AI Store Employee & MemoryEssentialIntermediate

Training Your AI: FAQs, Document Uploads & Store Rules

Learn how to feed your brand's sizing charts, return policies, shipping guidelines, and custom business rules into your AI employee's semantic knowledge base.

6 min readUpdated 2026-09-18
In This Guide
  • Upload PDF documents, text files, and custom FAQ pairs into your store's dedicated knowledge base.
  • Understand how semantic chunking and vector embeddings retrieve exact information during customer chats.
  • Define store rules for returns, exchanges, delivery timelines, and care instructions.
  • Use the interactive AI Knowledge Sandbox to test and refine how Aura answers questions before going live.
Prerequisites

Before configuring this feature, confirm that your store meets the following requirements:

Store Policy Documentation:Prepared text documents or PDFs of your store's policies, sizing charts, or FAQs.
conversora.io/admin/ai-employee
Training Your AI: FAQs, Document Uploads & Store Rules
Click to enlarge

Interface view for this workflow.

1. Adding Custom Question-and-Answer Pairs

The fastest way to teach your AI specific facts is by creating explicit FAQ pairs. FAQ pairs are ideal for concise, unambiguous rules such as delivery charges, return fees, and customer service contact numbers.

When creating an FAQ pair, write the question in the natural, colloquial phrasing that your customers commonly use. Aura’s semantic vector index automatically matches questions even if the customer uses different synonyms, slang, or regional phrasing (e.g., 'delivery charge koto?' vs 'what is shipping cost?').

1

Navigate to Knowledge Base

In the admin console, select 'AI Employee' and click on 'Knowledge Base'.

Path:Admin Sidebar → AI Employee → Knowledge Base
2

Click 'Add FAQ'

Click the '+ Add FAQ' button in the upper-right corner of the FAQ panel.

Path:Knowledge Base → FAQs Tab → Add FAQ
3

Enter Question & Answer

Provide the primary question and a comprehensive, friendly answer.

Path:FAQ Modal → Question & Answer Inputs → Save

2. Uploading Policy Documents & Sizing Charts

For comprehensive policy manuals, terms of service, or multi-page product care guidelines, uploading PDF or text documents is more efficient than creating dozens of separate FAQ pairs.

Conversora automatically processes uploaded documents through its edge chunking pipeline. You can view the indexed chunk count directly in the document list and toggle individual documents active or inactive at any time.

1

Open Document Uploader

Switch to the 'Documents' tab inside the Knowledge Base panel.

Path:Knowledge Base → Documents Tab
2

Upload File

Drag and drop your PDF or TXT file (up to 10MB) into the upload zone.

Path:Documents Tab → Drag & Drop Zone
3

Verify Processing Status

Wait 5-10 seconds for the status badge to transition from 'Processing' to 'Indexed'.

Path:Document Row → Status: Indexed

Practical Business Scenarios

How leading merchants implement this functionality in daily operations:

Seasonal Holiday Return Policy Update
Situation: During December, a merchant extends their standard 7-day return policy to 30 days for holiday gift purchases.
Best practice: Update the return policy FAQ in the Knowledge Base. The AI immediately begins quoting the 30-day window to all chat inquiries without modifying website code.

Troubleshooting & Common Issues

Diagnose and resolve frequent failure points quickly:

Problem: AI says 'I don't have information on that' for a policy contained in an uploaded PDF
Why this occurs: The PDF contains scanned raster images instead of selectable text, preventing the text extraction parser from reading its contents.
Resolution: Ensure your PDF contains selectable digital text, or copy and paste the policy text directly into the Knowledge Base text editor.

Frequently Asked Questions

How long does it take for new knowledge to take effect in AI chats?

Vector indexing completes within 5 to 10 seconds. Once the document shows 'Indexed', Aura incorporates the information on the very next customer message.