Skip to main content

Context Enrichment

Fino automatically analyzes your data sources to understand their structure, meaning, and business context. This semantic understanding powers more accurate natural language queries, better visualizations, and smarter insights.

Overview​

When you connect a data source or import a dataset, Fino's AI:

  1. Analyzes the schema - Examines tables, fields, and relationships
  2. Generates descriptions - Creates human-readable explanations
  3. Identifies patterns - Detects business metrics, entity types, and data categories
  4. Builds a glossary - Defines domain-specific terms and acronyms

This context is editable—you can review, refine, and enhance what Fino generates to improve accuracy.

Accessing Context​

For Connected Sources​

Navigate to Data → Sources, select a connection or dataset, then click the Context tab.

For Local Datasets​

Navigate to Data → Sources → FinoDB, select a dataset, then click the Context tab.

Connection-Level Context​

Connection-level context applies to an entire data source (all tables/indices within a connection).

Domain​

A badge identifying the business domain of your data source.

Examples: E-commerce, Healthcare, Finance, Logistics, HR

To edit: Click the domain badge → type new value → press Enter or click ✓

Summary​

An AI-generated description of what your data source contains and its purpose.

To edit: Click the summary text → modify in the textarea → click Save

Data Categories​

Logical groupings of related tables within your data source.

FieldDescription
NameCategory identifier (e.g., "Customer Data", "Order Management")
DescriptionWhat this category represents
TablesList of tables belonging to this category

To add: Click "+ Add Category" → fill in details → click Save

Key Metrics​

Important business metrics tracked in this data source.

FieldDescription
NameMetric name (e.g., "Monthly Revenue", "Customer Churn Rate")
DescriptionWhat this metric measures
SourceTable or calculation source

To add: Click "+ Add Metric" → fill in details → click Save

Entity Types (SQL connections only)​

Core business entities and their identifying fields.

FieldDescription
NameEntity name (e.g., "Customer", "Order", "Product")
DescriptionWhat this entity represents
IdentifierPrimary key or unique identifier field

To add: Click "+ Add Entity" → fill in details → click Save

Glossary​

Business terms, acronyms, and domain-specific vocabulary.

FieldDescription
TermThe word or acronym (e.g., "MRR", "ARR", "SKU")
DefinitionPlain-language explanation

To add: Click "+ Add Term" → fill in details → click Save

Database Context (SQL Connections)​

For SQL-based connections (BigQuery, Snowflake, PostgreSQL, etc.), Fino generates additional schema-level context that captures table structures and relationships.

Accessing Database Context​

  1. Navigate to Data → Sources
  2. Select a SQL connection
  3. Click the Context tab
  4. Switch to the Database sub-tab

Tables​

A list of all tables discovered in your database with their metadata.

FieldDescription
Table NameThe name of the table
Primary KeyThe column that uniquely identifies rows
DescriptionAI-generated explanation of the table's purpose
PurposeThe business function this table serves

Table Relationships (Joins)​

Fino automatically detects relationships between tables based on foreign key patterns and column naming conventions.

FieldDescription
From Table → To TableThe tables involved in the relationship
Join ColumnsWhich columns are used to join (e.g., order_id → id)
CardinalityThe relationship type (1:1, 1:N, N:1, N:N)

Views​

ERD View​

A visual Entity Relationship Diagram showing:

  • Tables as boxes with their primary keys
  • Relationship lines connecting related tables
  • Cardinality labels on connections
  • Hover tooltips for full details

Navigation:

  • Pan by dragging the canvas
  • Zoom with scroll wheel
  • Hover over tables or relationships for details

List View​

A browsable list of tables, each showing:

  • Table name and primary key
  • Outgoing relationships (this table references others)
  • Incoming relationships (other tables reference this one)

How Relationships Improve Queries​

When you ask Fino a question involving multiple tables:

  • Without relationships: Fino may not know how to join tables correctly
  • With relationships: Fino uses detected joins to automatically combine data across tables

Example:

  • Question: "Show me orders with customer names"
  • Fino knows orders.customer_id joins to customers.id (N:1 relationship)
  • Generates correct JOIN query automatically

Dataset-Level Context​

Dataset-level context applies to individual tables or indices within a connection.

