cotera

Getting started with columns

Columns are where you deploy intelligence and automation within your datasets. This guide helps you understand how to work with columns effectively and points you toward the detailed documentation for each column type.

Understanding Columns

Columns represent different types of work that can be performed on your data. They're how you transform raw business information into intelligent, automated operations. Each column processes every row in your dataset, adding its results as a new field alongside your existing data.

Before diving into column creation, make sure you understand the foundational concepts:

  • Datasets: The workspace where columns operate
  • Agents: The AI intelligence deployed through columns
  • Tools: External system connections available to columns

Column Types Overview

Cotera supports several types of columns, each designed for different kinds of work:

LLM Columns

Deploy AI agents to analyze and interpret your data. These columns can read text, process information, make decisions, and generate insights based on the business logic you define.

Best for: Sentiment analysis, content categorization, data extraction, decision-making

Learn more: Create an LLM column • LLM column tutorial

Expression Columns

Perform calculations and data transformations using your existing data. Similar to spreadsheet formulas, these handle mathematical operations, data cleaning, and logical operations without requiring AI.

Best for: Calculations, data formatting, conditional logic, aggregations

Learn more: Expression columns guide

Tool Columns

Connect directly to external tools and services to send data or fetch additional information. These columns enable integrations with CRMs, notification systems, APIs, and hundreds of other services.

Best for: System updates, notifications, data enrichment, external actions

Learn more: Create a tool column • Tool usage guide

Working with Columns

Creating Columns

Navigate to your dataset and access the column creation interface through the column editor. Choose the column type that matches your goal and configure its logic according to your business needs.

Testing and Refining

Use the preview feature to test your column configuration on sample data before deploying it across your entire dataset. This allows you to refine your logic without affecting production data.

Publishing Changes

Understand the difference between saving and publishing changes. Saving preserves your work without affecting live data, while publishing deploys your column to process your dataset.

Running Executions

Once configured, run executions to process your data according to the schedule and triggers you've defined. Columns automatically apply their logic to new and updated data.

Column Management

Organization

Keep your columns organized with descriptive names that clearly indicate their purpose. Rename columns as your workflows evolve to maintain clarity across your team.

Maintenance

Columns can be modified, tested, and refined over time. Make incremental improvements based on results without disrupting your underlying data or other columns.

Removal

When columns are no longer needed, delete them to keep your datasets clean and focused on active workflows.

Next Steps

Ready to build your first column? Start with these practical tutorials:

  1. Create your first LLM column: Step-by-step guide to building an intelligent column
  2. Create an LLM agent with tools: Learn how to extend agent capabilities with external tools
  3. Expression columns: Master data transformations and calculations

For comprehensive technical details about specific column types, explore the guides linked above or dive into the columns core concept to understand how columns fit into Cotera's broader architecture.