A powerful local-first desktop application that provides entrepreneurs with AI-powered legal assistance and business advisory through multiple AI models, 67 specialized legal practice areas, 67 MBA-level business advisory areas, intelligent multi-jurisdictional analysis with inter-provincial/interstate complexity awareness, and comprehensive business consulting capabilities spanning the full spectrum of business disciplines. With offline resilience and local data storage, your work remains accessible even when cloud services fail. The time & cost savings become a force multiplier.

⚠️ IMPORTANT DISCLAIMER: Atticus provides information, not legal advice. Always consult with licensed professionals for legal, financial, or business decisions. See RISK.md for comprehensive risk assessment.

Before using Atticus, please review these critical documents:

🚨 CRITICAL DISCLAIMER: This application provides information, not legal advice. AI outputs may contain errors, hallucinations, or outdated information. Always verify with licensed professionals. See RISK.md for complete risk disclosure.

1. Clone and install:

   git clone <repository>
   cd atticus
   npm install

2. Run in development:

   npm run dev

3. Configure providers - Get API keys and set up in Settings

📖 Full Documentation | 🎯 Quick Start Guide | 🔒 Privacy Policy | ⚠️ Risk Assessment | 🤖 Ethical AI Analysis

• Key Features
  • Multi-Provider AI Support
  • Multi-Model Selection
  • Real-Time Cost Transparency
  • AI Response Validation & Analysis
  • Multi-Jurisdictional Analysis
  • Legal Practice Area Detection
  • Business Advisory Detection
  • Advanced Configuration: YAML System Prompt Editing
  • Model Filtering & Customization
  • Rich Chat Interface
  • Privacy & Security Features
  • Offline Resilience & Business Continuity
  • Internationalization Support
• Architecture & Design
• Configuration
• Development
• Troubleshooting
• Privacy, Ethics & Risk
• Contributing
• License

Access 10 leading AI providers with 60+ models through one unified interface:

💡 Pro Tip: Select multiple models simultaneously to cross-validate legal analysis and reduce hallucinations!

Query multiple AI models simultaneously to reduce hallucinations and get diverse perspectives:

• Select 1-9+ models per conversation
• Parallel API execution for fast responses
• Model attribution badges on each response
• Compare answers across different AI systems
• Cross-validate legal analysis

See exactly what you're spending on each query with automatic token usage and cost reporting:

Real-time cost breakdown showing input/output token usage and pricing per query - complete transparency on API expenses

Why Direct Provider APIs Save Money:

Going directly to provider API endpoints (like OpenAI, Anthropic, Google) instead of using intermediary services provides significant cost savings:

• ✅ No markup fees - Pay only provider's base rate (e.g., GPT-4o: $2.50/1M input, $10/1M output)
• ✅ No subscription required - Pay per use only, no monthly minimums
• ✅ Volume discounts - Benefit from provider's tiered pricing as you scale
• ✅ Full control - Monitor and optimize costs in real-time
• ❌ Avoid 2-10x markups - Many AI chatbot services charge $20-50/month for what costs $2-5 in API usage

Cost Report Features:

• 📊 Collapsible display - Compact summary (cost, tokens, time) expands to full breakdown
• 💵 Exact pricing - Input tokens × rate + Output tokens × rate = Total cost
• 🎨 Color-coded tiers - Green (<$0.01), Amber ($0.01-$0.10), Red (>$0.10)
• ⚡ Performance metrics - Response time and tokens per second
• 🔍 Model attribution - See which model and provider generated each response

Example Cost Comparison:

💡 Pro Tip: Most queries cost $0.001 to $0.05 with direct API access. Even with multiple models selected, costs remain minimal compared to subscription services or specialized legal AI platforms.

Track API costs with complete transparency through the built-in Cost Ledger:

• Per-Message Breakdown - See input/output token costs for every API call
• Real-Time Totals - Track cumulative costs across entire conversations
• Color-Coded Tiers - Visual indicators for low/medium/high cost calls (green/amber/red)
• Detailed Analytics - Provider, model, timestamps, and individual cost components
• Export-Ready - Full ledger of expenses for budgeting and reporting

💡 Pro Tip: Use the Cost Ledger button (next to Audit Log) to review expenses and optimize model selection for cost-effective legal research.

Catch confabulations and validate accuracy with independent AI analysis of response clusters:

• Multi-Model Validation - Have an unused model independently analyze and judge responses from other models
• Consistency Checking - Identify conflicting information across different AI responses
• Accuracy Assessment - Detect potential inaccuracies, hallucinations, and confabulations
• Quality Ranking - Get comparative rankings of responses with justifications
• Trust Recommendations - Receive guidance on which responses are most reliable
• Iterative Analysis - Run multiple analyses with different validator models for consensus building

How it works:

1. Get responses from multiple models for your legal query
2. Click the Analysis button on the cluster action bar
3. Select an unused model to act as an independent validator
4. The validator model reviews all responses and provides detailed quality assessment
5. Results include consistency checks, accuracy ratings, completeness analysis, and recommendations

💡 Pro Tip: Use Analysis after getting responses from 2+ models to identify potential issues before relying on the advice. Multiple validator models can provide consensus validation for critical decisions.

