> ## Documentation Index
> Fetch the complete documentation index at: https://whitepaper.neurobro.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Core Technologies

> Technical infrastructure powering the Neurobro ecosystem

<Frame>
  <img src="https://mintcdn.com/neurobro/l4gM_i_EkePyUTVO/images/agent-base.webp?fit=max&auto=format&n=l4gM_i_EkePyUTVO&q=85&s=c37e15d66d28baa4c23569e1eba8e2c8" style={{ borderRadius: '0.5rem' }} width="1536" height="1024" data-path="images/agent-base.webp" />
</Frame>

## Overview

Neurobro's intelligence layer is built on two foundations:

* **Data**: The highest quality financial data - real-time, multi-source, and proprietary - covering on-chain activity, market feeds, news, and social signals
* **Agents**: Autonomous AI components that process data, generate insights, and serve users across all platforms

This architecture enables a single intelligence to power every product in the ecosystem - from Neurodex to the mobile app to NeuroAPI - while maintaining consistency and quality.

## Agents & Nevron

Behind Neurobro's intelligence lies a network of specialized "Nevrons" - modular AI agents working in harmony. 200+ Nevrons work together to form a unified financial intelligence.

Each Nevron handles specific tasks - from market analysis to sentiment tracking to on-chain research - forming the building blocks of our ecosystem. Nevrons communicate with each other, enabling complex multi-step analysis, decision-making, and research workflows.

Some Nevrons also interface with external systems: executing trades, posting analysis, and interacting with users across platforms.

To enable this architecture, we built our own agent framework called Nevron.

The Nevron framework is open source and available on [GitHub](https://github.com/axioma-ai-labs/nevron), with comprehensive [technical documentation](https://axioma-ai-labs.github.io/nevron/) to help you get started.

<Warning>
  Since Neurobro is being constantly updated, the core technologies may differ from the ones described in this section.
</Warning>

### Why Nevron?

It's very important to understand that Nevron is not a general-purpose framework, but a framework specifically designed for building specialized AI agents. Here's why to use it:

<AccordionGroup>
  <Accordion icon="puzzle-piece" title="Adaptability">
    Easy customization for different tasks or workflows through:

    * Modular components
    * Configurable parameters
    * Task-specific optimization
  </Accordion>

  <Accordion icon="arrows-rotate" title="Flexibility">
    Quick reconfiguration capabilities:

    * Dynamic workflow adjustment
    * Real-time task modification
    * Seamless integration options
  </Accordion>

  <Accordion icon="bolt" title="Efficiency">
    <CardGroup cols={2}>
      <Card title="Resource Optimization" icon="microchip">
        Optimal computing utilization
      </Card>

      <Card title="Reasoning Power" icon="brain">
        Enhanced decision-making
      </Card>
    </CardGroup>
  </Accordion>

  <Accordion icon="shield-check" title="Robust Performance">
    Reliable fact-based outputs through:

    * Multi-source verification
    * Error handling
    * Performance monitoring
    * Quality assurance
  </Accordion>
</AccordionGroup>

### Shared Resources

Neurobro maintains a unified state across platforms through shared resources. Many different Nevrons work together on the same foundation of data and knowledge.

<CardGroup cols={3}>
  <Card title="Dynamic Communication" icon="comments">
    Real-time agent collaboration
  </Card>

  <Card title="Knowledge Sharing" icon="database">
    Centralized information pool
  </Card>

  <Card title="Cross-Platform Sync" icon="arrows-rotate">
    Consistent state management
  </Card>
</CardGroup>

### Agent Components

Neurobro appears as a single agent, but from a technical perspective, it is more complex. Neurobro includes multiple components that work together to provide both intelligence and functionality.

Here's an overview of the architecture:

<Tabs>
  <Tab title="Overview">
    Each platform-specific Neurobro instance maintains its own:

    * Functionality scope
    * Data sources
    * Environmental interaction points
  </Tab>

  <Tab title="Architecture">
    The central instance of Neurobro present across all platforms has the following components:

    <CardGroup cols={3}>
      <Card title="Knowledge Base" icon="database">
        Advanced RAG systems
      </Card>

      <Card title="Embeddings" icon="network-wired">
        SOTA models
      </Card>

      <Card title="APIs" icon="plug">
        90% proprietary & 10% public
      </Card>
    </CardGroup>
  </Tab>
</Tabs>

Since the main intelligence is focused on the main product-Neurodex-see there for more details.

