Knowledge files

Files in the Knowledge base have two separate permissions: Use to answer lets the AI consult extracted content; Allow sending lets it share the file with a customer.

Demo organization: Casa Nativa. Interface shown in Spanish.

Add a fileCopy link

1. Open Knowledge base → Files.
2. Select Add file and choose a file.
3. Enter a Title and a Description for the agent explaining when to use it.
4. Select Use to answer and/or Allow sending.
5. Select Save file and wait for processing.
6. Open the file and check What the AI read, page by page, before relying on its contents.

The UI accepts JPG, PNG, PDF, Word, and Excel. Images can be up to 5 MB and documents up to 100 MB. Content ready means extraction succeeded; Available to send; content not readable means the agent may share it but cannot answer questions from its text.

What the agent knowsCopy link

In every conversation, on any channel, the agent receives the list of files with their title, their Description for the agent, how many pages each PDF has, and what it may do with each one. Write descriptions that say what the file contains and when to use it.

Ricky reads each PDF page by page, as it looks: the text, the tables, the prices, and what each photo shows. That is why a PDF takes longer to process than other files.

• The agent does not receive the files' content in every conversation: it opens or searches the file it needs, when it needs it.
• The agent can also look at a page of a PDF, or at an image, and ask about a detail: a photo, a table, or the layout.
• Files with Allow sending can be sent to the customer on WhatsApp only.

If you see Incomplete upload or File unavailable, delete the failed upload and try again. If extraction is wrong, correct the source file and upload it again. File permission does not replace selecting the agent tools.