AI Agent Builder

Design and build AI agents with tools, memory, and multi-step reasoning capabilities. Covers ChatGPT, Claude, Gemini integration patterns based on n8n's 5,000+ AI workflow templates.

Overview

This skill covers:

• AI agent architecture design
• Tool/function calling patterns
• Memory and context management
• Multi-step reasoning workflows
• Platform integrations (Slack, Telegram, Web)

AI Agent Architecture

Core Components

┌─────────────────────────────────────────────────────────────────┐
│                      AI AGENT ARCHITECTURE                       │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  ┌─────────────┐     ┌─────────────┐     ┌─────────────┐       │
│  │   Input     │────▶│   Agent     │────▶│   Output    │       │
│  │  (Query)    │     │   (LLM)     │     │  (Response) │       │
│  └─────────────┘     └──────┬──────┘     └─────────────┘       │
│                             │                                   │
│         ┌───────────────────┼───────────────────┐              │
│         │                   │                   │              │
│         ▼                   ▼                   ▼              │
│  ┌─────────────┐     ┌─────────────┐     ┌─────────────┐       │
│  │   Tools     │     │   Memory    │     │  Knowledge  │       │
│  │ (Functions) │     │  (Context)  │     │   (RAG)     │       │
│  └─────────────┘     └─────────────┘     └─────────────┘       │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Agent Types

agent_types:
  reactive_agent:
    description: "Single-turn response, no memory"
    use_case: simple_qa, classification
    complexity: low
    
  conversational_agent:
    description: "Multi-turn with conversation memory"
    use_case: chatbots, support
    complexity: medium
    
  tool_using_agent:
    description: "Can call external tools/APIs"
    use_case: data_lookup, actions
    complexity: medium
    
  reasoning_agent:
    description: "Multi-step planning and execution"
    use_case: complex_tasks, research
    complexity: high
    
  multi_agent:
    description: "Multiple specialized agents collaborating"
    use_case: complex_workflows
    complexity: very_high

Tool Calling Pattern

Tool Definition

tool_definition:
  name: "get_weather"
  description: "Get current weather for a location"
  parameters:
    type: object
    properties:
      location:
        type: string
        description: "City name or coordinates"
      units:
        type: string
        enum: ["celsius", "fahrenheit"]
        default: "celsius"
    required: ["location"]
    
  implementation:
    type: api_call
    endpoint: "https://api.weather.com/v1/current"
    method: GET
    params:
      q: "{location}"
      units: "{units}"

Common Tool Categories

tool_categories:
  data_retrieval:
    - web_search: search the internet
    - database_query: query SQL/NoSQL
    - api_lookup: call external APIs
    - file_read: read documents
    
  actions:
    - send_email: send emails
    - create_calendar: schedule events
    - update_crm: modify CRM records
    - post_slack: send Slack messages
    
  computation:
    - calculator: math operations
    - code_interpreter: run Python
    - data_analysis: analyze datasets
    
  generation:
    - image_generation: create images
    - document_creation: generate docs
    - chart_creation: create visualizations

n8n Tool Integration

n8n_agent_workflow:
  nodes:
    - trigger:
        type: webhook
        path: "/ai-agent"
        
    - ai_agent:
        type: "@n8n/n8n-nodes-langchain.agent"
        model: openai_gpt4
        system_prompt: |
          You are a helpful assistant that can:
          1. Search the web for information
          2. Query our customer database
          3. Send emails on behalf of the user
          
        tools:
          - web_search
          - database_query
          - send_email
          
    - respond:
        type: respond_to_webhook
        data: "{{ $json.output }}"

Memory Patterns

Memory Types

memory_types:
  buffer_memory:
    description: "Store last N messages"
    implementation: |
      messages = []
      def add_message(role, content):
          messages.append({"role": role, "content": content})
          if len(messages) > MAX_MESSAGES:
              messages.pop(0)
    use_case: simple_chatbots
    
  summary_memory:
    description: "Summarize conversation periodically"
    implementation: |
      When messages > threshold:
          summary = llm.summarize(messages[:-5])
          messages = [summary_message] + messages[-5:]
    use_case: long_conversations
    
  vector_memory:
    description: "Store in vector DB for semantic retrieval"
    implementation: |
      # Store
      embedding = embed(message)
      vector_db.insert(embedding, message)
      
      # Retrieve
      relevant = vector_db.search(query_embedding, k=5)
    use_case: knowledge_retrieval
    
  entity_memory:
    description: "Track entities mentioned in conversation"
    implementation: |
      entities = {}
      def update_entities(message):
          extracted = llm.extract_entities(message)
          entities.update(extracted)
    use_case: personalized_assistants

Context Window Management

context_management:
  strategies:
    sliding_window:
      keep: last_n_messages
      n: 10
      
    relevance_based:
      method: embed_and_rank
      keep: top_k_relevant
      k: 5
      
    hierarchical:
      levels:
        - immediate: last_3_messages
        - recent: summary_of_last_10
        - long_term: key_facts_from_all
        
  token_budget:
    total: 8000
    system_prompt: 1000
    tools: 1000
    memory: 4000
    current_query: 1000