Table Vault Knowledge: Structured Data for Smarter AI

What is Table Vault Knowledge?

Table Vault Knowledge is a powerful way to give Eloquens AI access to your structured data from spreadsheets and CSV files. Unlike traditional document uploads where data can be difficult to search and query, Table Vault organizes your information into searchable tables that AI can query like a database - making it perfect for product catalogs, pricing lists, customer data, inventory, and any structured information.

What Happens Behind the Scenes

When you upload CSV files to Table Vault, Eloquens AI transforms them into intelligent, queryable data sources:

Automatic Structure Detection: AI analyzes your CSV file and intelligently determines data types for each column

Smart Data Processing: Column names are automatically cleaned and standardized for optimal searchability

Intelligent Querying: AI can search, filter, and analyze table data to answer specific questions

Real-Time Access: Information is queried on-demand as emails are processed, ensuring AI uses current data

Precise Answers: Unlike document search that returns similar content, table queries return exact matches and specific data points

How Table Vault Differs from Traditional Knowledge

Traditional Document Upload (File Knowledge)

• Uploads PDFs, Word docs, Excel files as documents
• AI searches for relevant text passages
• Best for unstructured information and narrative content
• Returns approximate matches and context

Table Vault Knowledge (Structured Data)

• Processes CSV files into queryable tables
• AI performs precise queries like database searches
• Best for structured data like pricing, inventory, specifications
• Returns exact values and specific records

Example:

• Document approach: "Find information about Product X" → Returns paragraphs mentioning Product X
• Table Vault approach: "What's the price of Product X?" → Returns exact price from pricing table

How to Upload Data to Table Vault

Step 1: Prepare Your CSV File

Before uploading, ensure your CSV file is well-formatted:

• First row contains headers: Column names should be in the first row
• Consistent data: Each column should contain the same type of data
• Clean formatting: Avoid special characters in column headers
• UTF-8 encoding: Save your file with UTF-8 encoding for best compatibility

Tip: You can export CSV files from Excel, Google Sheets, or any database system.

Step 2: Navigate to Knowledge Manager

• Go to Knowledge Manager in the main navigation
• Click on the Tables section in the left sidebar
• You'll see a list of existing tables (if any)

Step 3: Upload Your CSV File

• Click the Upload button (or + icon)
• An upload wizard will appear

In the Upload Wizard:

1. Choose Your File

   • Click Select File or drag and drop your CSV file
   • Supported format: CSV files (.csv)
   • File is validated to ensure proper format

2. Name Your Table

   • Enter a descriptive table name
   • Use clear names like "product_catalog", "pricing_2024", or "customer_accounts"
   • Table name must be unique within your workspace
   • Only lowercase letters, numbers, and underscores are allowed

3. Review and Confirm

   • Review the file name and table name
   • Click Upload to process the file

Step 4: Processing and Confirmation

• Eloquens AI processes your CSV file
• Columns are analyzed and data types are assigned
• You'll see a success message with import statistics
• The table is now available in your workspace

Step 5: Associate Table with Workflow

Once your table is uploaded, connect it to workflows:

1. Go to Workflows → Select or create a workflow
2. Navigate to the Knowledge AI section
3. Click on the Tables tab
4. You'll see a list of all available tables in your workspace
5. Check the box next to each table you want to use in this workflow
6. Click Save changes or Continue

Note: You can associate multiple tables with a single workflow. AI will search all connected tables when processing emails.

How Table Vault Knowledge Works in Practice

When You Receive an Email

Question Analysis: Eloquens AI analyzes the email to identify questions or requests for information

Table Selection: AI determines which connected tables might contain relevant data

Intelligent Querying: AI formulates and executes queries against your tables like:

• "Find product X"
• "Get price for item Y"
• "Look up customer account Z"
• "Filter items by category"
• "Find records matching criteria"

Response Integration: Query results are incorporated into AI's response with specific, accurate data

What Gets Included in Responses

Exact Data Points: Specific values from table cells (prices, quantities, dates, etc.)

Multiple Records: Lists of matching items when queries return multiple results

Filtered Results: Data that meets specific criteria from the email

Structured Information: Tables or lists formatted clearly in the response

Managing Your Tables

To View Table Details

• Go to Knowledge Manager → Tables
• Click on any table name
• View table structure, columns, row count, and recent updates
• See which workflows are using the table

To Upload New Data

Creating a New Table:

• Follow the upload process described above
• Each CSV upload creates a new table

Updating Existing Data:

• Upload a new CSV file with the same table name
• Choose "Replace" mode to overwrite existing data
• Or choose "Append" mode to add new rows (if your system supports it)

To Remove a Table

• Go to Knowledge Manager → Tables
• Find the table you want to remove
• Click the Delete icon or menu option
• Confirm deletion

Warning: Deleting a table removes it from all workflows that use it. AI will no longer have access to that data.

