Round Robin Example

export SWARMS_API_KEY="your-api-key-here"

import requests
import json
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"
}

def run_roundtable(topic: str, max_loops: int = 1) -> dict:
    """Run a collaborative round-robin research discussion."""

    swarm_config = {
        "name": "Research Roundtable",
        "description": "Collaborative research with round-robin agent turns",
        "swarm_type": "RoundRobin",
        "task": topic,
        "agents": [
            {
                "agent_name": "Industry Researcher",
                "description": "Gathers market data and industry trends",
                "system_prompt": "You are an industry researcher. Provide data-driven market analysis, cite specific numbers and trends, and identify key players. Build on insights from other team members when available.",
                "model_name": "gpt-4.1",
                "max_loops": 1,
                "temperature": 0.4
            },
            {
                "agent_name": "Technology Analyst",
                "description": "Evaluates technical landscape and innovation",
                "system_prompt": "You are a technology analyst. Assess the technical landscape, evaluate emerging technologies, and identify innovation opportunities. Reference and build upon the research data shared by other team members.",
                "model_name": "gpt-4.1",
                "max_loops": 1,
                "temperature": 0.4
            },
            {
                "agent_name": "Strategy Advisor",
                "description": "Synthesizes insights into actionable strategy",
                "system_prompt": "You are a strategy advisor. Synthesize insights from the team into actionable strategic recommendations. Identify risks, opportunities, and provide a prioritized roadmap. Reference specific points made by other team members.",
                "model_name": "gpt-4.1",
                "max_loops": 1,
                "temperature": 0.5
            }
        ],
        "max_loops": max_loops
    }

    response = requests.post(
        f"{API_BASE_URL}/v1/swarm/completions",
        headers=headers,
        json=swarm_config,
        timeout=180
    )

    return response.json()

# Define the research topic
topic = """
Analyze the emerging autonomous AI agent market. Cover the current state of
the technology, major players and their approaches, enterprise adoption
barriers, and the most promising near-term use cases. Provide actionable
insights for a startup considering entering this space.
"""

# Run the roundtable discussion
result = run_roundtable(topic)

# Display the collaborative discussion
for output in result.get("output", []):
    agent = output["role"]
    content = output["content"]

    print(f"\n{'='*60}")
    print(f"{agent}")
    print(f"{'='*60}")

    # Handle content as string or list
    if isinstance(content, list):
        content = ' '.join(str(item) for item in content)

    print(str(content)[:800] + "...")

============================================================
Industry Researcher
============================================================
## Autonomous AI Agent Market Analysis

### Market Overview
The autonomous AI agent market reached an estimated $4.2B in 2024 and is
projected to grow at 45% CAGR through 2028. Key segments include:
- Developer tools & coding agents (35% of market)
- Customer service automation (28%)
- Enterprise workflow agents (22%)
- Research & analysis agents (15%)

### Major Players
- **OpenAI** (GPT-based agents, Assistants API)
- **Anthropic** (Claude, tool use framework)
- **Google** (Gemini agents, Vertex AI)
- **Startups**: Cognition (Devin), Adept, Induced AI, CrewAI, Swarms

### Enterprise Adoption
Current penetration: ~12% of Fortune 500 in production...

============================================================
Technology Analyst
============================================================
Building on the Industry Researcher's market data, let me assess the
technical landscape:

### Core Technology Stack
The agent frameworks broadly fall into three categories:
1. **Single-agent loops** (ReAct, function calling) — mature but limited
2. **Multi-agent orchestration** (Swarms, CrewAI, AutoGen) — growing fast
3. **Code-generation agents** (Devin, Cursor) — highest enterprise demand

### Key Technical Differentiators
Drawing from the market segments identified above:
- Tool use reliability (currently 85-92% accuracy)
- Context window management for long-running tasks
- Multi-step planning and self-correction capabilities
- Sandboxed execution environments for safety...

============================================================
Strategy Advisor
============================================================
Synthesizing the market data from our Industry Researcher and the technical
assessment from our Technology Analyst, here are my strategic recommendations:

### Entry Strategy (Priority Order)
1. **Target the multi-agent orchestration gap** — As noted, this segment is
   growing fastest at 45% CAGR, and tool use reliability (85-92%) leaves
   room for differentiation through better orchestration

2. **Focus on enterprise workflow agents** — The 22% market share with only
   12% Fortune 500 penetration signals massive headroom

3. **Build on open-source adoption** — CrewAI and Swarms have proven the
   community-first model works for developer tools

### Key Risks
- Commoditization risk as foundation model providers add native agent features
- Enterprise security and compliance requirements add 6-12 months to sales cycles...

# Run 2 loops — agents go around twice, refining their analysis each time
deep_result = run_roundtable(
    topic="Evaluate the competitive positioning of Anthropic vs OpenAI vs Google in the enterprise AI market. Assess technical capabilities, pricing strategy, ecosystem lock-in, and likely market share in 3 years.",
    max_loops=2
)

# Show the final contributions after 2 rounds of refinement
for output in deep_result.get("output", []):
    print(f"\n{output['role']}:")
    content = output["content"]
    if isinstance(content, list):
        content = ' '.join(str(item) for item in content)
    print(str(content)[:600] + "...")