> ## Documentation Index
> Fetch the complete documentation index at: https://neuroai-cf1c4abf.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Overview
> Discover the key technologies and capabilities that power Neuro AI
## Key technologies and capabilities
### Northstar LLM
Our proprietary large language model, trained on specific business data through simulations in virtual worlds on Google supercomputers. Performance comparable to Gemini 3.0.
**Technical specifications:**
* Trained using advanced reinforcement learning algorithms
* Optimized for audit and financial analysis tasks
* **96.8% accuracy** on FinQA benchmark
### Super agent (the most powerful Agent)
The platform's main executor, which:
* Works faster, better, and more scalably than competing solutions
* Capable of autonomously planning and executing complex action chains
* Manages other AI agents as a unified control center
* Intelligently delegates tasks to the most suitable agents
* Synthesizes outputs from all agents into a unified result
### Vibe tasking
A concept allowing users to convey the "vibe" (intention, idea, context), and the platform automatically executes the full work cycle from planning to result.
### Dynamic planner
A dynamic planner that helps plan and divide tasks into subtasks. Analyzes processes and builds an implementation plan.
**Capabilities:**
* Breaks down complex tasks into logical step sequences
* Coordinates execution among multiple agents
* Adapts the plan in real-time based on results
### Contextual workspace
A contextual workspace where all information is stored:
* **Memory** — short-term and long-term
* **Personal preferences** — how to better understand and analyze the user
* **All project data** — files, context, history
## Modular architecture
Neuro AI is built on a sophisticated modular architecture, where each module performs a specific function. Together, these modules ensure the system operates as a true autonomous agent.
### Key modules
**AI agent module**
* Functions as a specialized intelligent agent
* Performs tasks analogous to an employee's work in a specific role
* Capable of making independent decisions within its competence
**Language model module**
* Based on large language models (LLMs)
* Understands and generates text in natural languages
* Analyzes information and performs logical tasks
**Search module**
* Finds and extracts information from the internet
* Works with databases and other sources
* Provides current data for analysis
**Self-learning module**
* Learns from its own experience
* Improves performance over time
* Adapts to new situations
**File module**
* Manages file operations
* Reads, writes, organizes, and processes documents
* Works with various data formats
**Memory module**
* Maintains short-term and long-term memory
* Ensures context preservation between sessions
* Learns from past interactions
**Control module**
* Monitors and checks work results
* Ensures quality and accuracy before providing to user
* Manages output data quality
**External tools module**
* Integrates with external APIs, software systems, and services
* Allows Neuro AI to interact with the broad digital ecosystem
* Provides connectivity with 200+ services through MCP
**Planning module**
* Breaks down complex tasks into logical step sequences
* Coordinates execution among multiple agents
* Synchronizes the entire system's work
## Advanced technologies
### A2A (agent-to-agent) technology
Technology that allows multiple AI agents to communicate and collaborate with each other:
* Information exchange between agents
* Action coordination
* Joint solving of complex tasks
### MCP (model context protocol)
Ensures consistent context management:
* Unified context across different modules and agents
* Maintaining coherence throughout complex workflows
* 500+ integrations/apps with external services
### Autonomous execution engine
Once the goal is defined, Neuro AI's engine:
* Breaks the task into a logical sequence of steps
* Executes these steps using various agents and tools
* Works without constant user input
### Multi-agent collaboration
Neuro AI uses a team of specialized AI agents:
* **Research agent** — gathering information from the web
* **Data analysis agent** — processing information
* **Content generation agent** — composing reports
* Agents work in coordination, passing information and results to each other
### Integrated toolset
The virtual workspace includes:
* Web browser for interacting with sites
* Code editor for writing and running programs
* File system for working with documents
* Ability to interact with the digital world like a human
### Persistent memory
Unlike many AI models with limited context windows:
* Long-term memory system
* Remembers past interactions
* Stores user preferences
* Remembers project details across multiple sessions
* Learns and adapts over time
### Real-time monitoring
* Transparent view of system operations
* Ability to track progress in real-time
* User can intervene and provide feedback
## Technical performance metrics
**Measurable performance indicators:**
* **Sub-2-second latency** for complex queries
* **96.8% on FinQA benchmark**
These quantitative metrics build trust and confirm enterprise-level solution.
## Key differences from competitors
### What makes Neuro AI a true AI agent
**Four key capabilities of a true AI agent:**
**1. Perceives the world** — sees, hears, reads, and analyzes the environment. Processes text, images, video, and data from multiple sources.
**2. Understands context** — doesn't just react to keywords, but understands the full task context, including goals, constraints, and connections between information.
**3. Makes decisions based on experience and memory** — maintains short-term and long-term memory, learns from past interactions, and applies knowledge to new situations.
**4. Acts autonomously** — calls APIs, sends emails, manages software, controls systems, and executes complex workflows without step-by-step instructions.
### Comparison with traditional AI systems
| **Criterion** | **RAG systems** | **Voice assistants** | **Neuro AI** |
| ---------------------- | ----------------------------------------------------------------- | --------------------------- | ------------------------------------------------------------------------------- |
| **Technology essence** | "Smart search + text generation" | "Script + voice" | "Intelligent system that understands, remembers, plans, and acts" |
| **Perception** | Text from database only | Voice commands | Text, images, video, data from multiple sources |
| **Understanding** | Extracts fragments from vector database and inserts into response | Recognizes command keywords | Analyzes context, goals, connections between data |
| **Memory** | No long-term memory | No experience memory | Short-term + long-term memory, learning from experience |
| **Decision making** | Doesn't make decisions | Executes predefined scripts | Reasons, plans, makes independent decisions |
| **Actions** | Doesn't perform actions | Limited API commands | Calls APIs, writes reports, sends emails, manages CRM, updates systems |
| **Learning** | Doesn't learn from experience | Doesn't adapt | Learns from mistakes, accumulates knowledge, transfers experience between tasks |
| **Autonomy** | Requires constant management | Reacts only to commands | Works autonomously, executes complex multi-step tasks from start to finish |
### From conversation to action
**Alternatives:**
Claude, ChatGPT, Perplexity — require lots of copy-paste and manual work
**Our advantages:**
**1. Agent-first**
The user interacts with a single intelligent agent, which is the main entry point for solving any task.
**2. Intent-to-action**
The system can transform high-level natural language requests ("create a portfolio website", "analyze sales report") into a complex sequence of actions without the need for step-by-step instructions.
**3. True autonomy**
Neuro AI can work independently for extended periods, processing complex long-term tasks in the background, freeing users for strategic work.
**4. A workforce of agents**
This isn't one AI — it's a team of AI working together. The multi-agent approach provides a more powerful and flexible problem-solving model.
**5. Transparency and control**
The ability to monitor AI work and intervene when necessary builds trust and ensures the user remains in control of the process.
**6. Intelligence beyond imitation**
While RAG systems imitate knowledge and voice assistants imitate reactions, Neuro AI demonstrates true capabilities for reasoning, planning, and adaptation.
**The main difference:**
We didn't want to build another chatbot. We wanted to democratize AI power for ordinary people, without requiring them to be engineers or programmers. You simply set a task — the system determines the execution plan and follows it step by step in **fully autonomous mode**.
Neuro AI isn't built to help people work faster — it's built to **work instead of people**, automating entire workflows from start to finish.
## Explore features
Interact with AI through natural conversations
Deploy specialized AI agents for different tasks
Manage and automate complex workflows
Configure your Neuro AI