Why Nexus?

The Problem with AI Automation Today

Most AI automation tools have a critical flaw: your agents forget everything the moment a workflow ends.

| Challenge | Traditional n8n | With Nexus | |-----------|----------------|------------| | Memory | Agents forget after workflow ends | Persistent episodic + semantic memory | | Knowledge | Manual document management | Auto-indexed GraphRAG knowledge base | | Agents | Single-agent, linear workflows | Multi-agent collaboration & competition | | Context | Token limits force truncation | Intelligent context retrieval | | Specialization | Generic LLM prompts | 11 domain-trained AI services |

Built for AI Agents That:

• Remember — Past interactions inform future responses
• Reason — Query your entire knowledge base, not just recent context
• Collaborate — Multiple specialized agents working together
• Scale — From prototype to enterprise without replatforming

Architecture

graph TB
    subgraph "n8n Workflow"
        T[Trigger] --> N[Nexus Node]
        N --> R[Response]
    end

    subgraph "Adverant Nexus Platform"
        N --> API[Nexus API Gateway]
        API --> GR[GraphRAG Engine]
        API --> MA[MageAgent Orchestrator]
        API --> SP[Specialized Services]

        GR --> VS[(Vector Store)]
        GR --> KG[(Knowledge Graph)]
        GR --> ES[(Episodic Memory)]

        MA --> A1[Research Agent]
        MA --> A2[Analysis Agent]
        MA --> A3[Synthesis Agent]

        SP --> GEO[GeoAgent]
        SP --> LAW[NexusLaw]
        SP --> MED[NexusDoc]
        SP --> SEC[CyberAgent]
        SP --> VID[VideoAgent]
    end

    style GR fill:#4DB8D8,color:#fff
    style MA fill:#007AFF,color:#fff
    style SP fill:#5F6368,color:#fff

Core Capabilities

GraphRAG Knowledge Base

Free for open source users — 500 docs/month, no credit card required

Your AI agents finally remember. Store documents once, retrieve intelligently forever.

• Semantic Search — Find by meaning, not just keywords
• Graph Relationships — Understand how concepts connect
• Episodic Memory — Remember conversation context across sessions
• Multi-Modal — Text, PDFs, images, and structured data

MageAgent Orchestration

Multi-agent AI that actually works together

Spawn specialized agents that collaborate, compete, or work in parallel—with shared memory and coordinated goals.

| Mode | How It Works | Best For | |------|-------------|----------| | Collaborate | Agents build on each other's work | Research, analysis | | Compete | Multiple approaches, best answer wins | Creative tasks, alternatives | | Hybrid | Competition with collaborative synthesis | Complex decisions |

11 Specialized AI Services

| Service | Superpower | Use Case | |---------|------------|----------| | GeoAgent | Planetary-scale spatial analysis | Real estate, logistics, climate | | NexusLaw | Contract intelligence | Legal review, compliance | | NexusDoc | Clinical document understanding | Healthcare, insurance | | CyberAgent | Threat detection & analysis | Security operations | | VideoAgent | Video understanding & transcription | Media, surveillance | | NexusCRM | Customer intelligence | Sales, support | | Sandbox | Secure code execution | Data science, automation | | FileProcess | Document extraction at scale | OCR, data extraction | | Orchestration | Workflow automation | Complex pipelines |

Quick Start

1. Install the Node

Via n8n Community Nodes (Recommended)

Settings → Community Nodes → Install → n8n-nodes-adverant-nexus

Via npm (Self-hosted)

cd ~/.n8n/custom && npm install n8n-nodes-adverant-nexus

Via Docker

environment:
  - N8N_CUSTOM_EXTENSIONS=/home/node/.n8n/custom
volumes:
  - ./custom:/home/node/.n8n/custom

2. Get Your Free API Key

1. Sign up at dashboard.adverant.ai
2. Create an API key (starts with nx_live_ or nx_test_)
3. GraphRAG is free — no credit card required

3. Add Credentials in n8n

1. Go to Credentials → Add Credential
2. Search for Adverant Nexus API
3. Enter your API key and select environment
4. Click Save

How It Works

sequenceDiagram
    participant W as n8n Workflow
    participant N as Nexus Node
    participant G as GraphRAG
    participant M as MageAgent
    participant S as Services

    W->>N: User Query
    N->>G: Retrieve Context
    G-->>N: Relevant Knowledge + Memory
    N->>M: Orchestrate Agents
    M->>S: Delegate to Specialists
    S-->>M: Domain Insights
    M-->>N: Synthesized Response
    N->>G: Store New Memory
    N-->>W: Final Response

Example Workflows

Customer Support with Memory

graph LR
    A[Customer Message] --> B[Recall Memory]
    B --> C[Search Knowledge Base]
    C --> D[Generate Response]
    D --> E[Store Interaction]
    E --> F[Send Reply]

What this enables:

• Support agent remembers past conversations
• Retrieves relevant docs automatically
• Learns from every interaction

{
  "nodes": [
    {
      "name": "Recall Customer History",
      "type": "n8n-nodes-adverant-nexus.adverantNexus",
      "parameters": {
        "service": "graphrag",
        "operation": "recallMemory",
        "query": "={{ $json.customer_id }}",
        "limit": 10
      }
    },
    {
      "name": "Search Knowledge Base",
      "type": "n8n-nodes-adverant-nexus.adverantNexus",
      "parameters": {
        "service": "graphrag",
        "operation": "search",
        "query": "={{ $json.message }}",
        "limit": 5
      }
    }
  ]
}