<Info>
  Read more about the AI components in the [Nevron](/technical/neurodex/agents) section.
</Info>

### Large Language Models

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

Different LLMs are used for different purposes:

<Tabs>
  <Tab title="DeepSeek">
    <CardGroup cols={2}>
      <Card title="R1" icon="brain">
        Specialized long-running reasoning tasks (e.g. internal evaluation of the found alpha)
      </Card>

      <Card title="V3" icon="lightbulb">
        Fast direct communication with users
      </Card>
    </CardGroup>
  </Tab>

  <Tab title="Kimi">
    <Card title="Kimi k2" icon="brain">
      Direct communication with users
    </Card>
  </Tab>

  <Tab title="Meta">
    <Card title="Llama 4" icon="square-root-variable">
      Simple modular tasks (e.g. summarization, translation, etc.)
    </Card>
  </Tab>

  <Tab title="OpenAI">
    <CardGroup cols={2}>
      <Card title="GPT-4o" icon="wand-magic-sparkles">
        Base model for Nevron orchestration and function calling
      </Card>

      <Card title="o3 Series" icon="eye">
        Advanced reasoning tasks (e.g. internal evaluation of the found alpha)
      </Card>
    </CardGroup>
  </Tab>

  <Tab title="X AI">
    <Card title="Grok 2/3" icon="brain">
      Direct communication with users + sentiment analysis tasks
    </Card>
  </Tab>

  <Tab title="Google">
    <Card title="Gemini Pro" icon="expand">
      Large context processing (e.g. summarization of daily interactions)
    </Card>
  </Tab>

  <Tab title="Custom">
    <Card title="Fine-Tuned Models" icon="sliders">
      * Personality alignment
      * Writing tasks
      * Specified tasks for data processing
    </Card>
  </Tab>
</Tabs>

### Platform-Specific Examples

While Neurobro, as an AI agent that encompasses all the Nevrons, remains a single agent, its presence on different platforms significantly varies. Here are some examples of how Neurobro works across different platforms:

<AccordionGroup>
  <Accordion icon="terminal" title="Neurodex">
    * The main platform with full coverage of all Nevrons and their capabilities
  </Accordion>

  <Accordion icon="telegram" title="Telegram">
    * Real-time responses to user questions
    * Automated posting of found signals and analysis
  </Accordion>

  <Accordion icon="twitter" title="𝕏 (Twitter)">
    <Frame>
      <img src="https://mintcdn.com/neurobro/l4gM_i_EkePyUTVO/images/example-twitter.webp?fit=max&auto=format&n=l4gM_i_EkePyUTVO&q=85&s=5b6651cec4b4aefcc59ae317362992ff" style={{ borderRadius: '0.5rem' }} width="1176" height="704" data-path="images/example-twitter.webp" />
    </Frame>

    * "@0xNeurobro" mention monitoring and replying
    * Commenting on threads
    * Automated posting of found signals and analysis
  </Accordion>

  <Accordion icon="base" title="Baseapp">
    * Direct messaging with Neurobro in light mode for fast answers and balanced layer of Neurobro intelligence
    * Secure messaging with full e2e encryption via XMTP protocol
    * Seamless trading capabilities with integrated Baseapp XMTP swaps support
  </Accordion>

  <Accordion icon="mobile-screen" title="Mobile App">
    * Neurobro intelligence on the go via iOS and Android apps
    * Broader market coverage: crypto, stocks, forex, commodities, and prediction markets
    * Chat interface with daily AI questions ([tier-based](/neurobro/pass))
    * Swipeable personalized news feed (swipe left/right to curate, swipe up to chat about news)
    * Shared backend and agent infrastructure with Neurodex
  </Accordion>

  <Accordion icon="credit-card" title="x402 Micropayments">
    * HTTP 402-based micropayment protocol for AI services
    * **Full access to Neurobro's intelligence** - anyone can connect programmatically and leverage the complete capabilities of 200+ Nevrons
    * Pay-per-request model with no subscriptions or gatekeeping
    * Ideal for developers, traders, and AI builders who want to integrate Neurobro into their own applications
    * Open-source client examples available on [GitHub](https://github.com/axioma-ai-labs/x402-neurobro)
  </Accordion>
</AccordionGroup>

<Info>
  To learn more about the Neurobro AI Agents, refer to the [Neurodex Agents](/technical/neurodex/agents) section, since it's the main platform with full capabilities of the Neurobro AI Agents.
</Info>

## Data

> GIGO = Garbage in, garbage out

AI agents are only as good as the data they process. This is why we invest heavily in building a proprietary financial data infrastructure - the highest quality, most comprehensive data layer we can build. This is the second foundation of Neurobro's intelligence.

