Upload CSV files

​

Uploading CSVs

1

Select Upload CSVs

From the Import Data page, click the Upload CSVs card to start uploading your data.

You can also click Use Demo Data to explore AstroBee with sample data, or Connect Sources to link external systems like HubSpot or Salesforce.

2

Add files and name your data source

In the Upload Data modal:

1. Enter a Data Source Name to identify your uploaded data
2. Drag and drop your CSV files or click Browse to select them

• Supports CSV files up to 100MB each
• Upload multiple CSVs - each becomes a separate source table
• Remove accidentally uploaded files using the trash icon

Click Create Data Source to begin processing.

Pattern extraction and data layer generation typically take a few minutes depending on the number and size of your files. Feel free to close this tab and come back later. When it’s done, you can ask questions in plain language about your data!

3

Pattern extraction and data layer generation

AstroBee’s Pattern Extraction Agent automatically analyzes your data to discover relationships between tables.

The left panel shows three tabs:

• Discovered Joins - Relationships found between your tables
• Unmatched Fields - Fields that couldn’t be automatically joined (you can manually map these later when editing your data layer)
• Tables - Your source tables being processed

The right panel shows the Pattern Extraction Agent working as a sub-agent, analyzing columns and discovering joinable patterns across your data. It applies entity resolution and normalization to match records even when formats differ between sources.Click on any discovered join to see pattern details including:

• Source and target columns
• Extraction logic (exact match, fuzzy/substring match)
• Confidence and coverage scores
• Live matches showing actual matched values

4

Review your data layer

Once generation completes, review your integrated data layer.

The left panel shows the Tables tab listing your entities with descriptions and relationship counts. The right panel shows AstroBee’s chat where it summarizes what it built, including coverage statistics and potential analytics use cases.Explore your data layer using the tabs:

• Tables - View entities and their relationships
• Description - Read the auto-generated documentation about your integrated data
• Data Map - Visualize how your sources connect

​

CSV filename icon matching

• accounts_salesforce.csv → Salesforce icon
• campaigns_hubspot.csv → HubSpot icon
• analytics_ga4.csv → Google Analytics icon
• orders_pg.csv → PostgreSQL icon

​

Viewing source data

• Tables - List of all source tables with search
• Description - Auto-generated documentation
• Data Map - Visual representation of table relationships

• Column headers with data types (Text, Integer, Timestamp, etc.)
• Actual data from your source
• Download CSV button for exporting data

​

Adding new data to an existing data layer

1. The new files become a new data source alongside your existing sources
2. AstroBee automatically generates a new data layer version incorporating all sources
3. New tables appear grouped under their data source name

​

Troubleshooting

Trailing quote on quoted field is malformed

Issue: Quote characters (”) within your data aren’t properly escaped or closed.

Solution:

• Ensure quoted fields are properly closed with matching quotes
• Escape quotes within text by doubling them ("") or using a different quote character
• Check that your CSV follows consistent quoting rules

Quoted field unterminated

Issue: A quoted field starts with a quote but never closes it.Solution:

• Find the row mentioned in the error and ensure all quoted fields have closing quotes
• Remove line breaks within quoted fields or properly escape them
• Check for stray quote characters in your data

Too many fields / Too few fields

Issue: Rows have inconsistent numbers of columns compared to the header row.

Solution:

• Ensure all rows have the same number of columns as your header
• Check for extra commas or missing commas in the data
• Remove or fill empty columns to maintain consistent structure
• Look for line breaks within data fields that might split a single row

Unable to detect delimiter

Issue: The parser cannot automatically determine what character separates your columns (comma, semicolon, tab, etc.).Solution:

• Ensure your CSV uses standard comma separators (,)
• If using semicolons or tabs, convert to comma-separated format
• Check that delimiter usage is consistent throughout the file
• Avoid mixing different separators in the same file

Empty CSV or no data rows

Issue: CSV file appears to be empty or contains only headers with no actual data.

Solution:

• Ensure your CSV file contains actual data rows below the header row
• Check that the file wasn’t corrupted during export or transfer
• Verify the file has content beyond just column headers
• If exporting from another system, make sure to include the actual data records

File size or format issues

Issue: CSV files larger than 100MB or in unsupported formats.Solution:

• Split large files into smaller chunks under 100MB each
• Ensure your file has a .csv extension
• Save Excel files as CSV format before uploading
• Use UTF-8 encoding for files with special characters

Column headers renamed to generic names

Issue: Your column headers appear as column, column_1, column_2, etc. instead of the original names.Why this happens: AstroBee normalizes headers to be compatible with database systems. Headers containing only Unicode characters (e.g., 名前, 日期), special characters (e.g., ###, @@@), or starting with numbers get renamed to generic column names.Solution:

• Use alphanumeric headers with underscores (e.g., customer_name, order_date)
• If your headers are in another language, add an English prefix or suffix (e.g., name_名前)
• Headers like First Name or Order-ID work fine and become first_name and order_id

​

Next steps