexing is available on Pro+ and Pro Max. The 7-day Pro trial includes 20 space pages so you can try it before deciding.
The important detail is that space pages draw on their own monthly allowance, separate from the documents you add the normal way: 200 pages a month on Pro+, 500 on Pro Max. Indexing a 180-page space does not consume the document cap you were saving for the PDFs on your desk. Re-crawling a page you already indexed is free.
The free and Pro plans do not include whole-space indexing. Both can still read the Confluence pages you open individually, one at a time, and ask across them. See pricing for the full breakdown.
Beyond one space
Two neighboring problems have their own guides:
- If your question spans several Confluence spaces rather than one, open the relevant pages across each and bundle them into a Collection: how to search across multiple Confluence spaces.
- If it spans your whole company wiki and the tools around it - SharePoint files, Google Docs, Notion pages - the same Collection can hold all of them: how to find information in your company wiki.
Throughout, your wiki content stays in your account. It is never used to train AI models, and it is stored in a UK region isolated per account. See security and compliance for how that works on sensitive internal documents.
The short version
- Confluence's own search cannot tell a current page from a stale one, and server-side AI tools cannot see what you see.
- On Pro+ and above, index an entire space in one pass, through your own signed-in session. It only ever reaches pages you could open yourself.
- Space pages have their own monthly allowance (200 on Pro+, 500 on Pro Max), so a crawl never eats your document cap.
- Ask once across the whole space, then click each citation back to the exact page and paragraph before you trust it.