Skip to main content

Fino AI Assistant

Fino is Infino's natural language interface for querying your data stack. Type questions in plain English and get responses with summaries, data tables, visualizations, and technical queries. Works across all connected sources and datasets.

Where to find it

  • Fino AI Assistant (Home): the default landing page. It’s search-first and can start an inline conversation at the bottom of the page.
  • Threads: browse and manage all conversation threads via Query → Threads.

Creating Conversations

Start a conversation:

  1. Go to Fino AI Assistant (Home) in the sidebar
  2. Type your question into the main search box
  3. Press Enter (or choose Start a conversation from the search dropdown)
  4. The chat composer appears pinned at the bottom of the Home page (no navigation)

Fino can query across connected your data sources or datasets automatically.

Thread Management

Organizing conversations:

  • Auto-naming: Fino generates titles like "Flight Analysis - Airlines"
  • Rename threads: Click pencil icon → type new name
  • Status management: Three-dot menu → "Mark as Done" or "Mark as Ongoing"

To see all threads and manage them in bulk, use Query → Threads.

Filtering and sorting (top bar):

  • Sort dropdown: Change the order of conversations (e.g., by date, name)
  • Filters dropdown: Apply custom filters to find specific conversations
  • Quick filter tabs:
    • All: Shows all conversations with total count
    • Created by you: Conversations you started
    • Shared with you: Conversations others shared with you
    • Resolved: Completed or closed conversations

Response Format

Response Components

Every Fino response contains four sections:

  1. Summary: Natural language explanation of results
  2. Data Table: Sortable, filterable tabular data
  3. Visualization: Charts when data supports visual display
  4. Technical Query: Query DSL or SQL that was executed

Data Table Features

Table interactions:

  • Sort: Click column headers
  • Filter: Use column filter options
  • Cell viewer: Click truncated cells to see full content
  • Export: Download as CSV, Excel, or JSON
  • "Open in Explore": Send results to detailed data exploration interface

Chart Features

Chart interactions:

  • Clickable elements: Click bars, pie slices, or data points
  • Context menu: Right-click for copy/export options
  • "Add to Dashboard": Save visualization as dashboard panel

Visualization Overlay

Fino can overlay new data on existing charts by combining them with previous visualizations.

Overlay capability:

  • Context preservation: Maintains previous chart structure and data
  • Data layering: Combines current chart with previous visualization
  • Visual continuity: Merges multiple charts into a single view

How to trigger overlay:

  • Click "Overlay with previous" button on any chart (appears on latest charts only)
  • Button automatically sends: "Overlay this vizualisation with previous one"
  • Result: Current chart data combined with the previous visualization

Chart Comment Box

Contextual Follow-ups

When you click chart elements, Fino displays a Comment Box with AI-generated follow-up questions and key insights relevant to the selected portion of the visualization.

How it works:

  1. Click any chart element (bar, slice, point)
  2. Comment Box appears with:
    • Follow-up questions specific to your selection
    • Key insights about the selected data
  3. Click a suggestion or type your own question
  4. Fino responds with context from your chart selection

Example with flight data:

  • Click airline bar → Shows insights about that airline + questions like "Show routes for American Airlines"
  • Click delay metric → Shows delay insights + questions like "What causes delays for American Airlines?"
  • Click time period → Shows period insights + questions like "Show American Airlines performance in Q4"

Help Fino Feature

Data Context Enhancement

The "Help Fino" button appears when you have a dataset selected.

How to use:

  1. Select dataset from dropdown
  2. Click "Help Fino" button
  3. Add field descriptions: Explain what columns represent
  4. Provide data context: Business-specific information about the data
  5. Save metadata

Benefits:

  • More accurate interpretation of your questions
  • Better suggested follow-up questions
  • Contextually relevant chart types

Conversation Features

Context Memory

Fino remembers conversation history within each thread:

  • Follow-up questions: "Show me more details about that"
  • Progressive refinement: "Focus on the top result"
  • Pronoun references: "Break that down by month"

Query Syntax Toggle

At the bottom of each response, toggle between:

  • Query DSL: OpenSearch/Elasticsearch format
  • SQL: Standard SQL queries

Execution modes

Use the Ask | Generate | Raw controls above the chat composer:

  • Ask (default): generate a query (if needed), review, and execute
  • Generate: generate a query + review only (no execution)
  • Raw: execute the exact SQL/QueryDSL you provide

Integration Features

Dashboard Creation

From visualizations:

  • "Add to Dashboard" button on charts
  • Select target dashboard or create new one
  • Chart becomes persistent dashboard panel

Data Export

Export options:

  • Tables: CSV, Excel, JSON formats via Export button
  • Charts: Right-click → Copy image or export underlying data
  • Full responses: Copy/paste summaries and findings

Connection Status

WebSocket Indicator

Fino uses WebSocket for real-time communication:

  • Connection status: Visible in conversation interface
  • "Thinking" indicator: Shows when Fino is processing
  • Streaming responses: Results appear as they're generated

Troubleshooting

Common issues:

  • "No index selected": Choose an index from dropdown before asking questions
  • Slow responses: Complex queries take time; watch "thinking" indicator
  • Connection problems: Check WebSocket status indicator; refresh page if needed
  • No visualizations: Not all data supports charts; tables always provided
  • Unclear results: Fino works best with specific questions rather than very broad requests