Parallel Block

The Parallel block runs multiple branches of your workflow at the same time, then waits for all branches to complete before continuing. Use it to speed up workflows by running independent tasks concurrently.

Overview

When to Use

• Run independent API calls simultaneously
• Query multiple data sources at once
• Generate multiple AI outputs in parallel (e.g., summaries in different styles)
• Speed up workflows where tasks don't depend on each other

Configuration

The Parallel block is a container. Place blocks inside each branch. Each branch runs independently and the parallel block waits for all branches to finish.

Outputs

Example: Multi-Source Research

Goal: Research a topic across Wikipedia, ArXiv, and web search simultaneously.

Workflow:

                        ┌→ [Google Search] ─→ [Agent: Web Summary]   ─┐
[Starter] → [Parallel]  → [Wikipedia]     ─→ [Agent: Wiki Summary]    → [Agent: Combine] → [Response]
                        └→ [ArXiv]        ─→ [Agent: Paper Summary] ─┘

How it works:

1. User provides a research topic
2. Parallel block launches 3 branches simultaneously:
   • Branch 1: Google Search → summarize web results
   • Branch 2: Wikipedia → summarize encyclopedia content
   • Branch 3: ArXiv → summarize academic papers
3. All branches complete concurrently (3x faster than sequential)
4. Final Agent combines all summaries into a comprehensive report

Tips

• Only parallelize independent tasks — if branch B needs branch A's output, they can't run in parallel
• Great for multi-provider AI — run the same prompt on GPT-4o, Claude, and Gemini simultaneously to compare
• Watch rate limits — parallel API calls may trigger rate limiting on external services