Agent Components

Neurobro is present as a single agent, but from the technical perspective he is more complex. Neurobro includes multiple components that work together to provide both intelligence and functionality. Here is the overview and architecture note:

Each platform-specific Neurobro instance maintains its own:

  • Functionality scope
  • Data sources
  • Environmental interaction points

Knowledge Base

Embeddings

All embeddings are generated using OpenAI’s state-of-the-art text-embedding-3-large model, ensuring:

Precision

High-accuracy matching

Relevance

Context-aware results

Performance

Optimized processing

API Integration

For the most part, Neurobro uses proprietary APIs to get the most accurate and up-to-date information.

Proprietary APIs

~90% of visible value

Public APIs

~10% of functionality

Examples: Coingecko, DexScreener, Base BlockScout

Large Language Models

LLMs form the backbone of the AI, powering Neurobro with advanced intelligence capabilities.

Different LLMs are used for different purposes:

R1

Specialized long-running reasoning tasks (e.g. internal evaluation of the found alpha)

V3

Fast direct communication with users

Tools & Workflows

Each Nevron uses specialized tools and workflows to deliver optimal performance:

1

Tool Identification

Agent analyzes context and determines required functionality

2

Tool Execution

Selected tools represented by Nevrons process data and perform specific operations

3

Response Generation

Results are synthesized into coherent, contextually appropriate output

4

Feedback Integration

System learns from interaction outcomes to improve future performance

Neurobro itself uses a set of nevrons to perform complex tasks: chatting, analysis, research, etc.

Platform-Specific Examples

While Neurobro remains a single agent, his presece on different platforms significantly varies. Here are some examples of how Neurobro works on different platforms:

Want to learn more about our tool development? Check out the Nevron documentation.