The client's own website comes in through the Sitemap tab. Everything else the AI should know goes in through Sources and Data sheets: a brochure, a product manual, a partner's webpage, a price list. This article explains what to add where.
Before you start
You need edit rights on the company memory. See What AI Memory is and how it is organised.
Register and crawl the client's own website in the Sitemap tab, not in Sources. See Approving a website crawl.
Which place to use
What you have | Where it goes |
|---|---|
The client's own website | Sitemap tab |
Another website, or a single webpage | Sources > Website or Webpage |
PDFs, Word documents, text files | Sources > upload |
Prices, rates, stock, anything in rows and columns | Data sheets |
Add a website or a webpage
Open AI Memory and go to the Sources tab.
Click Add Source. The Add Sources window opens.
Under Or add from, choose Website (crawl a whole site) or Webpage (a single page).
For a website, fill in:
Website URL
Name
Max Pages: how many pages to crawl, up to 500. The default is 50.
Click Start Crawling.
For a webpage, fill in Page URL and optionally Name, then click Add Page.
If you see This domain is managed by your Sitemap., the address is the client's registered website. You can't add it here. Manage it in the Sitemap tab instead.
Upload files
On the Sources tab, click Add Source.
Drag files into the upload area, or click browse files. Supported formats: PDF, TXT, MD, DOCX, HTML, JSON, RTF, XLSX, XLS and CSV.
Or choose PDF, Documents, Excel or CSV under Or add from to upload only that kind of file.
Click Add Source (or Upload).
Tip: You can also drag files straight onto the Memory or Library tab. They are uploaded and you are taken to Sources.
The Source limit line in the window shows how many sources the company has against its limit.
Follow and manage sources
Each source shows a status:
Status | Meaning |
|---|---|
Pending | Waiting to start |
Crawling | Pages are being fetched |
Processing | Content is being read into the memory |
Completed | Done and available to the AI |
Failed | Something went wrong. Try Re-crawl, or delete and add it again |
Use the search box and the Websites, Documents and Data filters to find a source. The menu on each source offers:
View Pages: the pages or content collected.
Extract Memories: the memories created from the source.
Re-crawl: fetch the source again, for example after the page changed.
Delete: permanently removes the source and all its crawled pages. This can't be undone.
Data sheets
Data sheets are tables you maintain yourself: a price list, delivery rates, stock. When someone asks the AI a price, it looks the row up and answers only from what it finds. This makes data sheets the right place for any number the AI must get exactly right.
Import a spreadsheet
Open the Data sheets tab.
Click Import file and pick an Excel or CSV file (.xlsx, .xlsm, .xls or .csv), or drag the file onto the page.
Each file becomes a folder, with one sheet per tab.
Create a sheet from scratch
Type a name in the box at the bottom, for example "Price list 2026".
Click Create, open the sheet and paste or type your rows.
Click Save changes.
Use New folder to organize sheets. Each sheet shows its row count and when it was last updated. The date turns amber after 30 days as a reminder to check it's still current.
Build a sheet from the website
If pages in the Sitemap are classified as products, Data sheets can offer to build a product sheet from them. Click Sync to read those pages into one table, so the AI can sort and filter them (prices, years, what is in stock). A sheet built this way is marked from website. Use its refresh button to read the pages again and replace the rows.
The Library
More > Library lists the individual memories saved for the company, including the ones created from crawled pages and sources. Use the Scope, Type, Tags and Source filters to find one, or click Add to write a memory by hand.