Navigator
Your AI Command Center
Navigator is where you design, fine-tune, and deploy specialized AI models and workflows. Whether you're customizing a template or building from scratch, Navigator gives you complete control over your AI applications through an intuitive visual interface.
What is Navigator?
Navigator is your AI development environment - a visual workflow builder that lets you create sophisticated AI applications without complex coding. Think of it as your control center where you design how AI processes data, responds to users, and integrates with your existing systems.
Core Philosophy
Navigator operates on a simple principle: AI applications are workflows. Every AI solution - from a simple chatbot to a complex document analysis system - can be broken down into connected steps that process data and generate outputs.
Visual Design:
Drag and drop elements to build AI workflows
Visual connections show how data flows through your application
Real-time preview of your AI application as you build
Complete Control:
Customize every aspect of your AI's behavior and responses
Train custom models on your own data
Deploy anywhere from single devices to distributed clusters
What You Can Build
Conversational AI
Smart Chatbots: Create AI assistants that understand your business, products, and customers. Unlike generic chatbots, Navigator lets you build AI that knows your industry terminology and follows your specific guidelines.
Internal Knowledge Assistants: Build AI systems that can answer questions about your company's processes, documentation, and institutional knowledge.
Specialized Advisors: Create AI experts in specific domains - from technical support to creative writing assistance.
Document Intelligence
Document QnA Systems: Upload your documents and create AI systems that can answer questions about the content, with source citations and context.
Content Analysis: Automatically process, categorize, and extract insights from large document collections.
Research Assistants: Build AI that can read through research papers, reports, or documentation and provide summaries or answer specific questions.
Custom AI Models
Domain-Specific Training: Train AI models on your own data to create specialists that understand your industry, terminology, and specific use cases.
Workflow Automation: Create AI-powered processes that can handle routine tasks, make decisions, and integrate with your existing systems.
API-Powered Applications: Build AI services that other applications can use, creating intelligent features across your technology stack.
Key Navigator Concepts
Visual Workflow Design
Elements: Individual components that perform specific functions - AI models, data processors, API connectors, and more. Each element has configurable settings that control its behavior.
Connections: Lines between elements that show how data flows through your application. Data enters through one element, gets processed by others, and produces outputs.
Canvas: Your visual workspace where you arrange elements and connections to design your AI application's logic.
Templates vs Custom Builds
Featured Templates: Pre-built workflows created by webAI experts for common use cases. Perfect starting points that you can customize for your specific needs.
LLM Chatbot: Conversational AI assistant
Document QnA: Question-answering system for your documents
Custom Training: Workflows for training specialized models
Custom Workflows: Built from scratch using individual elements. Gives you unlimited flexibility to create exactly what you need for unique use cases.
Development to Deployment
Development Mode: Build and test workflows locally in Navigator. Perfect for experimenting, iterating, and refining your AI applications.
Deployment: When ready, deploy your workflows to clusters of devices where they run independently and can be accessed through Companion or APIs.
Navigator's Element Library
AI Model Elements
LLM Chat: The core conversational AI element. Configurable for different personalities, expertise levels, and response styles.
Document QnA: Specialized AI that can read and understand documents, then answer questions about their content with citations.
Custom Trained Models: Your own AI models, trained on your specific data and optimized for your use cases.
Data Processing Elements
Document Processors:
OCR: Extract text from images and scanned documents
File Readers: Handle various document formats (PDF, Word, etc.)
Text Cleaners: Prepare and format text for AI processing
Vector Processing:
Embedding Generators: Convert text into mathematical representations
Vector Databases: Store and search through large amounts of text efficiently
Similarity Search: Find relevant content based on meaning, not just keywords
Content Managers:
Chunking: Break large documents into AI-friendly pieces
Metadata Extraction: Pull important information from document properties
Content Filters: Clean and prepare data for processing
Integration Elements
APIs and Connectors:
Web APIs: Connect to external services and data sources
Database Connectors: Access structured data from various databases
File Systems: Read from and write to local and network storage
Input/Output Handlers:
User Interfaces: Create chat interfaces and forms
File Uploads: Handle user-submitted documents and media
Response Formatters: Control how results are presented to users
Customization and Control
AI Behavior Control
System Prompts: Define how your AI should behave, what tone to use, and what knowledge to emphasize. This is where you give your AI its "personality" and expertise.
Response Parameters:
Temperature: Control creativity vs consistency in responses
Length Limits: Set response size boundaries
Format Control: Define output structure and style
Training Integration: Connect custom-trained models to leverage your organization's specific knowledge and terminology.
Data and Privacy Control
Local Processing: All AI processing happens on your devices. Your data never leaves your environment, ensuring complete privacy and control.
Custom Data Sources: Connect to your existing databases, document stores, and knowledge bases without compromising data security.
Performance Optimization
Model Selection: Choose from various AI models based on your hardware capabilities and performance requirements. For most use cases that require training or inference on consumer hardware, we recommend using a 7B parameter model.
Resource Management: Navigator helps you balance AI capability with available computing resources, ensuring optimal performance on your hardware.
Scaling Options: Deploy to single devices for personal use or across clusters for team and enterprise applications.
Integration Capabilities
Companion Integration
Your Navigator-built applications automatically appear in Companion, giving users a familiar chat interface to interact with your AI systems.
Seamless Connection:
Deploy workflows and they immediately become available in Companion
Multiple AI models can run simultaneously with separate conversation threads
Real-time updates when you modify workflows
API and System Integration
REST APIs: Generate API endpoints from your workflows, allowing other applications to access your AI capabilities.
Webhook Support: Connect to external systems that can trigger your AI workflows based on events or schedules.
Database Integration: Read from and write to various database systems, making your AI applications part of your existing data infrastructure.
Team Collaboration
Shared Workflows: Team members can collaborate on AI applications, sharing templates and best practices across your organization.
Version Control: Track changes to your AI workflows and revert to previous versions when needed.
Deployment Management: Control which workflows are deployed where, managing AI applications across multiple devices and teams.
Getting Started with Navigator
If You're New to AI Development
Start with Templates: Use Featured Templates to understand how AI workflows are structured. Customize them gradually as you learn how different elements work together.
Focus on One Use Case: Pick a specific problem you want to solve rather than trying to build a general-purpose AI system.
Experiment Safely: Navigator's local processing means you can experiment with your data without privacy or security concerns.
If You're an Experienced Developer
Explore Custom Workflows: Build workflows from scratch to create exactly what you need for your specific use cases.
API Integration: Connect Navigator workflows to your existing applications and systems for seamless AI integration.
Advanced Training: Train custom models on your data to create specialized AI that understands your domain expertise.
For Teams and Organizations
Template Libraries: Create organizational templates that capture your best practices and approved AI patterns.
Cluster Deployment: Scale AI applications across your infrastructure while maintaining complete control over data and processing.
Governance and Control: Implement controls over who can create, modify, and deploy AI applications within your organization.
Last updated