Myotis
Italian civil and criminal files as a graph: documents, people, articles, precedents. An assistant that answers against that graph, not against the open web.
Case graphs from your fascicolo
Myotis reconstructs relationships between source documents, cited penal and civil articles, persons, and precedents, so you can explore a criminal or civil case visually, query paths in the graph, and work with an AI assistant that understands the full network.
Legal AI Investigations
Interactive case graph · Fascicolo Verdi (fictional demo)
Entity types
Fictional demo data · Hover nodes to highlight connections · Toggle full graph or data panel
Built for Italian legal studios
Myotis is a standalone version of Inputless Analytics built for Legal AI Investigations. Italian legal professionals upload civil and criminal appeals, indictments, and case files; the platform extracts entities, cites articles from the Codice Penale and civil codes, links persons, and renders an interactive investigation graph with explore tools and a graph-aware AI assistant.
Originally developed by Giovanni Errico and Leonardo Trisolini (2020-2022). Myotis is document intelligence for Italian legal work.
Built for Italian procedure
What it does
Built for Italian civil and criminal practice.
Case graph
Appeals, filings, people, articles, and precedents as one graph you can walk. Civil and criminal.
Document pass
Pulls parties, dates, cited articles, and holdings from contracts and fascicoli. A lawyer still reads the source.
Graph assistant
Ask against the case graph, in Italian legal vocabulary. Answers cite nodes. A lawyer still reads the source.
Search the store you have
Precedents and filings from the documents already in the matter. Seconds, from the store you already have.
Matter status
Deadlines, outcomes, and load across the studio. Useful for staffing. Not a management consultant deck.
Privilege and GDPR
Tenants stay separate. Access is logged. Client files stay under the studio’s rules.
Patterns from past matters
Flags similar outcomes and clauses from history. A prediction is a hint, not a forecast you can bill on.
What changes in the studio
Less first-pass reading
Entities and citations are extracted before the lawyer opens the PDF. Review still happens. It starts later in the pile.
Repetitive work off the clock
Intake, indexing, and first-pass extraction. Lawyers keep judgment, hearings, and the client.
Find the article that matters
Precedents and cited norms sit on the graph. You are not grepping the fascicolo at midnight.
Better Client Service
Live case updates. Document analysis that a lawyer can check.
Technical Stack
Docker, the graph store, and the models you already run in the studio.
Backend
- Django 3.2 - Web framework
- PostgreSQL - Primary database
- ArangoDB - Graph database for relationships
- Redis - Caching and message broker
- Celery - Asynchronous task processing
- spaCy - Natural language processing
- Language model on the case graph
Frontend
- Webpack - Asset bundling
- Bootstrap 5 - UI framework
- JavaScript - Interactive components
- WebSocket - Real-time communication
Infrastructure
- Docker - Containerization
- Nginx - Reverse proxy and load balancer
- Prometheus - Monitoring and metrics
- Grafana - Analytics dashboards
- ClamAV - Antivirus scanning
Security & Compliance
Data Protection
- GDPR-aligned data handling
- Secure multi-tenant architecture
- Client confidentiality protection
Audit & Recovery
- Complete activity logging and tracking
- Automated backup and recovery
- Antivirus scanning with ClamAV
Quick Start
Clone Repository
Clone the Inputless Analytics Standalone repository from GitHub
Docker Setup
Ensure Docker & Docker Compose are installed. Recommended: 8GB+ RAM, 50GB+ disk space
Configure Environment
Set up environment variables including OpenAI API key for GPT-4 integration
Deploy
Run docker-compose to start the development or production environment
Access Application
Access the web interface at http://localhost:3000 and configure your legal document database