# Theka — DAM with built-in AI Knowledge Layer

> Theka is a Digital Asset Management (DAM) platform with a built-in AI Knowledge Layer (Wiki AI / WikiLM) for enterprise document management.

## What is Theka

Theka is a multi-tenant SaaS platform that centralizes company files (contracts, data sheets, images, manuals) into one organized, versioned and protected system. It combines traditional DAM capabilities with an AI Knowledge Layer that auto-generates topic wikis, a navigable topic graph and a RAG chat with verifiable citations.

## Differentiators

- **Knowledge Layer (Wiki AI / WikiLM)**: The AI extracts recurring topics and atomic claims from documents, generates auto-updated wiki pages per topic, builds a navigable topic graph and feeds a RAG chat that always cites source documents.
- **Per-tenant pricing, not per-user**: Flat monthly fee per organization. Adding users does not increase costs (within tier caps).
- **BYO Storage**: Bring your own S3 / Azure Blob / SharePoint / MinIO / SFTP / Backblaze / DigitalOcean. Documents stay in your cloud, in your jurisdiction.
- **BYO AI**: Bring your own OpenAI / Anthropic / Gemini / Mistral API key. AI tokens are billed directly by the provider, not by Theka.
- **On-premise deployment available** for organizations with sovereignty requirements (healthcare, finance, public administration, defense).
- **Made in Italy**: Italian invoicing, Italian-language support, EU/Italian GDPR compliance.

## Core Features

- **Digital Asset Management (DAM)**: Upload, organize and manage documents by category, department, project and entity. Full version control with clear statuses (draft, review, approved, final, signed, archived). Download with audit trail.
- **Wiki AI / WikiLM (Knowledge Layer)**: Auto-generated wiki pages per recurring topic, with synthetic summary and atomic claims, each with [N] citations linking back to source documents. Updated incrementally as the knowledge base grows.
- **Topic Graph**: Force-directed graph of topics and their relationships, navigable like Obsidian. Topics can be promoted to Business Entities (clients, suppliers, products).
- **AI Chat (RAG)**: Chat interface powered by Retrieval-Augmented Generation that answers natural-language questions citing source documents. Uses tool-calling to search documents, query topics, and read document content.
- **AI Document Analysis**: Configurable AI tasks per category — metadata extraction, custom field auto-fill, summarization, OCR, embedding, knowledge synthesis. Provider-agnostic (OpenAI, Anthropic, Gemini, Mistral as swappable connectors).
- **Topic-Aware Search**: Three search modes combined — structured filters, full-text search, semantic search with vector embeddings (pgvector), plus topic and claim search.
- **Master Data**: Organizations, products, assets and other business entities with custom fields (text, number, currency, date, boolean, select), image slots, and unique external codes for ERP integration.
- **Image Variants & Slots**: Multi-size thumbnail generation. Named image positions for products and entities (e.g. "Front View", "Side View"). Configurable per asset/entity type.
- **Approval Workflows**: Configurable multi-stage approval processes with role-based permissions. Digital signature support, full audit on every transition.
- **Protocol Assignment**: Permanent, sequential protocol numbers for regulatory compliance. Atomic and thread-safe.
- **Multi-Tenant Architecture**: Complete data isolation between organizations. Tenant suspension with configurable reasons and scheduled reactivation.
- **Role-Based Access Control**: Three-level permissions — organization roles, category-level access, and per-document controls. Secure sharing links with expiration.
- **Audit Logging**: Every action is tracked — who viewed, modified, downloaded or approved a document, including AI chat queries, when and from where. Immutable.
- **17+ Built-in Connectors**: Storage (S3, Azure, SharePoint, MinIO, SFTP, Backblaze, DigitalOcean), AI (OpenAI, Anthropic, Gemini, Mistral), SSO (Microsoft Entra, Google), Search (Elasticsearch, OpenSearch), ERP integrations.
- **Notification System**: In-app, push and email notifications with per-category preferences and master toggle.
- **Browse Navigation**: Hierarchical browsing by categories, projects, entity types and entities. Assets visible at every level.

## Architecture

- **Frontend**: Next.js 16 (React, App Router) with TypeScript, Tailwind CSS, next-intl for i18n (Italian/English)
- **Backend**: Django REST Framework with PostgreSQL and pgvector
- **AI/ML**: Vector embeddings for semantic search, LLM integration (Anthropic, OpenAI, Google) for document analysis and RAG chat
- **Storage**: S3-compatible object storage with per-tenant isolation
- **Authentication**: JWT with refresh tokens, CSRF protection, optional MFA (TOTP), OAuth (Google, Microsoft)

## Data Privacy & Security

- **Tenant Isolation**: Each organization is a completely separate environment. Data is never shared between tenants.
- **Data Ownership**: Documents remain in the owner's repository with full control. Export or delete everything at any time.
- **AI is Optional**: AI features are opt-in, enabled by the organization administrator. No data is sent to third parties without explicit choice.
- **GDPR Compliant**: Full compliance with EU GDPR, ePrivacy Directive and Italian Privacy Code.
- **Security Headers**: HSTS, CSP, COOP, X-Frame-Options, X-Content-Type-Options, Trusted Types, strict Referrer-Policy.

## Target Users

- **Manufacturing**: Technical data sheets, certifications, supplier contracts — organized and accessible by department with granular permissions.
- **Services**: Complete project files with contracts, attachments and communications. Versioning and collaboration directly on documents.
- **Multi-site organizations**: Centralized system with controlled access per location. Each branch sees only what concerns them.
- **E-commerce**: Product image management with named slots (front, side, detail, ambient), variant generation, and entity-based organization.

## Public Landing Pages

- Home: https://theka-doc.com
- Asset Management (DAM core): https://theka-doc.com/asset
- Wiki AI / Knowledge Layer (WikiLM, topic graph, citations): https://theka-doc.com/wiki-ai
- Configurable AI (per-category tasks, BYO providers, cost monitoring): https://theka-doc.com/intelligenza-artificiale
- Integrations & Connectors (storage, AI, SSO, search, webhooks): https://theka-doc.com/integrazioni
- Security & Governance (multi-tenant, RBAC, audit log, MFA, SSO): https://theka-doc.com/sicurezza
- Pricing (per-tenant, BYO storage, BYO AI, on-premise): https://theka-doc.com/prezzi

## Support & Legal

- FAQ: https://theka-doc.com/support/faq
- Guides: https://theka-doc.com/support/guides
- Contact: https://theka-doc.com/support/contact
- Release notes: https://theka-doc.com/support/release-notes
- Privacy Policy: https://theka-doc.com/legal/privacy-policy
- Terms of Service: https://theka-doc.com/legal/terms-of-service
- Cookie Policy: https://theka-doc.com/legal/cookie-policy
- Full documentation: https://theka-doc.com/llms-full.txt

## Technical Details for AI Agents

- The platform serves content in Italian (default) and English (paths /it/* and /en/*; root path resolves to Italian)
- Public landing pages: home + 6 deep pages (asset, wiki-ai, intelligenza-artificiale, integrazioni, sicurezza, prezzi)
- Public support: FAQ, guides, contact, about, release notes
- Public legal: privacy policy, terms of service, cookie policy
- Authenticated pages require tenant context via subdomain (e.g. acme.example.com)
- API follows REST conventions at /api/v1/
- All dates are ISO 8601, all text is UTF-8
- Sitemap available at https://theka-doc.com/sitemap.xml
- robots.txt available at https://theka-doc.com/robots.txt