💬Get free consultation

What goes in each data source

Why add data sources

We've trained AI on your own data. By adding your own data sources, you can train AI assistant about your specific products, services, and business policies.

This helps the AI give more accurate and personalized responses to your customers.

Train AI

Type of data sources

There are two main types of data sources:

• Store data: Information already available in your Shopify store that can be automatically synced
• Custom knowledge: Additional information you manually add to enhance AI knowledge

What product information is synced to Chatty?

Available product information that is synced to Chatty:

• Specific product details (name, descriptions, images, vendor, type)
• Product variants (colors, sizes, styles)
• Price ranges and comparisons
• Product features and specifications
• Collection
• Product inventory
• Product metafields

Your product information will be updated daily at 12:00 AM PST.

What types of product metafields are supported?

We support all metafields in this list Shopify documentation on metafield types (opens in a new tab), except:

• Reference type:
  • collection_reference
  • customer_reference
  • file_reference
  • metaobject_reference
  • mixed_reference
  • page_reference
  • product_reference
  • product_taxonomy_value_reference
  • variant_reference
• Other:
  • Link
  • URL

Some main Shopify pages will be auto-synced when you activate AI. What are they?

• Shipping policy
• Return policy
• Privacy policy
• Terms of service
• FAQ
• Contact us
• About us

If your Shopify pages are not auto-synced, go to AI agent → Training data → Knowledge base, open Sync from your store, and click Sync store pages. Re-activating AI also triggers a sync.

---

1. Q&A

Best for: Specific customer questions you get repeatedly

What to add here:

• Policies: Return policy, shipping times, warranty info
• Business info: Store hours, location, contact methods
• Troubleshooting: "Product not working," "How to care for this item"

Don't add here:

• Product information: Price, description, size, type, variant, etc. - That's automatic
• Long documents - use Files instead
• General behavior instructions - use Custom Instructions

Where should I add product advice for questions like "Which size for wide feet", "Best for sensitive skin"?

With product advice, you should add information to product description. If you don’t want to show it in public stores, add it to product metafield and sync to Chatty.

Adding product care instructions using metafields (opens in a new tab)

Adding a pop-up size chart to your product pages (opens in a new tab)

Coming soon: Add FAQs for specific products.

---

2. General instructions

Best for: How your AI should behave and respond in general

What to add here:

• Brand voice & tone: "Be friendly but professional, use simple language"
• Response style: "Always ask if they need anything else," "Keep answers under 3 sentences"
• Industry instruction: “You are a professional skincare advisor, your goal is to help customers find the best choice for their skin.”
• Decision support: "Present options with clear pros/cons," "Highlight best value options"
• Sales approach: "Suggest complementary products," "Mention current promotions"
• Boundaries: "Don't give medical advice," "Always verify large orders with humans"
• Format preferences: "Use bullet points for multiple options"

Don't add here:

• Specific information - use Q&A instead
• Basic product info (price, description) - that's automatic
• Long policy text - use Q&A for policies
• Specific scenario: “When customer mention about table, suggest dinner table” - Use custom scenarios
• Refer actions: “If the customer asks about 'size guide' send them 'size-fit.txt' file”. AI can't take action, but send messages.

---

3. Scenarios

Best for: Specific situations or customer journeys

What to add here:

• Customer types: "First-time buyer," "Returning customer," "Bulk buyer"
• Shopping situations: "Gift shopping," "Last-minute order," "Special occasion"
• Problem resolution: "Order arrived damaged," "Wrong size received"
• Sales situations: "Cart abandonment," "Price comparison," "Bulk pricing inquiry"
• Seasonal flows: "Holiday shipping deadlines," "Back-to-school shopping"

Don't add here:

• Simple Q&A pairs - use Questions instead
• General behavior rules - use Instructions

---

4. URLs

Best for: Linking to existing helpful resources

What to add here:

• Size guides & fit charts
• Care instruction pages
• Video tutorials or demos (coming soon)
• FAQ pages with comprehensive info
• Policy pages (privacy, terms)

Don't add here:

• URLs that might change frequently
• Password-protected content
• URLs with interactive content, embedded content - iframe, API, external resources

---

5. Files

Best for: Bulk information and detailed documents

What to add here:

• Detailed care/maintenance guides
• Detailed product manuals
• Technical specification sheets
• Compatibility charts
• Training manuals for complex products
• FAQ compilations from other sources

Don't add here:

• Single Q&A pairs - use Questions instead
• Simple policies - use Q&A instead
• Files over 2MB or with complex formatting
• Images and tables are supported yet

⚠️ Notes

• Avoid duplication: Don't put the same info in multiple content types
• Test everything: Ask AI your questions after adding content

6. Smart recommendations

Best for: Tell AI which products to recommend in each case. Steps: AI product recommendations.

What to add here:

• Product list for each case: Best sellers, New arrivals, Sales, Special occasions