Focus legal analysis on specific jurisdictions or get comparative analysis across CUSMA/USMCA region and beyond:

• 🇨🇦 Canada - Canadian federal and provincial law, including inter-provincial trade and regulatory differences
• 🇺🇸 United States - U.S. federal and state law, including interstate commerce and jurisdictional conflicts
• 🇲🇽 Mexico - Mexican federal and state law, including interstate commerce and regulatory variations
• 🇺 European Union - EU regulations and member state law, including cross-border harmonization

Inter-provincial/Interstate Complexity Coverage:

Atticus intelligently addresses internal jurisdictional complexities within each country:

🇨🇦 Canadian Inter-provincial:

• Division of powers - Federal vs. provincial jurisdiction (Constitution Act, 1867)
• Interprovincial trade barriers - Agreement on Internal Trade (AIT), Canadian Free Trade Agreement (CFTA)
• Provincial regulatory differences - Securities regulation, employment standards, professional licensing
• Civil law vs. common law - Quebec civil law vs. other provinces' common law
• Provincial incorporation - CBCA (federal) vs. provincial business corporations acts
• Key variations - Ontario (commercial hub), Quebec (civil law), BC, Alberta differences

🇺🇸 Interstate Complexity:

• Federal vs. state jurisdiction - Commerce Clause, Supremacy Clause, federalism principles
• Interstate commerce - Dormant Commerce Clause restrictions on state protectionism
• State regulatory variations - Corporate law, employment, taxation, professional licensing
• Conflicts of law - Choice of law provisions, forum shopping considerations
• State incorporation - Delaware (corporate law hub), vs. home state incorporation
• Key variations - Delaware (corporate), California (employment/consumer), New York (finance), Texas (business-friendly)

🇲🇽 Mexican Interstate:

• Federal vs. state jurisdiction - Mexican constitutional framework
• Interstate commerce - Regulatory coordination between states
• State taxation variations - State-level tax and commercial regulations
• Border states - Special regulations for northern border zone (Zona Libre)
• Key variations - Mexico City (financial hub), Nuevo León (manufacturing), Jalisco (tech), border states

CUSMA/USMCA Coverage:

The system is fully optimized for CUSMA/USMCA (Canada-United States-Mexico Agreement) cross-border operations:

• Trade compliance - CUSMA rules of origin, tariff classifications, customs procedures
• Cross-border business - Entity formation, tax planning, regulatory compliance across CA/US/MX
• Supply chain - Manufacturing, logistics, and distribution across North America
• Labor mobility - TN visas (Canada/Mexico to US), professional credentials recognition
• Intellectual property - Patent, trademark, and copyright protection in all three jurisdictions
• Dispute resolution - CUSMA Chapter 31 (dispute settlement), investor-state mechanisms
• Interstate/interprovincial coordination - Addressing internal barriers that affect cross-border operations

Intelligent Prompt Modification:

• Single jurisdiction → Focused analysis with interprovincial/interstate guidance
• Multiple jurisdictions → Structured comparative analysis with cross-border complexity
• CA + US + MX → CUSMA-specific guidance with internal trade barrier considerations
• No selection → Global legal perspective

Comprehensive per-conversation configuration: select multiple AI models simultaneously and choose specific jurisdictions for legal analysis (Canada, US, Mexico, EU) for optimal results

Automatically detects 67 specialized legal practice areas with 5,000+ keywords and applies expert-level AI prompts:

🎯 Startup-Focused Coverage (90% Optimization):

The practice area system is heavily optimized for startup legal needs:

• 🏢 Founder agreements and equity splits (vesting, cliffs, acceleration)
• 💰 Fundraising instruments (SAFE notes, convertible debt, priced rounds Series A-C)
• 📊 Cap table management (dilution modeling, option pools, pro rata rights)
• 🌍 Multi-jurisdiction operations (Canada flip, Delaware C-Corp, Mexico maquiladora, EU expansion)
• 💼 Exit strategies (M&A structuring, IPO readiness, acqui-hire negotiations)
• 🏛️ Government funding (SR&ED tax credits, SBIR/STTR grants, Mexico CONACYT, Horizon Europe)
• 🛡️ IP protection (patent strategy, trademark portfolio, open source compliance)
• 📝 Commercial contracts (SaaS agreements, enterprise MSAs, channel partnerships)
• 🚚 CUSMA/USMCA compliance (rules of origin, tariff classifications, cross-border supply chain)

Detection Features:

• ✅ 5,000+ keyword triggers - Comprehensive coverage across all areas
• ✅ Automatic detection - AI analyzes your query and selects appropriate practice area
• ✅ Confidence scoring - Know when detection is certain vs. uncertain
• ✅ Alternative suggestions - See other potential matches (e.g., "Also consider: IP Law, Tax Law")
• ✅ Expandable keywords - View all trigger keywords per area in Settings
• ✅ Expert system prompts - Each area has specialized AI instructions for accurate guidance

Geographic Coverage:

• 🇨🇦 Canada: 80% coverage (strongest in corporate, startup, tax, IP, employment)
• 🇺🇸 United States: 92% coverage (comprehensive across all practice areas)
• �🇽 Mexico: 70% coverage (focus on CUSMA trade, maquiladora, corporate, labor, IP)
• �🇪🇺 European Union: 75% coverage (focus on GDPR, cross-border, international)
• 🌐 CUSMA Region: Full cross-border coverage for CA/US/MX operations

Each area includes:

• ✅ Specialized system prompts - Expert-level guidance tailored to practice area
• ✅ Keyword detection - Automatic area identification from your query
• ✅ Confidence scoring - Transparency about detection accuracy
• ✅ Alternative suggestions - See other potential matches
• ✅ Expandable keywords - View all trigger keywords per area

Automatically detects 67 specialized business advisory areas with 3,500+ keywords for comprehensive strategic guidance across the full MBA spectrum:

🎯 MBA-Level Advisory Coverage (67 Areas):

Each advisory area provides consulting-grade guidance matching top firms (McKinsey, BCG, Bain, Goldman Sachs, Deloitte), covering the full MBA curriculum and beyond.

