SWAI Synapse

Real-time event processing system

Think of how your eyes and ears help you notice things around you. That’s what Synapse does for SWAI agents - it helps them see and hear what’s happening.

Synapse is like the agent’s nervous system, keeping it alert and ready to respond to whatever happens nearby.

Synapse receives input from the external environment — including user actions, on-chain data, and off-chain signals — and routes it to Cortex for analysis and decision-making.

This process is core to real-time responsiveness within CTZN.

The cool thing about Synapse is that it uses a simple, consistent way to handle information. This means it can connect to all kinds of different systems as long as they speak the same language.

Synapse connects directly with Cortex (the decision-making module) to process information and determine appropriate responses in real-time.

Signal processing

1. Signal detection

   Continuously monitors and captures incoming signals from various sources

2. Priority assessment

   Evaluates signal importance based on setup instructions and current agent state

3. Queue management

   Organizes signals into priority queues for efficient processing and to prevent system overload during high-volume periods

4. Signal transmission

   Forwards prioritized signals to Cortex for decision-making and reasoning

Signal sources

Synapse’s flexible architecture allows for endless integration possibilities.

From social platforms to blockchain networks, any system that generates events can be connected to your SWAI agent.

Social interactions

Monitors engagement across platforms:

• Replies and mentions
• Likes and reactions
• Shares and retweets
• Direct messages

On-chain activity

Tracks blockchain events:

• Token transfers
• NFT sales and mints
• Smart contract interactions
• Governance proposals

External systems

Integrates with various platforms:

• E-commerce transactions
• Payment systems
• Content management
• Analytics platforms

Signal format

All events in Synapse are standardized into signals with this structure:

interface Signal {
    timestamp: number;    // When the event occurred
    description: string;  // Human-readable event description
    priority: number;     // Importance level (1-5)
    labels: string[];     // Event categories
    metadata: object;     // Additional context
}

All signals must follow this format to ensure consistent processing across the system.

Queue Management

Events are processed according to their priority level:

High Priority

Direct questions and critical alerts receive immediate attention

Medium Priority

Standard interactions are processed in order of arrival

Low Priority

Routine updates are handled when resources are available

This prioritization maintains consistency in agent behavior and ensures a coherent experience throughout interactions.