> ## Knowledge Base Index
> Fetch the complete knowledge base index at: https://help.upsellplus.com/sitemap.xml
> Use this file to discover available pages before exploring further.
> Pure-Markdown content can be obtained by appending a '.md' suffix to the content URLs listed in the sitemap (without the trailing slash).

# The AI Config Tab - Fine-Tuning Autopilot Recommendations

## Overview

The **AI Config tab** activates when you enable **Autopilot** on the Products tab. It lets you configure AI-driven recommendations instead of manually selecting products for each offer.

## Intent Settings

The **Intent** setting defines what type of recommendations to make:

- **Related Products:** Similar items in the same category (colors, sizes, variations)
- **Complementary Products:** Items that pair well together
- **Upgrade:** Premium or higher-tier versions of products
- **Cross-Sell:** Products from different categories the customer might need
- **Bundled Products:** Items that work together as complete solutions

## Max Products

Controls how many recommendations to show:

- **2-3 products:** High-conversion, focused upsells; ideal for Checkout offers
- **4-5 products:** Balanced approach; works for most use cases
- **6-8 products:** Maximum variety; better for Thank You Page offers

Test different numbers to find what works for your store.

## Inclusion Filters

Guardrails that tell AI which products can be recommended.

### Filter Types:

- **Collections:** Limit to specific product collections
- **Tags:** Use product tags as boundaries (e.g., `in-stock`, `seasonal-current`)
- **Product Types:** Restrict to specific product types
- **Metafields:** Use custom fields for advanced control (e.g., `margin_category = high-margin`)

### Filter Logic:

- **Same category:** OR logic (Collection A OR B)
- **Different categories:** AND logic (Collection A AND Tag "current")

Start with one filter and add more if needed. Too many filters can leave no products to recommend.

## Real-World Examples

**Fashion Store:** Complementary Products intent, 4 Max Products, filtered by Apparel collection + seasonal-current tag + in-stock tag

**Electronics:** Upgrade intent, 3 Max Products, filtered by Premium Accessories product type + high-margin metafield

**Grocery:** Bundled Products intent, 5 Max Products, filtered by (Organic OR Pantry Staples) AND (meal-kits OR recipe-ready)

## Monitoring Performance

Track these metrics:

- **Conversion Rate:** Target 2-5% depending on offer type
- **Average Order Value Lift:** Compare Autopilot vs manual selections
- **Relevance:** Review recommendations regularly for quality

**Iteration frequency:** Weekly reviews, monthly comparisons, quarterly adjustments for seasonal changes.

## Common Mistakes

- **Over-filtering:** Limits available products; expand your filters if conversion drops
- **Wrong Intent:** Test different intents against your goals; let data guide decisions
- **Max Products too high:** Causes decision paralysis; start with 4 and test lower
- **Ignoring seasonality:** Update filters quarterly as inventory changes

## Pro Tips

- A/B test different intents to find what resonates
- Use metafields for sophisticated control over recommendation quality
- Combine Autopilot with Smart Rules to segment recommendations by customer type
- Start with one offer, test, and expand once you see results

The AI Config tab automates product recommendations. Start simple, monitor performance, and iterate based on data.