Advisory Features:

• ✅ 670+ core competencies (10 per area × 67 areas)
• ✅ World-class frameworks - McKinsey 7-S, BCG Matrix, Porter's Five Forces, Balanced Scorecard, COSO ERM, Lean/Six Sigma
• ✅ Industry best practices - Proven methodologies from top consulting firms and Fortune 500 companies
• ✅ Strategic templates - Business model canvas, OKRs, financial models, pitch decks, GTM plans, strategic roadmaps
• ✅ Implementation guidance - Step-by-step action plans, timelines, success metrics, and performance indicators
• ✅ Comprehensive MBA coverage - Strategy, finance, marketing, operations, HR, technology, specialized functions
• ✅ Industry verticals - Retail, healthcare, professional services, franchise, family business, nonprofit
• ✅ Specialized expertise - IP management, competitive intelligence, quality management, tax strategy, investor relations
• ✅ 3,500+ keyword triggers - Automatic advisory area detection across all 67 domains

Sample Guidance Quality:

Each advisory area includes 500-1,200 lines of expert content:

• Strategic Planning: ~500 lines (Porter's Five Forces deep dive, competitive moats, strategic planning process)
• Financial Advisory: ~600 lines (VC fundraising stages, 409A valuations, cap table modeling, M&A sell-side)
• Marketing Strategy: ~1,200 lines (GTM frameworks, product-market fit, pricing strategy, growth marketing)
• M&A Advisory: ~500 lines (PMI phases, synergy tracking, cultural integration, IT systems integration)
• Government Grants: ~400 lines (SR&ED claiming, SBIR phases, Horizon Europe, compliance best practices)
• Product Legal: ~400 lines (Privacy-by-design, WCAG compliance, open source licensing, regulatory pathways)

Each area includes:

• ✅ Business frameworks - McKinsey, BCG, Porter's Five Forces
• ✅ Industry best practices - Proven methodologies
• ✅ Strategic templates - Business model canvas, OKRs
• ✅ Implementation guidance - Actionable recommendations

Comprehensive settings organized in four intuitive tabs:

• 🔌 Providers Tab - Configure AI models and API keys
• ⚖️ Practice Areas Tab - View all 67 legal practice areas with expandable keywords
• 💼 Advisory Areas Tab - Explore all 67 business advisory areas
• ℹ️ About Tab - Mission, coverage statistics, capabilities overview

Key Features:

• Expandable keywords - Click "+X more" to see all detection keywords
• Color-coded areas - Legal (blue) vs Advisory (gold) distinction
• Geographic coverage - 🇨🇦 80% Canada, 🇺🇸 92% US, �🇽 70% Mexico, �🇪🇺 75% EU
• CUSMA/USMCA optimized - Full North American free trade zone support
• Copyright & disclaimer - Legal notices and AI usage guidelines

Full control over AI behavior with direct YAML configuration editing for power users:

Advanced YAML editor in Settings allows direct modification of practice areas, advisory areas, and analysis configurations - customize system prompts, keywords, detection rules, and AI behavior at a granular level

Comprehensive Editing Capabilities:

• 📝 Practice Areas Configuration - Edit all 67 legal practice areas directly in YAML

  • Customize system prompts with expert-level legal instructions for each area
  • Modify keyword detection rules (5,000+ keywords across all areas)
  • Adjust area names, descriptions, and color coding
  • Enable/disable specific practice areas
  • Add new practice areas with custom detection logic
  • 🌍 Language Variants: Edit practices.en.yaml, practices.fr.yaml, practices.es.yaml for localized guidance

• 💼 Advisory Areas Configuration - Edit all 67 business advisory areas

  • Customize consulting-grade system prompts for strategic guidance
  • Modify business keyword detection (3,500+ keywords)
  • Adjust advisory focus areas and descriptions
  • 🌍 Language Variants: Edit advisory.en.yaml, advisory.fr.yaml, advisory.es.yaml for localized guidance
  • Add new advisory domains with custom expertise

• 🔍 Analysis Configuration - Control AI response validation behavior

  • Customize system prompts for multi-model validation
  • Define accuracy assessment criteria
  • Configure consistency checking logic
  • Adjust confabulation detection sensitivity
  • 🌍 Language Variants: Edit analysis.en.yaml, analysis.fr.yaml, analysis.es.yaml for localized validation

Advanced Features:

• 🔄 Factory Reset - Restore default configurations from https://jdai.ca/atticus/

  • Separate reset buttons for Practice Areas, Advisory Areas, and Analysis
  • Confirmation dialogs prevent accidental resets
  • Fetch latest official configurations from remote endpoint

• 🛡️ Customization Protection - Modified configurations are protected from automatic updates

  • customized: true flag prevents overwriting your changes on app restart
  • Clear visual indicators show factory vs. customized status
  • Manual control over when to adopt upstream updates

• ✅ Real-Time Validation - Instant feedback on YAML syntax and structure

  • JSON Schema validation ensures configuration integrity
  • Duplicate ID detection prevents conflicts
  • Required field validation (id, name, description, keywords)
  • Color format validation for UI consistency

• 📊 Validation Scripts - Pre-startup validation for all configurations

  • validate-practices-clean.js - Validates all 67 practice areas
  • validate-advisory-clean.js - Validates all 67 advisory areas
  • validate-providers-clean.js - Validates provider configurations
  • validate-analysis-clean.js - Validates analysis settings
  • Customization status reporting in validation output

Why Direct YAML Editing?

• 🎯 Precision Control - Fine-tune AI behavior beyond what UI controls offer
• 🚀 Power User Features - Implement custom practice areas for niche legal domains
• 🔬 Experimentation - Test different system prompts and detection strategies
• 📚 Version Control - Configuration files can be tracked in Git for team collaboration
• 🏢 Enterprise Customization - Adapt Atticus to firm-specific practice areas and terminology

Configuration File Locations:

• Development: c:\JDAI\atticus\public\config\
• Production: %APPDATA%\atticus\config\ (Windows) after first edit
• Language-Specific Files: practices.en.yaml, practices.fr.yaml, practices.es.yaml (and same for advisory.*, analysis.*)
• Universal Files: providers.yaml (no language variants needed)

💡 Pro Tip: Use the expandable keyword viewer in the UI to understand detection patterns before editing. Make incremental changes and test with the validation scripts to ensure configuration integrity.

⚠️ Caution: Direct YAML editing is powerful but requires understanding of YAML syntax and JSON Schema validation. Always use the "Reset" buttons if configurations become corrupted. The factory configurations from https://jdai.ca/atticus/ are always available as a fallback.

