Prerequisites:
- Node.js (v22+)
- Docker
- Docker Compose
Then install code generation dependencies:
npm installStart the application:
# Edit docker-compose.yml line 135: Replace 'your_api_token' with your actual OpenWebUI API token
./start-dev.shAccess the application:
- Client (Angular frontend) at http://localhost:4200
- API Gateway at http://localhost:8080
- Full stack running with real backend services
Note: Before running, edit
docker-compose.ymlline 135 and replaceyour_api_tokenwith your actual OpenWebUI API token.
An interactive co-creation assistant that turns a vague event idea into a polished concept—fast. The planner drags in any relevant documents (past reports, industry papers, brand decks). The AI digests that material, asks follow-up questions until all gaps are filled, and delivers a full concept package: theme, format, agenda outline, speaker & sponsor suggestions, and ticket-price ideas, exportable as PDF and JSON. Because the AI does the heavy lifting—researching, matching, structuring—it cuts days of manual work to minutes, keeps all context in one place, and lets the planner iterate (“add a workshop,” “make it hybrid”) with a single prompt.
- Event-agency planners who prepare multiple client proposals.
- Corporate planners who must respect internal policies yet still wow stakeholders.
- LangChain + Weaviate RAG over user-supplied documents—no external scraping—so every suggestion is grounded in the customer’s own industry context.
- Adaptive dialogue powered by an LLM that follows nine standard conception steps, probes for missing details, remembers answers, and supports unlimited refinements.
- Creative synthesis prompt chains craft themes, agendas, and curated speaker/sponsor lists that reflect both uploaded content and the evolving conversation.
- Continuous learning—each new debrief or guideline embedded today improves tomorrow’s concepts automatically.
Co-create a fresh pitch – The planner uploads last year’s debrief and a market white paper, then says “Target 300 attendees, hybrid preferred.” The AI summarises the docs, asks two clarifiers (duration, networking preference), and returns a one-day concept. The planner adds, “Include a hands-on workshop and make the theme more visionary.” The AI revises the agenda and title, then offers a ready-to-share PDF.
Compliance-aware brainstorm – A corporate planner supplies the company’s policy handbook and audience personas. The AI filters speaker suggestions to fit policy, proposes an online format for global reach, and crafts sponsor packages aligned with brand guidelines. When the planner asks, “Shorten it to a half-day and add a panel,” the AI updates the concept instantly.
Learning loop – After an event, the planner uploads debrief notes (“need stronger networking, ticket price felt high”). Next time, the AI automatically proposes an interactive networking segment and adjusted ticket tiers, then asks, “Anything else you’d like to refine?”—keeping the focus on creative improvement instead of administrative grind.
The system follows a modular microservice architecture with clearly separated concerns across backend services, a modern web frontend, and scalable data infrastructure.
| Layer | Technology | Purpose |
|---|---|---|
| API Gateway | Spring Boot 3 | JWT authentication, routing, OpenAPI docs |
| User Service | Spring Boot 3 | User management, authentication, JWT tokens |
| Concept Service | Spring Boot 3 | Event concept CRUD, PDF generation, AI integration |
| GenAI Service | Python 3.12 + LangChain | Document ingestion, RAG pipeline, content creation |
| Web Client | Angular 19 | Chat UI, adaptive flow, PDF viewer |
| Relational DB | PostgreSQL | Stores users, concepts, agenda items, speakers |
| Vector DB | Weaviate | Embeddings for trends & document chunks |
| Object Store | MinIO | Uploaded files and generated PDFs |
| Observability | Prometheus + Grafana | Metrics and dashboards |
| Orchestration | Docker + Kubernetes | Containerization and scalable deployment |
This diagram shows the key objects and their relationships as identified during analysis.
This diagram illustrates the main interactions between users and the system.
This diagram provides a high-level overview of the system’s components and their interactions.
The project is split into several main directories:
/api: OpenAPI specifications (single source of truth)/client: Angular 19 frontend/gateway: API Gateway (Spring Boot) - Routing, JWT validation, OpenAPI docs/user-svc: User Service (Spring Boot) - Authentication, user management/concept-svc: Concept Service (Spring Boot) - Event concept CRUD, PDF export/genai-svc: GenAI Service (Python/Flask/LangChain) - Document processing, RAG pipeline
This project follows an API-first development approach. All API changes start with updating the OpenAPI specifications in the /api directory.
/api # API specifications (single source of truth)
├── gateway.yaml # API Gateway specification
├── user-service.yaml # User Service specification
├── concept-service.yaml # Concept Service specification
├── genai-service.yaml # GenAI Service specification
├── scripts/ # Code generation scripts
└── README.md # API documentation
- Local Development: Run
./start-docs.shto launch Swagger UI for all services - Documentation Hub:
docs/index.htmlprovides unified access to all API docs - Individual Specs: Each service OpenAPI spec can be viewed independently
- Design: Update OpenAPI specifications first
- Validate: Automated linting and specification validation
- Generate: Code generation from specifications
- Test: API contract testing and validation
- Deploy: Automated CI/CD pipeline deployment
Engineering Process Evidence:
- ✅ Documented Requirements: Clear API contracts in
/apidirectory - ✅ Architecture Models: UML diagrams (Analysis Object Model, Use Case, Component)
- ✅ API-First Workflow: Specifications drive implementation
- ✅ Automated Validation: Pre-commit hooks and CI pipeline checks
- Unit Tests: JUnit (Java), pytest (Python), Jasmine (Angular)
- Integration Tests: Database operations, API endpoints, service interactions
- CI Pipeline: GitHub Actions runs tests automatically on PR and merge
- Coverage Tools: JaCoCo (Java), pytest-cov (Python), ng test (Angular)
- On Pull Request: Lint, unit tests, API validation
- On Merge: Full test suite, integration tests, deployment
- API-First Validation: OpenAPI spec validation and code generation verification
See Testing Strategy for detailed coverage and methodology.
- Node.js (v22+), Java JDK 21+, Docker, Git
- All dependencies auto-installed by setup scripts
# Development (default)
./start-dev.sh
# Staging environment
CLIENT_BUILD_ENV=staging docker-compose up --build
# Production environment
CLIENT_BUILD_ENV=production docker-compose up --buildFor individual service development, see respective service READMEs:
- Gateway Service -
./gradlew bootRun --args='--spring.profiles.active=local' - User Service -
./gradlew bootRun - Concept Service -
./gradlew bootRun - GenAI Service -
python app.py - Client App -
ng serve
- API Specifications - OpenAPI specs and API-first development
- API Documentation Hub - Interactive Swagger UI for all services
- Infrastructure & Deployment - Terraform, Ansible, CI/CD setup
- Monitoring - Prometheus, Grafana, alerting
- Concept Service - Event concepts, PDF generation
- User Service - User management, authentication, JWT tokens
- GenAI Service - LangChain integration, AI-powered features
- Gateway Service - API routing, JWT authentication
- Client App - Angular frontend, environment configuration
- Testing Strategy - Test coverage, CI automation, GenAI testing
- GenAI Testing - Conversation history and concept suggestion tests
- Helm Deployment - Kubernetes deployment guide
Based on CODEOWNERS:
- @lenni108: API specs (shared), Angular client, concept service, infrastructure, documentation
- @sfdamm: Gateway, GenAI service, CI/CD workflows, Helm charts, API specs (shared), documentation
- @ClaudiaDuenzinger: User service, monitoring stack, documentation