Skip to content
Yamidoo

Answering from your content

Yamidoo answers customer questions using your own content — the pages, help articles, and docs you already publish. Instead of writing canned responses, you add sources in the dashboard and Yamidoo grounds every answer in them, linking back to where it came from.

How it works

Section titled “How it works”

  1. Add sources in the dashboard.

  2. Yamidoo indexes them — content is fetched, split, embedded, and stored as a searchable knowledge base. Re-run indexing to pick up changes.

  3. Customers ask questions in the widget, and Yamidoo answers from the indexed content, citing the source page.

Source types

Section titled “Source types”

You can add content from several kinds of sources:

Source What it pulls in
Sitemap Every page listed in an XML sitemap. Recommended for whole sites.
Website crawl Pages found by following links from a start URL. Depth 1–5 (default 2); max pages default 25, within your plan’s page budget.
URL list Specific pages — one or many.
Q&A Question-and-answer pairs you write directly.
Text Plain text or Markdown you paste in.
Documents Uploaded files: PDF, DOCX, PPTX, HTML, TXT, Markdown, CSV, TSV. Up to 20 files, 20 MB each.

Choosing pages from a sitemap

Section titled “Choosing pages from a sitemap”

After Scan, tick the sections to index, such as /docs or /features. A section that splits into groups shows them underneath, so you can take /docs/article and leave /docs/article-category. Groups that are archive listings start unticked.

For finer control, two optional lists take paths with * as a wildcard, one per line:

  • Also include pages matching, e.g. /docs/article/* for doc articles only, or /features/*.
  • Skip pages matching. Category, tag, author and paging archives (*/category/*, *-category/*, */tag/*, *-tag/*, */author/*, */page/*) are listed by default: they are lists of posts and rarely answer a question. Remove a line to index those too.

The count updates as you type, and Show the pages that will be indexed lists them before you save.

Patterns on crawls and URL lists

Section titled “Patterns on crawls and URL lists”

Website crawl has the same two fields. Only index pages matching keeps the crawl reading every page to find links, but indexes just the matching ones, so a crawl from your homepage can collect only /docs/article/*. Skip pages matching is never fetched or followed. Archives are skipped by default.

URL list accepts * in a line: example.com/docs/article/* adds every matching page from the site’s sitemap, and is re-checked on each sync so new pages are picked up. The optional skip field drops some of what those lines match.

When a site has a sitemap, a sitemap source or a * line is faster than a crawl with an include pattern, which has to read the pages in between.

Editing a source

Section titled “Editing a source”

The pencil on a sitemap, website crawl or URL-list source opens its settings: sections, include and skip patterns, page and depth limits, and the URL list. Save and re-sync applies them straight away: pages that no longer match are removed and new matches are added. The sitemap or start URL itself can’t be changed; add a new source for a different URL.

Deleting a source frees its pages from your page budget immediately.

Page budget

Section titled “Page budget”

Indexed pages count against one org-wide pool shared by all your projects: Free 100, Starter 500, Plus 2,000, Pro 5,000 pages. On the Free plan you can add one source of each type per project.

Keeping sources fresh

Section titled “Keeping sources fresh”

You can re-sync any source by hand at any time. Auto-refresh depends on your plan: never on Free, monthly on Starter, weekly on Plus, daily on Pro.

Choosing good sources

Section titled “Choosing good sources”
  • Public docs and help center — the highest-signal source.
  • Pricing, FAQ, and policy pages — common question targets.
  • Q&A pairs — the fastest way to fix a specific wrong or missing answer.

Keeping answers accurate

Section titled “Keeping answers accurate”

When an answer is wrong or outdated, fix the source — update the page and re-index, or add a Q&A pair for a precise correction. To find what’s missing, open the project’s Content gaps page: it collects the questions the assistant couldn’t answer well and groups them into topics you can write for.

Next steps

Section titled “Next steps”