Our main data sources:

### Agent Memory

Agent Memory is similar to human memory, but for AI Agents. It consists of opinionated data points about the ecosystem the agent operates in. We currently use a blend of vector stores and graph databases to store this representation of memory.

<AccordionGroup>
  <Accordion icon="database" title="Vector Stores">
    <CardGroup cols={2}>
      <Card title="Qdrant" icon="server">
        Primary vector database
      </Card>

      <Card title="Weaviate" icon="layer-group">
        Secondary vector store
      </Card>
    </CardGroup>
  </Accordion>

  <Accordion icon="graph" title="Graph Databases">
    <CardGroup cols={2}>
      <Card title="Neo4j" icon="database">
        Primary graph database
      </Card>
    </CardGroup>
  </Accordion>
</AccordionGroup>

A crucial part of the vectors are embeddings. All embeddings are generated using OpenAI's state-of-the-art **text-embedding-3-large** model, ensuring:

<CardGroup cols={3}>
  <Card title="Precision" icon="bullseye">
    High-accuracy matching
  </Card>

  <Card title="Relevance" icon="magnifying-glass">
    Context-aware results
  </Card>

  <Card title="Performance" icon="gauge">
    Optimized processing
  </Card>
</CardGroup>

### On-Chain Data

We track 3,000+ whales on Base chain and analyze their activity in near real-time. Raw blockchain transactions are decoded and enriched with custom ML labels and technical data (pricing, volume, liquidity).

Our proprietary ML labeling system classifies whales into behavioral categories, tracks trading patterns, and performs performance analysis. Human-in-the-loop validation ensures data quality - critical because raw on-chain data contains significant noise, failed transactions, and scam activity.

This data feeds directly into specialized Nevrons for analysis and is also exposed to users through the Smart Money Dashboard on Neurodex.

<Info>
  See the [Smart Money Dashboard](https://neurodex.ai/smart-money) for more details.
</Info>

<CardGroup cols={2}>
  <Card title="PostgreSQL" icon="database">
    Primary relational database
  </Card>

  <Card title="MongoDB" icon="database">
    Document database for flexible schemas
  </Card>
</CardGroup>

### News, Social & Sentiment

We aggregate financial news, social media signals, and market commentary from across the web. RSS feeds, X API integrations, and other sources provide a continuous stream of market-relevant information.

Since raw social and news data is extremely noisy, specialized Nevrons filter, aggregate, and deduplicate information - maintaining a clean, up-to-date knowledge base.

<CardGroup cols={2}>
  <Card title="PostgreSQL" icon="database">
    Primary relational database
  </Card>

  <Card title="MongoDB" icon="database">
    Document database for flexible schemas
  </Card>
</CardGroup>

### Market Data

Technical market data (pricing, volume, liquidity, order books) is sourced from a combination of proprietary and third-party APIs including Coingecko, DexScreener, Base BlockScout, and Alchemy.

### Proprietary Data Layer

The majority of Neurobro's value comes from proprietary data processing and analysis. We do not disclose full details of our data sources for competitive reasons.

<CardGroup cols={2}>
  <Card title="Proprietary APIs" icon="lock">
    \~90% of visible value
  </Card>

  <Card title="Public APIs" icon="globe">
    \~10% of functionality

    Examples: Coingecko, DexScreener, Base BlockScout
  </Card>
</CardGroup>

## Conclusion

Neurobro's intelligence is built on two pillars: specialized AI agents and a proprietary financial data infrastructure.

The combination of a comprehensive data layer with asymmetric access to market information, plus an AI architecture designed to exceed the limitations of standalone LLMs, is what enables Neurobro to deliver institutional-grade financial intelligence to everyone.