Page cover

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.


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