360 Group Launches World's First L4 Multi-Agent Swarm AI

On August 2nd, the 360 Group announced the rebranding and upgrade of its Nano AI to Multi-Agent Swarm, marking it as the world's first system to achieve Level 4 (L4) autonomy in agent systems. This breakthrough transitions AI productivity from 'individual combat' to 'group collaboration,' heralding a new era of outcome-driven AI.

The Four Stages of Agent Evolution

Before the Multi-Agent Swarm, agents evolved through three stages:

1. L1 Chat Assistants: Basic chat tools like GPTs, offering suggestions or companionship but limited to 'toy-level' functionality.
2. L2 Low-Code Workflow Agents: Semi-finished tools built with low-code models, transitioning from toys to tools.
3. L3 Autonomous Planning Agents: Expert agents (e.g., Nano AI's 50,000+ vertical agents) capable of deep reasoning and tool use but limited in cross-domain tasks.

L4 Multi-Agent Swarm: A Game-Changer

The L4 Multi-Agent Swarm solves the bottleneck of individual agents by enabling team collaboration. Key innovations include:

• Swarm Collaboration Framework: Allows multiple agents to form flexible groups, nest in layers, and collaborate on complex tasks.
• Unified Command System: Agents operate under a shared goal, optimizing roles for tasks like producing a 10-minute cinematic masterpiece.

Technical Breakthroughs

Nano AI's Swarm Engine supports:

• Unlimited tool calls and context length.
• Execution of super-long tasks (e.g., 1,000+ steps over 2 hours).
• Industry-leading success rates, avoiding the 'too many cooks' dilemma.

Real-World Applications

Nano AI has launched 10+ multi-agent swarms for scenarios like:

• Video Production: 'One-sentence blockbuster' generation (20 minutes vs. 2 hours manually).
• E-Commerce: Multilingual product promo videos with a single click.
• Content Creation: AI-powered MCN teams for自媒体 businesses.