Show only the models you use and optimize them for specific domains:

• Enable/disable models per provider to reduce clutter
• Domain specialization - Configure each model for:
  • ⚖️ Practice Areas (legal) - Optimized for legal queries
  • 📊 Advisory Areas (business) - Optimized for business consulting
  • 🔄 Both - Available for all queries (default)
• Smart filtering - Models automatically filter based on detected query type
• Real-time detection - Legal questions show legal-optimized models, business questions show business-optimized models
• Visual indicators - Clear feedback showing which domain is active
• Per-provider configuration
• Settings persist across sessions

Benefits:

• Faster model selection with fewer irrelevant options
• Better results by matching models to their strengths
• Cost optimization using appropriate models for each task

• Markdown rendering for formatted responses
• Conversation management - save, search, and organize chats
• Conversation search - Find conversations by title, content, or practice area
• Editable titles - Rename conversations for better organization
• Per-conversation model selection - choose different models for different tasks
• Real-time area detection - automatic practice area and advisory identification
• PDF export - download conversations with formatting
• Document upload - PDF, Word, images (multimodal models)
• Sidebar actions - Quick access to PDF export, search, and delete

Search & Organization:

• 🔍 Quick search dialog - Filter conversations by title or message content
• ✏️ Inline title editing - Click edit button to rename conversations
• 📁 Auto-save - All conversations persist locally
• ⌨️ Keyboard shortcuts - Enter to save, Escape to cancel

All conversations are stored locally on your device with full search capabilities, editable titles, and organized sidebar for easy access to your legal and business advisory history

• ✅ Local storage - API keys stored on your machine only
• ✅ Direct API calls - no intermediaries or third-party servers
• ✅ Full control - you own your data
• ✅ Secure credentials - encrypted local storage
• ✅ Offline resilience - work continues when internet/cloud services are down

Atticus works when the cloud doesn't. Unlike web-based AI tools that become completely unusable during outages, Atticus maintains critical functionality:

• 📚 Access Your Work - All conversations, research, and analysis remain accessible locally even when internet is down
• 📝 Continue Working - Review past conversations, search through history, export PDFs, and prepare queries offline
• 🔍 Search & Review - Full-text search across all conversations works without internet connectivity
• 💾 Local-First Architecture - All data stored on your machine; you're never locked out of your own work
• 📊 Cost Analysis - Review expense ledgers and token usage reports offline

Real-World Impact: During AWS outages (Dec 2021, Jun 2022, Dec 2022), Azure incidents (Jan 2023, Jul 2024), or Cloudflare disruptions (Jun 2022, Nov 2023), web-based AI tools become completely inaccessible. Atticus users continue reviewing research, preparing queries, and accessing critical legal analysis without interruption.

💡 Business Continuity: When you need to reference legal research during a pitch, court preparation, or client call, cloud outages can't block access to your work. Your data stays with you, not trapped in someone else's datacenter.

Atticus supports 3 languages with complete UI and configuration localization:

• English 🇬🇧 - Primary language with full support
• Français 🇫🇷 - Complete French interface and legal/business guidance
• Español 🇪🇸 - Complete Spanish interface and legal/business guidance

Features:

• Automatic language detection and switching
• Language-specific YAML configurations for practices, advisory, and analysis
• Locale-aware date, number, and currency formatting
• All 67 legal practice areas and 67 business advisory areas translated
• Complete settings UI in all 3 languages
• 2,300+ lines of professional translations

Architecture: The internationalization system includes:

• React Context Provider for language state (LanguageContext.tsx)
• Comprehensive translations file (translations.ts with 2,239 lines)
• Language-aware config loading (configLanguage.ts)
• Locale-specific formatting utilities (formatting.ts)

Configuration Files: Language-specific YAML files are available for localized guidance:

• practices.en.yaml, practices.fr.yaml, practices.es.yaml
• advisory.en.yaml, advisory.fr.yaml, advisory.es.yaml
• analysis.en.yaml, analysis.fr.yaml, analysis.es.yaml
• providers.yaml (universal, no language variants needed)

💡 Pro Tip: Switch languages in Settings to experience Atticus in French or Spanish - all legal and business guidance adapts to your selected language!

Atticus uses direct API access to AI providers rather than consumer web apps (ChatGPT, Claude.ai, etc.). This architectural decision provides critical privacy and enterprise benefits:

• ❌ Consumer Apps: Your conversations are stored on provider servers, used for model training (unless opted out), and subject to provider data policies
• ✅ API Endpoints: Zero data retention by default - providers don't store or train on your API data per enterprise agreements

Real Impact: Sensitive legal documents, financial data, trade secrets, and client information stay private. No risk of your proprietary strategies appearing in future model training.

Atticus automatically detects PII and sensitive information, providing real-time warnings and anonymization options to protect confidential data before sending to AI providers

• ❌ Consumer Apps: Consumer-grade terms of service, no SLAs, limited compliance certifications
• ✅ API Endpoints: Enterprise agreements with GDPR/HIPAA compliance options, SOC 2 Type II certifications, BAAs available

Real Impact: Meet regulatory requirements for handling confidential client data, PII, and protected health information.

• ❌ Consumer Apps: Limited visibility into data handling, hard to audit usage, conversations tied to personal accounts
• ✅ API Endpoints: Complete audit trails, usage tracking per API call, programmatic access control, local data storage

Real Impact: Full visibility for compliance audits, client billing justification, and internal security reviews.

• ❌ Consumer Apps: Subscription per user ($20-200/mo), rate limits, restricted features, limited model selection
• ✅ API Endpoints: Pay only for what you use, access to all models including latest releases, no rate limit surprises

Real Impact: Better economics for variable usage patterns, access to specialized models, no paying for unused seats.

• ❌ Consumer Apps: Manual copy-paste workflows, browser-based only, no automation, siloed from your tools
• ✅ API Endpoints: Desktop application integration, automation capabilities, multi-provider flexibility, offline data storage

Real Impact: Professional workflow integration, use multiple AI providers simultaneously, maintain local conversation history.

Bottom Line: For legal and business advisory work involving confidential information, API endpoints provide the privacy guarantees, compliance certifications, and data control that consumer apps cannot match.