To Connect/Disconnect Tables from Workflows

To Connect:

• Go to your workflow → Knowledge AI → Tables tab
• Check the box next to the table
• Save changes

To Disconnect:

• Go to your workflow → Knowledge AI → Tables tab
• Uncheck the box next to the table
• Save changes

Data Processing and Column Handling

Automatic Column Name Standardization

When you upload a CSV, Table Vault automatically:

• Converts column names to lowercase
• Replaces spaces with underscores
• Removes special characters
• Creates database-friendly column names

Example Transformations:

• "Product Name" → "product_name"
• "Price ($)" → "price"
• "QTY Available" → "qty_available"

Data Type Detection

Table Vault intelligently determines the type of data in each column:

• Text/String: Names, descriptions, categories
• Numbers: Prices, quantities, IDs
• Dates: Timestamps, order dates, updated dates
• Boolean: Yes/No, True/False values

This automatic detection ensures queries work correctly and efficiently.

Best Practices

Preparing CSV Files

Clear Headers: Use descriptive column names that clearly indicate what data they contain

Consistent Formatting: Ensure all values in a column use the same format (e.g., all dates formatted the same way)

No Empty Columns: Remove columns that don't contain useful data

Complete Data: Fill in as much data as possible; empty cells may affect query results

Naming Tables

Be Descriptive: Use names that clearly indicate what data the table contains

Version Information: Include dates or versions if you maintain multiple versions ("pricing_2024_q1")

Category Prefixes: Group related tables with prefixes ("products_catalog", "products_inventory")

Avoid Generic Names: Instead of "data" or "table1", use "customer_accounts" or "order_history"

Organizing Tables

Separate Concerns: Create different tables for different types of data rather than one large table

Related Data: Keep related information in the same table (e.g., all product details together)

Update Regularly: Refresh tables when source data changes to keep AI responses accurate

Archive Old Data: Remove or replace outdated tables to prevent AI from using stale information

Using Tables in Workflows

Start Small: Begin with one or two essential tables

Test Thoroughly: Send test emails to verify AI can correctly query your tables

Monitor Usage: Review workflow results to see how table data is being used

Refine Data: Based on usage, adjust which columns and rows you include in tables

Common Use Cases

Product Catalog

Data to Include: Product names, descriptions, prices, SKUs, availability, specifications

AI Capabilities: Answer product questions, provide pricing, check availability, recommend similar items

Customer Information

Data to Include: Account numbers, contact details, service tiers, contract dates, preferences

AI Capabilities: Look up account details, verify customer information, check service levels

Pricing Lists

Data to Include: Item names, base prices, volume discounts, promotional pricing, valid dates

AI Capabilities: Quote accurate prices, explain pricing tiers, identify applicable discounts

Inventory Management

Data to Include: SKUs, quantities, warehouse locations, reorder points, suppliers

AI Capabilities: Check stock levels, identify low inventory, provide availability estimates

Service Catalogs

Data to Include: Service names, descriptions, pricing, duration, requirements, availability

AI Capabilities: Describe services, quote service packages, check scheduling availability

Privacy and Data Security

Your table data is protected through:

Workspace Isolation: Tables are only accessible within your workspace

Secure Storage: Data is stored securely in encrypted databases

Access Control: Only workflows you explicitly configure can access table data

No External Sharing: Table data is never shared outside your workspace

Audit Trail: All table queries are logged for review

What You'll See in Action

When Table Vault Knowledge is active:

Precise Responses: AI provides exact data from your tables rather than general information

Structured Answers: Responses include formatted lists, specifications, and data points

Multiple Results: When relevant, AI returns multiple matching records

Context-Aware: AI combines table data with other knowledge sources for comprehensive responses

Source Attribution: Clear indication when information comes from table data

Comparing Table Vault to Other Knowledge Types

Feature	Table Vault	Documents	Web Sources	Q&A
Data Type	Structured	Unstructured	Mixed	Curated
Query Style	Exact match	Similarity search	Similarity search	Exact match
Best For	Catalogs, pricing, lists	Policies, guides	Current info, external content	Common questions
Update Method	Upload new CSV	Upload new files	Auto-refresh	Manual edit
Response Precision	Very High	Medium	Medium	Very High

Troubleshooting Common Issues

Table Upload Fails

• Verify CSV file is properly formatted with headers in the first row
• Check file encoding (should be UTF-8)
• Ensure no corrupted or binary data in the file
• Try with a smaller file to isolate the issue

AI Not Using Table Data

• Verify table is connected to the workflow (Knowledge AI → Tables)
• Check that email questions clearly relate to table content
• Ensure table data is relevant to the types of questions asked
• Review table column names for clarity

Incorrect Query Results

• Review CSV data for consistency and completeness
• Check that column names accurately describe their content
• Verify data types are appropriate for the information
• Update table with corrected data and test again

This feature integrates seamlessly with your other Knowledge AI capabilities including documents, web sources, Q&A, and MCP integrations to provide comprehensive, accurate information for your automated email responses.