Skip to main content

@Mentions and References

Direct the AI to specific data objects with intelligent autocomplete.

What are @Mentions?​

@Mentions let you explicitly reference Sources, Views, Input Objects, and even specific columns by typing @ followed by the name. This tells the AI exactly which data to use, avoiding ambiguity and ensuring accurate results.

As you type, smart autocomplete suggests matching objects—you usually only need a few letters.

@ Autocomplete

Type @ to see intelligent suggestions

Why @Mentions Matter​

Precision: No guessing—the AI knows exactly which dataset you mean.

Speed: Fast autocomplete means less typing.

Context: Reference Views created earlier in the conversation.

Multi-source analysis: Combine different datasets clearly in one request.

Column-level reference: Mention specific columns to guide transformations.

How to Use @Mentions​

Mentioning Data Objects​

Type @ and start typing the object name:

@[orders] Show total revenue
Join @[customers] with @[orders] on customer_id
Using @[monthly_revenue], calculate year-over-year growth

Autocomplete shows matches after just 2-3 letters. Press Enter or click to select.

@ Selection

Autocomplete finds matches with minimal typing

Mentioning Columns​

Reference specific columns within your request:

@[orders] Group by @[region] and sum @[total_amount]
Filter @[customers] where @[last_purchase_date] is in the last 30 days

Column mentions help the AI understand exactly which fields to use, especially when column names are similar across datasets.

Multiple Mentions in One Request​

Combine several references:

Join @[orders] with @[products] and @[categories] to show
revenue by category
Compare @[q4_2023_sales] with @[q4_2024_sales]

The AI understands relationships between all mentioned objects.

What You Can Mention​

Sources - Your uploaded files or database connections

  • Blue nodes in the graph
  • Original data imports

Views - AI-created transformations

  • Colored nodes (by View type)
  • Previously created analyses

AI Tables - LLM-enhanced Views

  • Purple nodes
  • Data with intelligent columns

Input Objects - Variables and scenarios

  • Yellow nodes
  • Configurable parameters

Columns - Specific fields within objects

  • Works for Sources, Views, and AI Tables
  • Helps clarify which field you mean

Mentionable Objects

All colored nodes in your graph are mentionable

Autocomplete Features​

Fast Matching​

Autocomplete finds matches with minimal typing:

  • Type @or → suggests "orders"
  • Type @cust → suggests "customers", "customer_segments"
  • Type @rev → suggests "revenue_by_month", "revenue_summary"

Smart Filtering​

Autocomplete prioritizes:

  1. Recent objects you've worked with
  2. Exact prefix matches
  3. Contains matches anywhere in the name

Context Awareness​

If you're mid-sentence, autocomplete understands:

Show revenue from @[or...

Knows you want a data object, suggests Sources and Views

@[orders] group by @[re...

Knows you want a column, suggests columns from the orders object

Common Use Cases​

Joining datasets: Join @[customers] with @[orders] on customer_id

Filtering existing Views: @[high_value_customers] filter to only this year

Building on prior work: Using @[monthly_trends], forecast next quarter

Combining analyses: Compare @[forecast_v1] with @[forecast_v2]

Column-specific operations: Sort @[products] by @[profit_margin] descending

Multi-step references: Take @[filtered_orders], group by @[category], then show top 5

Tips & Best Practices​

Always use @mentions: Even if there's only one dataset, mentioning it explicitly helps the AI understand context.

Be specific with similar names: If you have "orders_2023" and "orders_2024", always use @mentions to avoid confusion.

Mention columns when ambiguous: If multiple datasets have "id" or "name", reference the specific column: @[orders.order_id]

Use full names if autocomplete is unclear: Type more letters to narrow suggestions.

Check the selected object: After autocomplete fills in the name, verify it's the right one before sending.

Reference Views by meaningful names: Rename Views to descriptive names (like "active_users") so they're easier to mention later.

@Mention Shortcuts​

Tab to accept: Hit Tab to accept the first autocomplete suggestion Arrow keys to navigate: Use up/down arrows to select from the list Escape to cancel: Press Escape if you don't want to mention anything Space after selection: Autocomplete adds ] and a space automatically

Autocomplete Navigation

Keyboard shortcuts for fast @mention selection

Troubleshooting​

Autocomplete not showing?

  • Make sure you've typed @ first
  • Wait a moment for suggestions to load
  • Check that the object exists in your Workbook

Wrong object suggested?

  • Type more letters to narrow the search
  • Scroll through suggestions with arrow keys
  • Verify the object name in the left sidebar

Column mention not working?

  • Try format: @[dataset.column_name]
  • Check spelling of the column name
  • View the data object to confirm the column exists