leadiosa.
Start free
Docs · Product

Knowledge base

Your help articles become the AI’s answers. Write them inline — or let your WooCommerce catalog sync in automatically. When something isn’t covered, it hands off instead of guessing.

How it works

Every article in your workspace knowledge base is split into chunks. Each chunk gets a vector representation (an "embedding") stored alongside the article. When a visitor asks something, we embed the query, find the most similar chunks, and inject them into the AI agent's prompt next to the conversation history.

The AI is instructed to answer from that material. If nothing relevant is found, it says so explicitly rather than fabricating an answer.

Adding content

Two ways to populate the knowledge base:

  • Write articles. Click Knowledge → New article and write (or paste) the content. Best for policies, FAQs, and answers the AI should reuse verbatim.
  • Sync your WooCommerce catalog. The WordPress plugin pushes your products in automatically — see the next section.

Edits re-embed the changed article automatically — there is nothing to rebuild or re-index by hand.

Tip
Start with your top 10–20 most-asked questions as short articles — that covers the bulk of repeat support volume. Add more as you spot gaps in the AI insights digest under Analytics.

WooCommerce catalog

On WordPress stores running WooCommerce, the Leadiosa plugin syncs your product catalog into the knowledge base — names, prices, descriptions, availability. Visitors ask "do you have this in size M" and the AI answers from your real catalog instead of guessing.

  • The plugin connects to your workspace during its first-run wizard — no keys to copy by hand.
  • Catalog content lives in a dedicated auto-managed entry; re-syncs replace it in place, so your hand-written articles are never touched.
  • Price or stock changed? Re-sync from the plugin and the AI answers with the new numbers.

Full plugin setup lives under Integrations.

Retrieval and sources

On each AI turn we retrieve the top five most similar chunks; chunks below a similarity floor are dropped, so weak matches never reach the prompt.

When the AI drafts a reply for an operator, the sources it used are listed next to the draft with similarity scores — you can check what it leaned on before sending.

Limits

PlanKnowledge baseArticles
FreeNot available — no AI on Free—
ProIncludedUnlimited (fair use)
Pro PlusIncludedUnlimited (fair use)

See AI agent for how the AI actually uses what you put here.

Under the hood (for the curious)

Embeddings are produced by a platform-managed model (OpenAI by default) — we handle the provider relationship so you don't have to shop for an API key. Vectors are stored per-workspace and isolated from every other workspace; nothing in your knowledge base reaches another customer's AI.

Knowledge-base content is treated the same way as conversation content for privacy — see Privacy Policy.

PreviousAI agentNext Integrations