Skip to main content
This example demonstrates how to build a document review system where each agent processes the output of the previous one - perfect for multi-stage processing workflows.

​
Step 1: Setup

import requests
import os

API_BASE_URL = "https://api.swarms.world"
API_KEY = os.environ.get("SWARMS_API_KEY", "your_api_key_here")

headers = {
    "x-api-key": API_KEY,
    "Content-Type": "application/json"
}

​
Step 2: Define Your Sequential Pipeline

Create a 4-stage document review pipeline where each agent builds on the previous agent’s work: 
def review_legal_document(document_text: str) -> dict:
    """Process a legal document through sequential review stages."""

    swarm_config = {
        "name": "Legal Document Review Pipeline",
        "description": "Sequential legal document review",
        "swarm_type": "SequentialWorkflow",
        "task": f"Review this legal document:\n\n{document_text}",
        "agents": [
            {
                "agent_name": "Document Parser",
                "description": "Extracts key information from legal documents",
                "system_prompt": "You are a legal document parser. Extract: document type, parties involved, key dates, financial terms, and obligations for each party.",
                "model_name": "gpt-4o",
                "max_loops": 1,
                "temperature": 0.2
            },
            {
                "agent_name": "Risk Analyst",
                "description": "Identifies potential legal risks",
                "system_prompt": "You are a legal risk analyst. Based on the parsed document, identify HIGH, MEDIUM, and LOW risk items. Explain each risk and suggest mitigations.",
                "model_name": "gpt-4o",
                "max_loops": 1,
                "temperature": 0.3
            },
            {
                "agent_name": "Compliance Checker",
                "description": "Verifies regulatory compliance",
                "system_prompt": "You are a compliance specialist. Review the document and risk analysis to check regulatory compliance (GDPR, industry regulations). Provide a compliance score (1-10).",
                "model_name": "gpt-4o",
                "max_loops": 1,
                "temperature": 0.2
            },
            {
                "agent_name": "Summary Generator",
                "description": "Creates executive summary",
                "system_prompt": "You are an executive assistant. Create a 2-minute read summary with: document overview, key terms, top 3 risks, compliance status, and recommendation (Sign/Negotiate/Reject).",
                "model_name": "gpt-4o",
                "max_loops": 1,
                "temperature": 0.4
            }
        ],
        "max_loops": 1
    }

    response = requests.post(
        f"{API_BASE_URL}/v1/swarm/completions",
        headers=headers,
        json=swarm_config,
        timeout=120
    )

    return response.json()

​
Step 3: Run the Pipeline

# Sample contract
contract = """
SERVICE AGREEMENT

This Agreement is entered into as of January 15, 2025 by and between:
Provider: TechCorp Solutions Inc., a Delaware corporation
Client: Acme Industries LLC, a California LLC

1. SERVICES: Provider agrees to deliver cloud infrastructure management services.

2. TERM: 36 months, auto-renewing for 12-month periods unless terminated with 30 days notice.

3. FEES: $15,000/month, due within 30 days. Late payments incur 2% monthly interest.

4. LIMITATION OF LIABILITY: Provider's total liability shall not exceed fees paid in preceding 3 months.

5. TERMINATION: Provider may terminate immediately for breach. Client may terminate with 90 days notice and payment of remaining term fees.
"""

# Run review
result = review_legal_document(contract)

# Display results
for output in result.get("output", []):
    print(f"\n--- {output['role']} ---")
    print(output['content'][:500] + "...")

print(f"\nTotal cost: ${result['usage']['billing_info']['total_cost']:.4f}")
Expected Output: 
--- Document Parser ---
Document Type: Service Agreement
Parties: TechCorp Solutions Inc. (Provider), Acme Industries LLC (Client)
Key Dates: Effective Jan 15, 2025, 36-month term...

--- Risk Analyst ---
HIGH RISK: Asymmetric termination rights - Client must pay remaining fees to exit...

--- Compliance Checker ---
Compliance Score: 7/10
Missing: Data processing terms, GDPR provisions...

--- Summary Generator ---
RECOMMENDATION: NEGOTIATE - Address termination clause before signing...

Total cost: $0.0891
Each agent in SequentialWorkflow receives the previous agent’s output. Design your prompts so each stage builds on prior work.