Atticus provides enterprise-grade legal and business advisory capabilities:

• 10 providers with 60+ models integrated
• Latest models (as of November 2025): GPT-5.1, Claude 4.5 Sonnet, Gemini 2.5 Pro, Grok 4, Llama 3.3
• Context windows up to 2M tokens (xAI Grok 4)
• Ultra-fast inference up to 3000 tokens/sec (Cerebras)
• Vision capabilities across 6 providers
• Function calling support for structured outputs

• 67 specialized practice areas with expert system prompts
• 5,000+ keyword triggers for automatic detection
• Geographic coverage: 🇨🇦 80% Canada | 🇺🇸 92% US | 🇲🇽 70% Mexico | 🇪🇺 75% EU
• CUSMA/USMCA focus: Full cross-border coverage for North American operations
• Startup optimization: 90% coverage for entrepreneurial legal needs
• Confidence scoring with alternative area suggestions

• 67 advisory domains with consulting-grade guidance
• ~25,000 lines of expert advisory content (expanded coverage)
• 3,500+ keyword triggers for automatic detection
• 670 core competencies across all advisory areas (10 per area)
• World-class frameworks: McKinsey, BCG, Bain, Goldman Sachs, Deloitte methodologies
• Full MBA coverage: Strategy, finance, marketing, operations, HR, technology, specialized functions
• Industry verticals: Retail, healthcare, professional services, franchise, family business, nonprofit

• Legal practice areas: ~8,500 lines (practices.yaml)
• Business advisory areas: ~25,000 lines (advisory.yaml)
• Total expertise: 33,500+ lines of structured guidance
• Combined keywords: 8,500+ triggers for intelligent detection

• 4 major jurisdictions: Canada, United States, Mexico, European Union
• CUSMA/USMCA Region: Comprehensive North American free trade zone coverage
• Three-dimensional complexity analysis: International + cross-border + interprovincial/interstate
• Intelligent prompt modification based on jurisdiction selection
• Comparative analysis for multi-jurisdiction queries
• Internal complexity awareness: Provincial/state variations, trade barriers, conflicts of law
• Cross-border transaction and compliance support

• Node.js 18 or higher
• npm or yarn

Clone or navigate to the project directory:

cd atticus

Install dependencies:

npm install

Run in development mode:

npm run dev

Build for production:

npm run build

The built application will be in the release folder.

Atticus uses a template-based provider system with 9 pre-configured providers. Here's how to set them up:

1. Launch Atticus

2. Click the Settings button (⚙️) in the sidebar

3. Browse the four tabs:

   Providers Tab - Configure AI models:

   • OpenAI (GPT-5.4, GPT-5.2, GPT-5.1, GPT-5, GPT-5 Mini, GPT-5 Nano)
   • Anthropic (Claude 4.5/4.6 Sonnet, Claude 4.5 Haiku, Claude 4.5/4.6 Opus)
   • Google (Gemini 3/3.1 Pro, Gemini 3 Flash, Gemini 3.1 Flash Lite)
   • Azure OpenAI (GPT-4o, GPT-4o mini, GPT-4)
   • xAI (Grok 4, Grok 4.1 Fast Reasoning, Grok 4.1 Fast Non-Reasoning)
   • Mistral AI (Mistral Large 2, Mistral Medium, Mistral Small, Mixtral 8x22B, Mixtral 8x7B)
   • Cohere (Command R+, Command R, Command, Command Light)
   • Groq (Llama 3.3 70B, Llama 3.1 70B, Llama 3.1 8B, Mixtral 8x7B, Gemma 2 9B)
   • Perplexity (Sonar Large Online, Sonar Small Online, Sonar Large Chat, Sonar Small Chat)
   • Cerebras (GPT OSS 120B, Llama 3.3 70B, Llama 3.1 8B, Qwen 3 32B, Qwen 3 235B, ZAI GLM 4.6)

   Practice Areas Tab - View all 67 legal practice areas:

   • Expandable keyword lists (click "+X more")
   • Color-coded by category
   • Comprehensive coverage details

   Advisory Areas Tab - Explore all 67 business advisory areas:

   • Strategic planning, finance, marketing, operations
   • Technology, risk, sustainability, M&A
   • Government grants and product compliance
   • Expandable keyword lists

   About Tab - Learn about Atticus:

   • Mission and vision
   • Geographic coverage statistics
   • Privacy and security guarantees
   • Important legal disclaimers

4. For each provider you want to use:

   • Click "Get your [Provider] API Key →" link to obtain an API key
   • Paste your API key in the input field
   • Click Activate to configure the provider
   • Once activated, you can:
     • Select a default model from the dropdown
     • Enable/disable specific models using checkboxes (only checked models appear in conversation selectors)
     • Set as Active Provider to make it the default for new conversations

5. Configure Model Filtering (optional):

   • After activating a provider, use the "Available Models" checkboxes
   • Uncheck models you don't want to see in conversation model selectors
   • This reduces clutter and shows only your preferred models

6. Configure Domain Specialization (optional):

   • For each enabled model, choose where it should be used:
     • ⚖️ Practice - Only for legal queries
     • 📊 Advisory - Only for business queries
     • 🔄 Both - Available everywhere (default)
   • Models automatically filter based on detected query type
   • Example: Configure Claude for legal, o1 for business strategy, GPT-4o for both

Status Badges:

• 🟢 Configured - Provider has an API key and is ready to use
• 🟡 Active - Currently set as the default provider for new conversations

Example: Configuring Google Gemini models with domain specialization and selective model enabling

You'll need to obtain API keys from the providers you want to use:

• OpenAI: https://platform.openai.com/api-keys
• Anthropic: https://console.anthropic.com/settings/keys
• Google AI: https://makersuite.google.com/app/apikey
• Azure OpenAI: Through your Azure portal
• xAI: https://console.x.ai/
• Mistral AI: https://console.mistral.ai/
• Cohere: https://dashboard.cohere.com/api-keys
• Perplexity: https://www.perplexity.ai/settings/api
• Groq: https://console.groq.com/keys
• Cerebras: https://cloud.cerebras.ai/

Security Notes:

• API keys are stored locally in: %APPDATA%/atticus/user-config.json (Windows)
• Keys are never transmitted except to the respective AI provider
• You are billed directly by each provider based on your usage

