Overview

Probably is built on a local-first architecture that keeps your data on your devices while leveraging AI capabilities through secure, encrypted connections.

Local-first software prioritizes user agency and data ownership:

• Data Ownership: Your data belongs to you and stays on your devices
• Privacy by Design: No surveillance, tracking, or data mining
• Offline Capability: Core functionality works without internet connection
• Performance: Local processing is faster than cloud alternatives
• Reliability: No dependence on external services for basic operations

Probably implements local-first principles through:

Local Data Processing

• All data analysis happens on your machine
• Database connections are direct from your computer
• No data passes through external servers
• Full control over what data is processed

Selective AI Integration

• AI providers only receive specific queries you authorize
• Sample data and chart information sent only when explicitly using AI features
• Responses are processed locally
• All AI interactions are encrypted and logged

Transparent Operations

• Clear visibility into what data is accessed
• Audit trail of all AI interactions
• User control over every external connection
• Open architecture with no hidden data flows

Complete Data Control

• Your sensitive business data never leaves your premises
• No risk of data breaches at cloud providers
• Compliance with strict data privacy regulations
• Protection against surveillance and data mining

Encrypted External Communication

• All AI provider communications use encryption
• API keys stored securely on your device
• No persistent data storage on external servers
• Zero-knowledge architecture for AI interactions

Local Processing Speed

• Database queries execute at native speed
• No network latency for data operations
• Immediate response for most analytical tasks
• Efficient use of local computing resources

Reduced Bandwidth Usage

• Only essential AI queries use internet
• No large data transfers to cloud services
• Efficient caching of AI responses
• Minimized network dependencies

Offline Capability

• Core analysis functions work without internet
• Local data sources always accessible
• Cached AI responses available offline
• No downtime due to external service issues

Service Independence

• Not dependent on any single cloud provider
• Multiple AI provider support reduces vendor lock-in
• Local backup and restore capabilities
• User control over service dependencies

Your Data Sources → Local Processing → AI Queries (when needed) → Local Results
     ↓                    ↓                     ↓                      ↓
[Databases]         [Your Computer]      [Encrypted APIs]      [Your Interface]
[Files]             [Analysis Engine]    [AI Providers]       [Visualizations]
[Spreadsheets]      [PXL Runtime]        [Secure Channels]    [Export Options]

Data Processing Engine

• Runs entirely on your machine
• Handles all data transformations
• Executes PXL expressions locally
• Manages visualization generation

Security Layer

• Encrypts all external communications
• Manages API key storage
• Implements access controls
• Provides audit logging

Caching System

• Stores AI responses locally
• Reduces external API calls
• Improves performance
• Respects data privacy

AI Provider Queries

• Only send specific analysis requests
• Never include raw sensitive data
• Use encrypted communication channels
• Receive only relevant AI responses

Database Connections

• Direct connections from your machine
• No data routed through external services
• Standard database security protocols
• Your existing access controls apply

All Your Data

• Source databases and files
• Analysis results and insights
• Visualizations and reports
• User configurations and preferences

Processing Operations

• Data cleaning and transformation
• Statistical calculations
• Aggregation and filtering
• Visualization generation

Sensitive Information

• Customer data and PII
• Financial information
• Proprietary business data
• Internal analytics and metrics

AI Analysis Requests

• Natural language questions and analysis requests
• Variable metadata (names, types, sample values)
• Chart data and visualizations for analysis
• Complete column data when using semantic functions (extract, classify, score)

Always with Your Control

• Explicit user authorization required
• Clear disclosure of what data is shared
• Option to disable external AI entirely
• Full audit trail of all interactions

GDPR Compliance

• Data minimization by design
• User control over data processing
• Right to data portability
• Clear consent mechanisms

Industry Standards

• SOC 2 Type II compatible architecture
• HIPAA-friendly data handling
• PCI DSS compliant payment data protection
• ISO 27001 security framework alignment

Transparency Features

• Complete audit logs of data access
• Clear documentation of data flows
• User-controlled privacy settings
• Regular security updates and patches

Data Governance

• User-defined data retention policies
• Configurable access controls
• Backup and recovery under user control
• No vendor lock-in for data export

Probably combines local-first benefits with AI capabilities:

• Local processing for all data operations
• Selective AI integration for advanced analysis
• User control over external service usage
• Transparent operations with full audit trails

Hardware Specifications

• Sufficient RAM for dataset processing
• Local storage for data and cache
• Network connectivity for AI services
• Standard business computer specifications

Software Dependencies

• Supported operating systems
• Database drivers as needed
• Network security compliance
• Standard business software stack

Individual Users

• Desktop application installation
• Personal data source connections
• Individual AI provider accounts
• Local configuration management

Team and Enterprise

• Shared data source configurations
• Centralized license management
• Team collaboration features
• Enterprise security integration

Hybrid Approaches

• Mix of local and shared data sources
• Configurable AI service usage
• Flexible deployment options
• Scalable architecture design