Agents

Agents are AI-powered workers that analyze information and take actions based on what they find. Unlike traditional AI chatbots that only respond to questions, Cotera agents can actively process data, make decisions, and execute tasks - whether working conversationally or automatically processing your business data.

What Are Agents?

Think of an agent as an AI employee that can read through information, understand what it means, and then do something useful with that knowledge. Agents combine the analytical power of AI with the ability to take real actions in your business systems.

Cotera offers two types of agents, each suited to different work patterns:

Chat Agents

Conversational agents you interact with directly through dialogue. Chat agents are perfect for:

• Ad-hoc research and analysis
• Exploring questions dynamically
• Investigating data on demand
• Testing approaches before automation
• Tasks requiring human guidance

Chat agents can access your data, use tools, and help you explore information interactively, combining the flexibility of conversation with the power to take concrete actions.

Dataset Agents

Automated agents deployed as columns within datasets that process every row of data according to your business logic. Dataset agents are ideal for:

• Systematic data processing at scale
• Automated analysis and categorization
• Scheduled workflows and monitoring
• Structured output generation
• Repeatable business processes

Dataset agents transform raw business information into structured intelligence, making AI insights part of your standard data infrastructure.

Common Agent Capabilities

Both chat and dataset agents share these fundamental capabilities:

Analyze and Understand

Agents might:

• Read customer reviews and identify which ones mention safety concerns
• Analyze support conversations to determine customer sentiment
• Process order data and flag high-value customers at risk of churning
• Review financial transactions and categorize them by expense type

Take Actions

But agents don't just analyze - they can act on their findings by sending alerts, updating systems, or triggering workflows.

How Agents Work

Connected to Your Data

Agents can access data you've made available through Cotera's dataset connections, giving them full context about your business. Chat agents access this data on demand during conversations, while dataset agents process it systematically.

Powered by Leading AI Models

Agents use state-of-the-art AI models from providers like OpenAI, Anthropic, and Google Gemini. You choose which model works best for each specific task.

Equipped with Tools

Agents can be connected to a vast library of tools that extend their capabilities:

• Data fetching tools: Google Search, database lookups, API calls, and more
• Action tools: Send Slack messages, update Klaviyo properties, append to Google Sheets, plus hundreds of other integrations
• Analysis tools: Image processing, document parsing, data validation, and additional specialized functions

Cotera integrates with hundreds of different tools and can add custom tool connections on demand to meet your specific business needs.

Flexible Input Processing

Agents can work with multiple types of input:

• Text data from your warehouse
• Images and documents
• Structured data like JSON or database records
• Real-time data from connected systems

Agent Capabilities

Autonomous Decision Making

Agents can make complex decisions based on information they process. You can give them as much or as little control as needed - from simple binary classifications to complex multi-step workflows.

Scheduled Execution (Dataset Agents)

Set dataset agents to run automatically on schedules that match your business needs - hourly, daily, weekly, or triggered by data changes.

Structured Outputs

Agents return results in formats that integrate seamlessly with your systems - strings, numbers, booleans, JSON objects, arrays, or any data warehouse-compatible type.

Chained Operations

Multiple agents can work together in sequences, where one agent's output becomes another agent's input, creating sophisticated automated workflows.

Agents vs Traditional Automation

Traditional business automation requires rigid, pre-programmed rules. Agents bring intelligence to automation:

Traditional Rule: "If customer rating is below 3, flag for review" Agent Approach: "Read customer feedback and identify any mentions of safety, quality, or service issues, regardless of rating, and assess urgency level"

This intelligence allows agents to handle nuanced, context-dependent tasks that would be impossible with standard automation.

Common Use Cases

Customer Intelligence

• Analyze support conversations for sentiment and intent
• Identify at-risk customers from behavioral signals
• Extract product feedback themes from reviews

Content Processing

• Categorize incoming support tickets by complexity and urgency
• Extract key information from contracts or documents
• Moderate user-generated content for compliance

Business Operations

• Flag unusual financial transactions for review
• Route leads to appropriate sales representatives
• Monitor social media mentions for brand reputation

Data Enrichment

• Add missing information to customer profiles
• Standardize and clean imported data
• Generate summaries of complex datasets

Research and Analysis

• Investigate questions across multiple data sources
• Synthesize information from documents and databases
• Explore trends and patterns interactively

Choosing Between Chat and Dataset Agents

Use Chat Agents when you need:

• Interactive exploration and research
• Flexible, guided analysis
• Ad-hoc investigation of questions
• Human oversight at each step
• Dynamic decision-making during conversation

Use Dataset Agents when you need:

• Automated, systematic processing
• Consistent application of business logic at scale
• Integration with existing data infrastructure
• Scheduled, repeatable workflows
• Results that feed into reporting and analysis

Many workflows benefit from both - using chat agents to explore and refine approaches, then deploying dataset agents to automate proven processes at scale.

Getting Started with Agents

Chat agents are available immediately for interactive work, while dataset agents live in columns within your datasets, making their outputs immediately available alongside your existing data.

Whether you need exploratory research, simple data classification, or complex multi-step business processes, agents provide the intelligence layer that transforms information into automated business value.