1. Click New Conversation (➕) in the sidebar header
2. Type your legal question or business advisory request
3. Atticus will automatically detect the practice/advisory area and provide specialized assistance
4. Upload documents using the paperclip icon (for supported models)

1. Click the Search (🔍) button in the sidebar header
2. Type keywords to filter by:
   • Conversation title
   • Message content
   • Practice area or advisory area
3. Click any result to open that conversation
4. Press Escape or click outside to close

1. Hover over any conversation in the sidebar
2. Click the Edit (✏️) icon in the bottom-right corner
3. Type a new descriptive title
4. Press Enter to save or Escape to cancel
5. Conversations are auto-sorted with edited titles

1. Hover over a conversation in the sidebar
2. Click the document icon in the bottom-right corner
3. Choose where to save the PDF

• Legal and business advice: Always review AI-generated content with licensed professionals
• Specific questions: Use detailed, specific questions for better responses
• Document context: Upload relevant documents for context-aware assistance
• Multi-model validation: Use 2-3 models to cross-validate important answers
• Domain specialization: Configure models for specific domains (legal/business) to optimize results and reduce costs
• Keyword awareness: Click "Expand keywords" in Settings to understand detection triggers
• Organization: Rename conversations with descriptive titles for easy retrieval
• Search effectively: Use the search dialog to quickly find past conversations
• Security: Keep API keys secure and never share them
• Startup focus: Leverage startup-specific areas for founder, fundraising, and growth topics

Atticus uses keyword detection to identify both legal practice areas and business advisory areas. Here are some examples:

• Corporate: merger, acquisition, shareholder, securities, incorporation
• Litigation: lawsuit, complaint, motion, discovery, trial
• IP: patent, trademark, copyright, infringement, licensing
• Real Estate: property, deed, lease, zoning, mortgage
• Employment: discrimination, harassment, wrongful termination, FMLA
• Contracts: agreement, breach, indemnity, warranty, NDA
• Criminal: prosecution, defense, felony, arrest, Miranda
• Tax: IRS, audit, deduction, tax planning, capital gains
• Startup Law: founder agreement, vesting, SAFE note, 83(b) election, cap table
• Venture Capital: term sheet, liquidation preference, anti-dilution, Series A

Core Business (11 areas):

• Strategic Planning: competitive analysis, SWOT, business model canvas, strategic roadmap, Porter's Five Forces
• Financial Advisory: valuation, financial modeling, capital structure, M&A advisory, 409A valuation
• Marketing Strategy: go-to-market, brand positioning, customer acquisition, growth marketing, CAC/LTV
• Operations: lean manufacturing, Six Sigma, supply chain optimization, KPIs, process improvement
• HR & Organizational: talent acquisition, performance management, organizational design, culture, OKRs
• Digital Transformation: cloud migration, AI/ML adoption, digital strategy, innovation, microservices
• Risk Management: enterprise risk, business continuity, compliance frameworks, resilience, COSO ERM
• Sustainability: ESG reporting, carbon footprint, sustainable operations, impact measurement, GRI
• M&A Advisory: post-merger integration, due diligence, synergy realization, cultural integration
• Government Grants: SR&ED tax credits, SBIR grants, Horizon Europe, innovation funding, R&D credits
• Product Legal: privacy-by-design, WCAG accessibility, GDPR compliance, open source licensing

Growth & Innovation (11 areas):

• Entrepreneurship: lean startup, MVP, product-market fit, founder vesting, pitch deck, seed funding
• International: market entry, CUSMA/USMCA, foreign market, localization, cross-border trade
• Analytics: business intelligence, data analytics, KPI dashboard, A/B testing, predictive modeling
• Supply Chain: supply chain optimization, inventory management, warehousing, 3PL, logistics
• Sales & RevOps: pipeline management, RevOps, CRM, sales enablement, MEDDIC, sales methodology
• Governance: board of directors, fiduciary duty, audit committee, D&O insurance, corporate governance
• Innovation & R&D: innovation management, R&D portfolio, design thinking, patent strategy
• Customer Experience: journey mapping, NPS, customer satisfaction, service design, CX strategy
• Org Design: organizational design, change management, transformation, operating model
• Negotiations: BATNA, deal structuring, contract negotiation, value creation, negotiation tactics
• Crisis: crisis management, business continuity, disaster recovery, incident response

Specialized Functions (22 areas):

• Brand: brand positioning, brand architecture, brand equity, rebranding, brand strategy
• Pricing: pricing optimization, revenue management, value-based pricing, price elasticity
• Product: product roadmap, product-market fit, product lifecycle, feature prioritization
• Process: lean, Six Sigma, DMAIC, continuous improvement, kaizen, process reengineering
• Treasury: capital structure, cash management, hedging, treasury operations, corporate finance
• Turnaround: restructuring, financial distress, bankruptcy, debt restructuring, turnaround
• CI: competitive intelligence, competitor analysis, market intelligence, war gaming
• BD: business development, strategic partnerships, channel partnerships, alliance management
• IP: patent strategy, IP portfolio, technology licensing, patent filing, IP commercialization
• Quality: TQM, ISO 9001, quality assurance, Six Sigma quality, quality management systems
• Real Estate: workplace strategy, facilities management, lease negotiation, corporate real estate
• Tax: tax planning, transfer pricing, international tax, R&D tax credits, tax optimization
• Insurance: risk transfer, captive insurance, claims management, D&O insurance, risk financing
• IR: equity story, analyst coverage, shareholder communications, investor relations strategy
• Public Affairs: government relations, lobbying, regulatory advocacy, policy positions
• Franchise: franchising, franchise development, franchisee selection, royalty structure
• Family: succession planning, family governance, generational transition, family business
• Nonprofit: impact measurement, fundraising, grant writing, social enterprise, mission-driven
• Retail: merchandising, store operations, omnichannel, same-store sales, retail analytics
• Healthcare: practice management, value-based care, revenue cycle, EHR, HIPAA compliance
• Prof Services: consulting firm, leverage model, partner compensation, utilization, firm economics
• Tech Transfer: technology commercialization, licensing, spinout, proof of concept, IP licensing

