Recommend Items That Maximize a Goal

Aito's _recommend endpoint finds the items most likely to achieve a specific outcome for a specific user. It combines collaborative filtering with Bayesian inference — no feature engineering or model training required.

Personalized recommendations — products ranked by purchase probability for each individual user

Try Live Recommendations

These queries run against a demo grocery store with real user behavior data. Notice how different users get different results from the same product catalog.

{
  "from": "impressions",
  "where": {
    "context.user": "veronica",
    "product.id": {
      "$and": [
        {
          "$not": "banana-organic-001"
        },
        {
          "$not": "quinoa-bulk-001"
        }
      ]
    }
  },
  "recommend": "product",
  "goal": {
    "purchase": true
  },
  "select": [
    "$p",
    "name"
  ],
  "limit": 5
}

How It Works

1. Record Interactions

Store user-item interactions (views, clicks, purchases) in a table:

{
  "context": { "user": "larry", "query": "milk" },
  "product": { "name": "Oat Milk", "tags": ["dairy-free"] },
  "purchase": true
}

2. Write a Recommend Query

Specify the user context, what to recommend, and what goal to optimize for:

{
  "from": "impressions",
  "where": { "context.user": "larry" },
  "recommend": "product",
  "goal": { "purchase": true },
  "limit": 5
}

3. Get Ranked Results

Items ranked by purchase probability for that specific user. Use $not to exclude items already in cart.

Key capabilities:

• Per-user personalization — different users get different rankings from the same catalog
• Goal-based optimization — optimize for purchases, clicks, or any boolean outcome
• Exclusion filters — remove already-seen or already-purchased items with $not
• Context-aware — factor in search queries, time of day, or any contextual signal

Where This Is Used

• E-commerce & Retail — product recommendations, personalized search, cross-sell suggestions

Get Started

Start for free → Upload interaction data and get personalized recommendations immediately.

Try the full e-commerce demo → See recommendations in action in a complete grocery store app.