# Knowledge Management

## Where to add knowledge

In the console open **Knowledge** ([https://base.hi.speak-to.ai/knowledge](https://base.hi.speak-to.ai/knowledge)).

### Talk to add knowledge (web)

On the web console, use **Talk to add knowledge** ([https://base.hi.speak-to.ai/knowledge/voice](https://base.hi.speak-to.ai/knowledge/voice)): speak naturally, polish the transcript, and save it as a knowledge article. Same idea on the invite-only iOS companion Voice tab. Usage is metered on your plan. (Visitor speak-to-chat on the public Twin is not live yet.)

### Upload files

Drag and drop or browse. Supported formats include:

- Text: `.txt`, `.md`, `.html`
- Documents: `.pdf`, `.docx`, `.rtf`, `.odt`, `.epub`

Files are processed into searchable chunks. Large uploads may take a few minutes.

### Quick Add

Add a calendar or booking link, video URL, or short note without a full file — useful for Calendly, YouTube, or FAQ snippets. The Twin can share those links in answers; inline YouTube or image embeds in chat are not live yet.

### Explore

Browse and edit indexed chunks. Fix titles or remove passages that confuse visitors.

### Suggested

After you have documents, **Reflect** surfaces gaps. Approve or dismiss suggested snippets to fill holes.

### Website import

Paste a public site or section hub URL to draft knowledge for your approval (markdown mirrors help SPA sites).

### Email ingest

Forward email from your account address to your twin's ingest address (shown in Knowledge). Messages land in a review queue before indexing. This is for *you* adding content, not for emailing your audience from the Twin.

## Best practices

- Start with one strong doc: FAQ, bio, or recent deck
- Prefer clear headings and short sections
- Update when your messaging changes — no retraining required

---

**Next:** [How Responses Work](/guide/ai-twin-owners/how-responses-work)