Total Coverage: 67 legal areas + 67 advisory areas = 134 specialized domains with 8,500+ detection keywords

Modern stack with modular design:

• Frontend: React 18 + TypeScript + Tailwind CSS
• Desktop: Electron 28 (Node.js)
• State Management: Zustand with conversation-scoped persistence
• AI Integration: Multi-provider template system with parallel execution
• Practice Area System: Modular keyword-based detection (67 legal + 67 advisory = 134 specializations)
• Jurisdiction System: Multi-jurisdictional analysis (CA, US, MX, EU) with interprovincial/interstate complexity
• Multi-Model Architecture: Parallel API execution via Promise.all()
• Model Domain Specialization: Real-time query detection with smart model filtering (practice/advisory/both)
• Configuration: YAML-based with remote updates and JSON Schema validation
• PDF Generation: jsPDF with custom legal formatting
• Markdown Rendering: react-markdown with syntax highlighting
• Build Tool: Vite 5.4

atticus/
├── public/
│   └── config/          # YAML configuration files
│       ├── providers.yaml      # AI provider configurations
│       ├── practices.yaml      # Legal practice areas (67 areas)
│       └── advisory.yaml       # Business advisory areas (67 areas)
├── src/
│   ├── electron/         # Electron main process
│   │   ├── main.ts       # Main process entry point
│   │   └── preload.ts    # Preload script (IPC bridge)
│   ├── components/       # React components
│   │   ├── ChatWindow.tsx       # Main chat interface
│   │   ├── Sidebar.tsx          # Conversation list with search
│   │   └── Settings.tsx         # 4-tab settings (Providers, Practice, Advisory, About)
│   ├── config/          # Configuration files
│   │   ├── jurisdictions.ts     # Jurisdiction configs (CA, US, EU)
│   │   └── providerTemplates.ts # AI provider templates
│   ├── modules/         # Feature modules
│   │   └── practiceArea/        # Practice area detection module
│   │       ├── definitions.ts   # Practice area definitions (67)
│   │       ├── detector.ts      # Auto-detection logic
│   │       ├── examples.ts      # Example prompts per area
│   │       ├── types.ts         # Type definitions
│   │       ├── PracticeAreaManager.ts  # Main manager class
│   │       └── index.ts         # Module exports
│   ├── services/        # API services
│   │   ├── api.ts              # AI API integration service
│   │   ├── practiceLoader.ts   # Load legal practice areas from YAML
│   │   └── advisoryLoader.ts   # Load advisory areas from YAML
│   ├── store/           # Zustand state management
│   │   └── index.ts             # Global app state with conversation management
│   ├── types/           # TypeScript type definitions
│   │   └── index.ts             # Global types (Conversation, Jurisdiction, etc.)
│   ├── i18n/            # ✨ Internationalization (NEW in v0.9.20)
│   │   ├── LanguageContext.tsx   # React context for language state
│   │   ├── translations.ts       # 2,239 lines of EN/FR/ES translations
│   │   └── index.ts              # i18n exports and utilities
│   ├── utils/           # Utility functions
│   │   ├── configLanguage.ts     # ✨ Language-aware config loading
│   │   └── formatting.ts         # ✨ Locale-aware formatting utilities
│   ├── App.tsx          # Main React component
│   ├── main.tsx         # React entry point
│   └── index.css        # Global styles (Tailwind)
├── docs/                # Comprehensive documentation
│   └── ...
├── dist-electron/       # Built Electron files
├── release/             # Production builds
├── .vscode/            # VS Code settings
├── package.json         # Dependencies and scripts
├── vite.config.ts       # Vite build configuration
├── tailwind.config.js   # Tailwind CSS configuration
├── tsconfig.json        # TypeScript configuration
└── LICENSE              # Dual license (Commercial + Apache 2.0)

• ToDo

⚠️ Important Security Information

• Your API keys are stored locally in: %APPDATA%/atticus/user-config.json (Windows)
• Conversations are saved in: %APPDATA%/atticus/conversations/
• Never share your configuration files
• Each provider bills you directly based on usage
• Review your API provider's usage dashboard regularly

Atticus is an AI assistant tool and NOT a substitute for professional legal or business advice.

• Always consult with a licensed attorney for legal matters
• AI responses may contain errors or outdated information
• Do not rely solely on AI for legal decisions
• Verify all information with current laws and regulations
• Attorney-client privilege does NOT apply to AI conversations

• npm run dev - Start Vite dev server
• npm run build - Build for production

• Verify your API key is correct
• Check your internet connection
• Ensure the endpoint URL is correct (for custom providers)
• Check the provider's status page for outages

• Ensure the model supports multimodal/RAG capabilities
• Check file size limits (varies by provider)
• Verify the file format is supported

• If a model times out, its response will show an error badge
• Reduce the number of selected models if responses are slow
• Check individual provider status if specific models fail

This is a sophisticated AI query tool. For issues or improvements, please contact the development team.

Atticus is built with a privacy-first architecture and transparent risk disclosure:

Local-First Architecture: Your data stays on your device. No telemetry, no cloud storage, no tracking.

Key Privacy Features:

• ✅ All conversations stored locally on your device
• ✅ API keys encrypted using OS-level keychain
• ✅ No data collection or telemetry by Atticus
• ✅ Zero tracking or analytics
• ✅ Complete user control over data deletion
• ⚠️ Data sent to third-party AI providers (user-selected)

Privacy Rating: 7.5/10 (Strong with Third-Party Dependencies)

📄 Full Privacy Policy & Data Protection Assessment

Atticus is designed with responsible AI principles at its core:

EU AI Ethics Guidelines Compliance:

• ✅ Human Agency & Oversight - Users maintain full control, persistent disclaimers
• ✅ Privacy & Data Governance - Local-first storage, user data ownership
• ✅ Transparency - Clear model attribution, explicit AI limitations
• ⚠️ Technical Robustness - Dependent on third-party AI providers
• ⚠️ Diversity & Fairness - Bias mitigation through multi-model querying

Core Ethical Commitment: Transparency, user autonomy, and explicit acknowledgment of limitations take precedence over feature expansion.