Data Source Overview​

High-level information about the specific dataset.

FieldDescription
CategoryClassification of this dataset's content
DescriptionWhat data this dataset contains

To edit: Click the description text → modify → click Save

Field Descriptions​

AI-generated descriptions for each field/column in the dataset.

Need Review Workflow​

Fields are flagged for review based on AI confidence scores:

  • Need Review - Low confidence, user review recommended
  • Fields move out of "Need Review" when you click Looks Good or edit them

Field Card Details​

Each field card shows:

  • Field name - The column/field identifier
  • Type - Data type (text, number, boolean, date, etc.)
  • Description - AI-generated explanation of what this field contains
  • Synonyms - Alternative names for this field (helps with natural language queries)

Editing Fields​

  1. Click a field card to expand it
  2. Edit description - Modify the AI-generated text
  3. Add synonyms - Type alternative names and press Enter
  4. Remove synonyms - Click the × on any synonym badge
  5. Click Save to persist changes

Approving Fields​

  • Click Looks Good to approve a field without changes
  • This increases the confidence score and removes it from "Need Review"

Filters​

Pre-defined query filters that can be applied to this dataset.

FieldDescription
NameFilter identifier
SQL ExpressionThe WHERE clause condition
DescriptionWhat this filter does

To add: Click "+ Add Filter" → fill in details → click Save

YAML Editor​

For advanced users, a YAML editor provides direct access to the full context structure.

Accessing YAML Editor​

Click the YAML button in the context view header.

Use Cases​

  • Bulk editing multiple items at once
  • Copy/paste context between data sources
  • Version control friendly format
  • Programmatic updates

Validation​

The editor validates your YAML before saving and shows errors if the structure is invalid.

How Context Improves Queries​

Natural Language Understanding​

When you ask Fino a question like "Show me top customers by revenue":

  • Without context: Fino guesses which fields represent "customers" and "revenue"
  • With context: Fino uses your defined entities, metrics, and field descriptions to generate accurate queries

Synonym Resolution​

If you define "order_id" with synonyms ["order number", "purchase id"]:

  • Asking about "order numbers" correctly maps to the order_id field

Business Metric Accuracy​

Defined key metrics ensure consistent calculations across queries and dashboards.

Best Practices​

1. Review AI-Generated Context​

The AI does a good job, but domain expertise improves accuracy:

  • Check that field descriptions match your actual data meaning
  • Add business-specific synonyms for key fields
  • Correct any misidentified entity types or metrics

2. Prioritize High-Impact Fields​

Focus review time on:

  • Fields used in common queries
  • Key metrics and KPIs
  • Customer/user identifiers
  • Financial and revenue fields

3. Keep Glossary Updated​

Add terms as your team uses them:

  • Internal acronyms (e.g., "CAC" = Customer Acquisition Cost)
  • Industry jargon
  • Company-specific terminology

4. Use Descriptive Categories​

Well-organized data categories help users discover relevant tables:

  • Group by business function (Sales, Marketing, Operations)
  • Group by data type (Transactional, Reference, Analytics)

Troubleshooting​

Context Not Generating​

  • Check connection status - Ensure the data source is accessible
  • Verify permissions - Fino needs read access to analyze schemas
  • Wait for processing - Large schemas may take a few minutes

Edits Not Saving​

  • Check for validation errors - Required fields must be filled
  • Refresh the page - Try re-loading if saves seem stuck
  • Check network - Ensure you have connectivity

AI Descriptions Seem Wrong​

  • Edit directly - Click to modify any description
  • Add context - More synonyms and glossary terms improve future generations
  • Use YAML editor - For bulk corrections

Next Steps​

  • Review your context - Start with high-priority data sources
  • Add business glossary - Define your domain terminology
  • Test with queries - Ask natural language questions to see context in action
  • Iterate - Refine context as you discover improvements