Parallel Agents

Parallel agents unleash the power of concurrent processing, running multiple agents simultaneously to tackle independent tasks. Think of it as your AI team working in parallel - while one agent analyzes sentiment, another extracts topics, and a third generates summaries, all at the same time.

Unlike sequential agents that process step-by-step, parallel agents maximize speed and efficiency by utilizing multiple processing streams. Perfect when you have independent tasks that don't need to wait for each other.

Quick Start Example

Here's a comprehensive parallel analysis agent that evaluates user feedback from multiple angles simultaneously:

import { LlmAgent, ParallelAgent } from "@iqai/adk";

// Parallel analysis agents for user feedback - each analyzes different aspects simultaneously
const sentimentAgent = new LlmAgent({
  name: "sentiment_agent",
  description:
    "Analyzes the overall sentiment and emotional tone of user feedback",
  instruction: `Analyze the sentiment of the user feedback. Classify as positive, negative, or neutral, and provide a confidence score. Identify specific emotional indicators.`,
  outputKey: "sentiment_analysis", // Key for storing sentiment results
});

const topicAgent = new LlmAgent({
  name: "topic_agent",
  description: "Categorizes feedback into relevant topics and themes",
  instruction: `Identify the main topics and themes in the user feedback. Categorize into areas like product features, customer service, pricing, usability, etc. List specific topics mentioned.`,
  outputKey: "topic_categories", // Key for storing topic results
});

const issueAgent = new LlmAgent({
  name: "issue_agent",
  description:
    "Identifies specific problems, bugs, or concerns mentioned in feedback",
  instruction: `Extract and categorize any specific issues, problems, or concerns mentioned in the feedback. Prioritize critical issues and suggest severity levels.`,
  outputKey: "identified_issues", // Key for storing issue results
});

const suggestionAgent = new LlmAgent({
  name: "suggestion_agent",
  description: "Extracts user suggestions and improvement ideas",
  instruction: `Identify any suggestions, recommendations, or improvement ideas mentioned in the feedback. Organize them by category and feasibility.`,
  outputKey: "user_suggestions", // Key for storing suggestion results
});

const satisfactionAgent = new LlmAgent({
  name: "satisfaction_agent",
  description: "Rates overall user satisfaction and likelihood to recommend",
  instruction: `Rate the user's overall satisfaction on a scale of 1-10. Assess likelihood to recommend and identify key satisfaction drivers or pain points.`,
  outputKey: "satisfaction_rating", // Key for storing satisfaction results
});

// Parallel analysis system - demonstrates concurrent execution pattern
const feedbackAgent = new ParallelAgent({
  name: "feedback_agent",
  description:
    "Comprehensive parallel analysis of user feedback across multiple dimensions",
  subAgents: [
    sentimentAgent,
    topicAgent,
    issueAgent,
    suggestionAgent,
    satisfactionAgent,
  ],
});

// Export the parallel agent for use
export { feedbackAgent };

Visual Flow

How Parallel Processing Works

Parallel agents maximize efficiency by running all sub-agents simultaneously:

⚡ Concurrent Execution

1. All agents start together - No waiting for previous agents to finish
2. Independent processing - Each agent works on the same input independently
3. Simultaneous completion - Results become available as each agent finishes
4. Non-deterministic order - Faster agents finish first, slower ones catch up

📊 Performance Benefits

• Time savings: N agents running for T seconds = T total time (not N×T)
• Resource utilization: Maximizes CPU/GPU usage across multiple agents
• Scalability: Add more agents without increasing total processing time

🔄 Result Aggregation Each agent writes to unique keys in shared session state. Later agents (or your application) can access all results together for final processing.

Real-World Use Cases

📝 Multi-Perspective Content Analysis
Sentiment + Topics + Keywords + Priority Assessment (all simultaneously)

🔍 Comprehensive Research
Web Search + Academic Papers + News + Social Media (parallel data gathering)

📊 Financial Analysis
Technical Analysis + Fundamental Analysis + News Sentiment + Risk Assessment

🛠️ Quality Assurance
Grammar Check + Fact Verification + Style Review + Compliance Check

🌍 Multi-Language Processing
Translation + Sentiment + Cultural Context + Localization (per language)

📊 Data Validation
Format Check + Business Rules + Data Quality + Duplicate Detection

When to Choose Parallel Agents

✅ Choose Parallel When:

• Independent analysis - Multiple perspectives on the same data
• Speed is critical - Need results as fast as possible
• Resource optimization - Want to maximize processing power
• Multi-source data - Gathering from different sources simultaneously
• Comparative analysis - Need multiple approaches to the same problem

❌ Don't Use Parallel When:

• Tasks depend on each other's intermediate results
• You need guaranteed execution order
• Sequential processing is part of the business logic
• Resource constraints limit concurrent execution
• Results from one agent affect how another should process

Related Topics