Download Full Workflow →

AI Research Assistant

graph LR
    A[Research Query] --> B[MageAgent Orchestrate]
    B --> C[Research Agent]
    B --> D[Analysis Agent]
    B --> E[Writing Agent]
    C --> F[Synthesize]
    D --> F
    E --> F
    F --> G[Final Report]

What this enables:

• Multiple agents research in parallel
• Specialized roles (researcher, analyst, writer)
• Collaborative synthesis of findings

{
  "service": "mageagent",
  "operation": "orchestrate",
  "task": "Research the impact of AI on healthcare and provide a comprehensive analysis with recommendations",
  "maxAgents": 5,
  "context": {
    "outputFormat": "detailed_report",
    "includeSourceCitations": true
  }
}

Download Full Workflow →

Document Processing Pipeline

graph LR
    A[Upload Document] --> B[FileProcess Extract]
    B --> C[GraphRAG Store]
    C --> D[Auto-Tag & Index]
    D --> E[Available for Search]

What this enables:

• Automatic document ingestion
• Intelligent chunking and indexing
• Instant searchability

{
  "service": "fileprocess",
  "operation": "submitUrl",
  "fileUrl": "https://example.com/document.pdf",
  "filename": "quarterly-report.pdf",
  "options": {
    "extractTables": true,
    "preserveFormatting": true
  }
}

Download Full Workflow →

Webhook Triggers

Real-time event notifications for all services:

| Category | Events | |----------|--------| | Documents | stored, deleted, updated, processed, failed | | Memory | stored, recalled | | Agents | started, completed, failed, spawned | | Files | submitted, processing, completed, failed | | Video | submitted, processing, transcribed, completed | | Security | threat_detected, scan_completed, vulnerability_found |

How Nexus Compares

| Capability | Zapier | Make | n8n (native) | Nexus | |------------|--------|------|--------------|-----------| | General Automation | ✅ | ✅ | ✅ | ✅ | | Persistent AI Memory | ❌ | ❌ | ❌ | ✅ | | Knowledge Base (RAG) | ❌ | ⚠️ | ⚠️ | ✅ | | GraphRAG | ❌ | ❌ | ❌ | ✅ | | Multi-Agent Orchestration | ❌ | ❌ | ⚠️ | ✅ | | Specialized AI Services | ❌ | ❌ | ❌ | ✅ (11) | | Self-Hosted Option | ❌ | ❌ | ✅ | ✅ | | Free Tier | ⚠️ | ⚠️ | ✅ | ✅ |

✅ Full support • ⚠️ Limited/workaround • ❌ Not available

Pricing

| Plan | Price | What's Included | |------|-------|-----------------| | Open Source | Free | GraphRAG (500 docs/mo), 10K searches | | Starter | $9/mo | GraphRAG + à la carte services | | Pro | $29/mo | GraphRAG + MageAgent + discounts | | Business | $99/mo | All services, unlimited, priority support |

All plans include: API access, webhook triggers, community support

View Full Pricing →

Complete Operations Reference

| Operation | Description | |-----------|-------------| | storeDocument | Store a document with automatic chunking and indexing | | storeMemory | Store episodic memory for conversation context | | recallMemory | Retrieve relevant memories for a query | | search | Semantic search across all documents | | listDocuments | List all stored documents | | getDocument | Retrieve a specific document by ID | | deleteDocument | Remove a document from the knowledge base | | enhancedRetrieve | Advanced retrieval with graph traversal |

| Operation | Description | |-----------|-------------| | orchestrate | Spawn and coordinate multiple AI agents | | analyze | Deep analysis with specialized agents | | synthesize | Combine multiple sources into coherent output | | collaborate | Agents work together on a shared goal | | compete | Multiple agents compete for best solution | | modelSelect | Dynamically select optimal model for task |

| Service | Operations | |---------|------------| | FileProcess | submitFile, submitUrl, getStatus, getResult, cancelJob | | Sandbox | execute, getLanguages | | Orchestration | submitTask, getStatus, cancelTask, listAgents | | GeoAgent | spatialQuery, spatialJoin, earthEngineAnalyze, timeSeries, vertexPredict | | CyberAgent | threatAnalysis, vulnerabilityScan, securityReport, iocLookup | | VideoAgent | submitVideo, transcribe, analyzeFrames, getStatus, getResult | | NexusLaw | contractAnalysis, clauseExtraction, riskAssessment, documentCompare | | NexusCRM | customerInsights, leadScoring, sentimentAnalysis, churnPrediction | | NexusDoc | medicalAnalysis, clinicalExtraction, icdCoding, patientSummary |

Support & Resources

| Resource | Link | |----------|------| | Documentation | adverant.ai/docs | | Dashboard | dashboard.adverant.ai | | Getting Started Guide | docs/GETTING_STARTED.md | | Example Workflows | examples/ | | GitHub Issues | Report Bugs | | Discord Community | discord.gg/adverant | | Email Support | [email protected] |

Contributing

We welcome contributions! Please read our Contributing Guide for details on:

• Setting up your development environment
• Code style and commit conventions
• Adding new services and operations
• Submitting pull requests

Security

Found a vulnerability? Please report it responsibly via SECURITY.md or email [email protected].

License

MIT License — see LICENSE for details.