Knowledge table

Written By Stanislas

Last updated 2 days ago

Overview

Knowledge Tables let you create simple, organized tables for any type of business data and connect them directly to your AI agents. Store your leads, product catalogs, FAQs, or customer information in structured tables that your agents can query, read, and update during conversations.

Unlike unstructured documents in your knowledge base, Knowledge Tables provide structured, database-like storage with defined columns and data types. This gives your agents precise, reliable access to business-critical data without the complexity of external databases.

Knowledge Tables are available on all plans (Pro, Team Starter, Team Growth, and Custom) with no technical limits on the number of tables you can create.

Prerequisites

Before connecting your first Knowledge Table to an agent, you need:

  • A Swiftask account with access to the Knowledge section

  • Basic understanding of your data structure (what columns and data types you need)

  • Your data ready to import (optional, but recommended for bulk operations)

Connecting a Knowledge Table to an agent

This is the primary way to give your agents access to structured data. You can either connect an existing table or create a new one directly from the agent's skill configuration.

The agent can then interact with it: reading the table, adding new columns or rows; based on the access you grant it.

Adding the Knowledge Table skill to an agent

Step 1: Open your agent settings and navigate to the Skills section.

Step 2: Search for or scroll to the Swiftask - Knowledge Table (Beta) tool and click to add it.

Step 3: A configuration panel opens. Fill in the required fields:

  • Name: Display name for this skill in your agent (e.g., "Lead Database", "Product Catalog")

  • Description: What this skill should be used for (e.g., "Look up and update lead information")

  • Knowledge table slug: Select your table from the dropdown menu

  • Access mode: Choose the appropriate permission level (see "Understanding access modes" below)

Step 4: Click Save to activate the skill. Your agent can now access the Knowledge Table.

Understanding access modes

The Access mode determines what permissions your agent has on the Knowledge Table. Choose the mode that matches your use case:

Read Only

  • Agent can query and view data

  • Agent cannot add, edit, or delete rows

  • Agent cannot change the table structure (add/remove columns)

  • Use case: FAQs, product catalogs, reference data that doesn't change during conversations

Read And Write Rows

  • Agent can query, view, add, and edit rows

  • Agent can delete rows

  • Agent cannot change the table structure (add/remove columns)

  • Use case: Customer records, lead qualification, status updates where agents need to modify data but not schema

Read/Write Rows And Columns (Schema)

  • Agent has full control: query, edit rows, add rows, delete rows, add new columns, and delete columns

  • Agent cannot rename existing columns

  • Use case: Advanced automation, dynamic table management where agents need to adapt the structure itself as your business evolves

Creating a Knowledge Table directly from the agent skill

If you don't have a Knowledge Table yet, you can create one directly from the agent's skill configuration without leaving the page.

Step 1: While configuring the Swiftask - Knowledge Table skill, click the + Create a new table button below the Knowledge table slug dropdown.

Step 2: A modal opens. Enter a table name (e.g., "Orders", "Leads", "Customers") and optionally add a description.

Step 3: Click Create. The new table is created and automatically selected in the Knowledge table slug field.

Step 4: You can now configure the Access mode and save the skill. The table is ready for columns and data, and you can ask the agent to create columns and rows for you.

Creating a Knowledge Table

You can create Knowledge Tables from the Knowledge section in your workspace. Tables are empty when first created; you define the structure by adding columns, then populate them with data.

Creating a new table

Step 1: In the left sidebar, click Knowledge, then select Knowledge Tables.

Step 2: On the Knowledge Tables overview page, click the + New table button in the top right.

Step 3: A dialog opens. Enter a table name (e.g., "Employees", "Products", "FAQ Database") and optionally add a description, then click Create.

Step 4: Your new table is created and ready for structure. You'll see an empty table with system columns (ID, Created at, Updated at) and a search bar at the top.

Adding columns to your table

Your Knowledge Table starts with three system columns that are always present and cannot be modified: ID, Created at, and Updated at. You add custom columns to store your business data.

Step 5: Click + button (red circle) in the column header area.

Step 6: Configure each column:

  • Column name: Use clear, descriptive names (e.g., "LeadName", "Email", "Status", "Price")

  • Data type: Choose from:

  • TEXT: Names, descriptions, free text

  • NUMBER: Quantities, IDs, numeric values

  • BOOLEAN: Yes/no, true/false values

  • DATE: Dates without time

  • DATETIME: Dates with time

  • JSON: Complex structured data

  • SELECT_OPTIONS: Dropdown/choice fields

  • REFERENCE: Links to other tables

Click Save to add the column.

Step 7: Repeat this process for each column you need. Your table structure is now defined.

