Document Upload

If you have ready documents, you can import them by uploading instead of typing the content one by one. HaveAI extracts the text from the document. Then you can review the information prepared from this text and save it for your assistant to use.

Supported Formats and Limits

The panel's file picker lists the PDF, DOCX, DOC, TXT and MD (Markdown) formats. The size limit per file is 25 MB. You can send up to 10 files in one upload.

If your old DOC file cannot be processed, try re-saving it as DOCX or as a PDF with selectable text. If your Excel file does not appear in the file picker, you can export the needed table to a PDF with selectable text or to TXT. Before exporting, check that headings, rows and values remain understandable.

Why Text Extraction Matters

HaveAI extracts text from documents. Scanned, photo-based PDFs or documents made only of images may not provide enough text. Documents whose text you can select and copy are more suitable.

Keep headings in tables and lists. Clearly state which product a price belongs to and which service a duration belongs to. Also add important information found in an image as text.

Importing Information from a Document

Open the document section

Open the document section under AI Training in the panel.

Select or drag the file

Select the file or drag it into the upload area.

Wait for processing

Wait for the file to be processed. At this stage the document's text is extracted.

Open the Train option

When processing is complete, open the document's Train option. HaveAI can analyze the content and prepare brand information, FAQ, product and service, or free-form information records.

Review the prepared records

Review the prepared records. Fix wrong, missing, outdated or unnecessary information; select the ones you want to use.

Save your selection

Save your selection. Check the records in the sections they were imported to and try the related questions on the AI Test screen.

Processing and adding to training are separate stages

Completing the upload alone does not mean the information in the document is ready to be used by the assistant. Processing the file and adding it to training are separate steps.

Tracking Statuses

In the document list, check the file's processing and training statuses separately. Even if a file is processed, the training step may still be pending. With the Review option offered after training, you can go over the prepared content again.

The list may also show technical information such as the number of chunks. When judging whether the assistant is ready to use the content, look not only at the chunk count but also at the training status and the saved information.

If processing fails, check the file's format, size and whether it contains text. After fixing the issue, you can use the reprocess option or upload again in a suitable format.

For Better Results

Use headings, short paragraphs and clear lists. Do not leave conflicting prices or conditions in the same document. When a document changes, also update the records the assistant uses; do not assume that uploading a new file automatically replaces all previous records.

Redaction of Secret Information

HaveAI automatically masks secrets it can recognize in the extracted training text, such as API keys, passwords and private keys. This control recognizes certain information formats; it does not mean all secret or personal data will be found.

Check your document before uploading and remove sensitive information not needed for training. Remember that masking in the training text does not mean the source file you uploaded is also cleaned.