🤖 Full Ethical AI Analysis & Assessment

Atticus maintains high code quality standards with continuous technical debt remediation:

Type Safety:

• ✅ Strong TypeScript typing throughout codebase
• ✅ Eliminated any types in favor of proper interfaces
• ✅ Comprehensive type definitions for all data structures
• ✅ Proper interface definitions for legal/advisory areas (LegalPracticeArea[])
• ✅ Type-safe compliance violation structures

Error Handling:

• ✅ Structured logging with createLogger() utility
• ✅ No silent error swallowing - all catch blocks documented
• ✅ Consistent error propagation patterns
• ✅ Comprehensive audit logging for security events

Code Organization:

• ✅ Clean separation of concerns (UI, services, types, utilities)
• ✅ Consistent naming conventions and patterns
• ✅ Well-documented public APIs
• ✅ Regular refactoring to reduce cognitive complexity

Recent Improvements (v0.9.19):

• Replaced all console.log calls with structured logger
• Fixed empty catch blocks with explanatory comments
• Added ComplianceViolation interface for type safety
• Improved Attachment and FileUploadResult type definitions
• Enhanced error handling in Settings and LogViewer components

CRITICAL: Atticus provides information, not advice. Understanding the risks is essential:

Major Risk Categories:

1. Unauthorized Practice of Law - AI outputs are not legal advice (Residual Risk: HIGH)
2. Accuracy & Reliability - AI hallucinations and outdated information (Residual Risk: MEDIUM-HIGH)
3. Data Privacy - Third-party AI provider exposure (Residual Risk: MEDIUM)
4. Professional Liability - No insurance coverage for AI-generated outputs (Residual Risk: HIGH)
5. Security Vulnerabilities - API key theft, supply chain risks (Residual Risk: MEDIUM)

Required Actions:

• ⚠️ Always verify AI outputs with licensed professionals
• ⚠️ Never rely solely on Atticus for high-stakes decisions
• ⚠️ Understand jurisdiction - Laws vary significantly by location
• ⚠️ Review third-party AI providers - Each has different privacy policies
• ⚠️ Maintain confidentiality - Don't share sensitive client information

⚠️ Full Risk Assessment & Mitigation Framework

Dual licensed under:

• Commercial License - For proprietary/commercial use (see LICENSE-COMMERCIAL.md)
• Apache 2.0 License - For open source use (see LICENSE-APACHE.md)

See LICENSE file for full details.

✨ Major Feature: Internationalization Support

• 🌍 Multi-Language Support - Full UI and configuration support for English, Français, and Español
• 📝 2,239 Lines of Translations - Complete professional translations across all interface elements
• 🎯 Language-Specific Configurations - Dedicated YAML files for each language:
  • practices.en.yaml, practices.fr.yaml, practices.es.yaml - Legal practice areas in 3 languages
  • advisory.en.yaml, advisory.fr.yaml, advisory.es.yaml - Business advisory areas in 3 languages
  • analysis.en.yaml, analysis.fr.yaml, analysis.es.yaml - Analysis configurations in 3 languages
• 🔧 New i18n Architecture:
  • React Context Provider (LanguageContext.tsx) for language state management
  • Comprehensive translations file (translations.ts with 2,239 lines)
  • Language-aware config loading (configLanguage.ts)
  • Locale-specific formatting utilities (formatting.ts) for dates, numbers, currencies
• ✅ Locale-Aware Formatting - Automatic date, number, and currency formatting per language
• 🇨🇦🇫🇷🇪🇸 All 134 Specialized Domains Translated - All 67 legal practice areas and 67 business advisory areas available in 3 languages

📦 AI Model Updates

• Total Model Count: Increased from 43 to 45 models across 10 providers
• OpenAI (6 models): Added GPT-5.4 (1M context, ultimate performance) and GPT-5.2 (latest advanced model)
• Anthropic (5 models): Added Claude 4.6 Opus and Claude 4.6 Sonnet (latest generation)
• Google (4 models): Updated to Gemini 3/3.1 series (newer generation models)
• xAI (3 models): Updated Grok Fast variants to 4.1 generation
• Cerebras (6 models): Updated Qwen model naming to "Qwen 3 235B Instruct"

🎨 Model Configuration Enhancements

• Domain specialization now supports all 45 models
• Language-aware model recommendations based on query type
• Updated provider status badges with multi-language support
• Improved model display names for international users

📚 Documentation Improvements

• Complete internationalization documentation added to README
• Architecture section updated with i18n module details
• Configuration documentation expanded to explain language-specific files
• Model tables updated with latest versions and capabilities

🔒 Privacy & Compliance

• No changes to data privacy model - remains local-first
• All translations follow EU AI Act transparency requirements
• Localization maintains PII protection across all languages

⚙️ Technical Enhancements

• Improved language detection and switching
• Optimized translation performance
• Enhanced config loading for multi-language support
• Maintains backward compatibility with existing configurations

Known Limitations:

• Automatic language detection based on OS locale
• Some specialized legal/business terminology may have multiple valid translations
• Manual language switching available in Settings

Code Quality & Technical Debt Improvements:

• ✅ Enhanced type safety across codebase (eliminated any[] types)
• ✅ Replaced console.log with structured logging in all components
• ✅ Fixed empty catch blocks with proper error documentation
• ✅ Added proper TypeScript interfaces for legal/advisory areas
• ✅ Improved error handling in Settings and LogViewer
• ✅ Created ComplianceViolation interface for GDPR/HIPAA/CCPA tracking

UI Improvements:

• ✅ Simplified top buttons to icons-only for cleaner interface
• ✅ Log Viewer now displays full date and time (not just time)
• ✅ Enhanced button tooltips for better usability

Architecture:

• ✅ Consistent logging patterns throughout application
• ✅ Better type definitions for file uploads and attachments
• ✅ Improved code maintainability and readability