Adding data to your table

Manual entry

Step 8: Click + Add row at the bottom left to add a new row manually. An inline editor opens where you can fill in the data for each column. Click Save when done.

CSV import (recommended for bulk data)

Click the Import button in the top right, select a CSV file from your computer, and all rows will be added automatically. Make sure your CSV headers match your column names exactly (case-sensitive).

Exporting data

Click the Export CSV button in the top right to download your table as a CSV file. This is useful for backups, analysis, or sharing with other systems.

Creating an agent from a Knowledge Table

Once your Knowledge Table is created and populated, you can quickly create a new agent pre-configured to use that table.

Step 1: From your Knowledge Table, click the Create agent button in the top right.

Step 2: A modal opens with:

  • Agent name: Enter a name for your new agent (e.g., "Lead Qualifier", "Support Bot")

  • Template selection: Choose from available agent templates (Blank template, Knowledge assistant, or specialized templates)

Step 3: Click Create Agent. You'll be redirected to the agent configuration page, and the Swiftask - Knowledge Table skill is automatically added with this table selected.

Step 4: Configure the remaining agent settings and the skill's Access mode as needed, then save.

Practical use cases

Knowledge Tables unlock powerful agent capabilities. Here are diverse, real-world scenarios:

Product support agent with pricing lookup

Create a "Products" table with columns like ProductID, Name, Category, Price, and StockStatus. When a customer asks "What's the price of your premium plan?", the agent queries the table and provides the exact, up-to-date price. Use Read Only access to prevent accidental modifications.

Lead qualification chatbot

Build a "Leads" table with LeadName, Email, Company, IndustryType, and QualificationScore columns. As your sales agent qualifies leads through conversation, it updates the QualificationScore directly in the table. Use Read And Write Rows access so the agent can add new leads and update scores without restructuring the table.

Multi-language FAQ agent

Create an "FAQs" table with QuestionEN, AnswerEN, QuestionFR, AnswerFR, Category, and LastUpdated columns. When users ask questions in French or English, the agent retrieves the appropriate answer from the table. Use Read Only access to ensure consistency.

Appointment booking assistant

Set up an "AvailableSlots" table with Date, Time, ServiceType, AgentID, and IsBooked columns. When customers request appointments, the agent checks available slots, books one by updating IsBooked to true, and confirms the appointment. Use Read And Write Rows to keep your calendar synchronized automatically.

Content moderation agent

Create a "UserSubmissions" table with SubmissionID, Content, SubmittedBy, CreatedDate, and ModerationStatus columns. As the agent reviews submissions, it updates the ModerationStatus (Approved, Rejected, NeedsReview). Use Read And Write Rows to keep the moderation workflow current.

Vendor management system with dynamic schema

Create a "Vendors" table with columns like VendorName, ContactEmail, PaymentTerms, and PerformanceScore. Use Read/Write Rows And Columns (Schema) access so the agent can not only update vendor information but also add new columns when new vendor attributes become important (e.g., CertificationLevel, EnvironmentalRating). This allows the system to grow with your business needs.

Employee directory and HR queries

Create an "Employees" table with columns like Name, Role, Department, HireDate, Manager, and IsActive. When an employee asks "Who is my manager?", the agent looks up the information instantly. Use Read Only access to protect sensitive HR data from accidental modifications.

Tips & best practices

Use clear, descriptive column names: Names like "EmployeeID", "HireDate", "IsActive" are clearer than "ID", "Date", "Status". This helps agents understand and process data correctly.

Choose the right data type: Using DATE for hire dates and BOOLEAN for active status helps agents interpret and validate data properly.

Keep tables focused: Create one table per domain (employees, products, orders). Avoid mixing unrelated data in the same table.

Start with Read Only: If you're unsure about permissions, begin with read-only access and upgrade to read/write later if needed.

Test with your agent: After adding the Knowledge Table skill, test the agent's ability to query and interact with the data before deploying to production.

Use CSV import for bulk data: If you have large datasets, use CSV import instead of adding rows manually. It's faster and less error-prone.

Use CSV export for backups: Regularly export your Knowledge Table as CSV to create backups or migrate data to other systems.

Troubleshooting

Agent can't find the Knowledge Table

  • Confirm the Knowledge table slug in the tool configuration matches the table name exactly

  • Ensure the Knowledge Table is created and has at least one row of data

Agent can't update rows even though I set Read And Write Rows

  • Verify the access mode is set correctly in the skill configuration

  • Check that the agent has the skill properly saved

CSV import fails

  • Verify your CSV file has headers that match your table column names exactly (case-sensitive)

  • Check that data types in your CSV match the column types (e.g., dates in DATE columns must be in a valid date format)

Additional resources