docs: Add system architecture diagram and fix README inaccuracies

- Add comprehensive system architecture diagram (graph.png) to README
- Rename section from 'MCP Architecture' to 'System Architecture' for broader scope
- Fix incorrect claims about batch processing, GPU acceleration, and zero-shot classification
- Correct output formats to reflect actual implementation (JSON only)
- Update repository URL and example files to match actual codebase
- Remove inaccurate claims about multilingual workflows and batch processing
- Add missing example file (communicator_email_tester.py) to documentation

These changes ensure README accurately reflects the actual capabilities and
implementation of the Sum-It-Up Agent system.

Author: iosifidisvasileios

README.md

@@ -48,7 +48,7 @@ The project is designed around **independent services** with clear boundaries, s
 ### 1) Install
 
 ```bash
-git clone https://github.com/yourusername/sum-it-up-agent.git
+git clone https://github.com/iosifidisvasileios/sum-it-up-agent.git
 cd sum-it-up-agent
 poetry install
 ```
@@ -86,7 +86,7 @@ python -m src.sum_it_up_agent.app /path/to/audio.mp4 "summarize this meeting and
 At runtime, the agent performs a pipeline like:
 
 1. **Ingest & normalize** audio/video
-2. **(Optional) diarization** for speaker separation
+2. **Speaker diarization** for speaker separation
 3. **Transcription** (Whisper-family models)
 4. **Meeting-type classification** (zero-shot / ensembles)
 5. **Instruction-aware summarization** using meeting templates + user instructions
@@ -96,7 +96,9 @@ Outputs can be saved as structured artifacts (e.g., JSON) for downstream workflo
 
 ---
 
-## MCP Architecture
+## System Architecture
+
+![System Architecture](graph.png)
 
 ```
 Sum-It-Up App (Singleton)
@@ -122,19 +124,18 @@ Sum-It-Up App (Singleton)
 ### Audio processing
 - Multi-format input (MP3, MP4, WAV, M4A, FLAC, …)
 - Optional speaker diarization
-- High-quality transcription with GPU acceleration where available
-- Batch-friendly workflows (parallelizable stages)
+- High-quality transcription with Whisper models
 
 ### Topic classification
-- Zero-shot meeting type detection (no training data required)
+- Meeting type detection using ensemble methods
 - Confidence scoring / uncertainty-aware output
 - Support for common meeting classes (planning, retro, interview, support call, etc.)
 
 ### Summarization
 - File-backed prompt templates for meeting-specific structure
 - Multiple summary types (standard, action items, decisions, key points, executive, …)
 - Multi-provider backends (hosted and local)
-- Output formats: JSON, TXT, CSV (and extensible)
+- Output formats: JSON (and extensible)
 
 ### Communication
 - **Email delivery** with HTML-formatted summaries
@@ -269,13 +270,14 @@ python -m unittest discover -s tests
 - `tests/examples/summarizer_examples.py`
 - `tests/examples/topic_classification_examples.py`
 - `tests/examples/prompt_parser_example.py`
+- `tests/examples/communicator_email_tester.py`
 
 ---
 
 ## Roadmap
 
 ### Near-term
-- More robust multilingual workflows (transcription + summarization)
+- Enhanced audio processing workflows (transcription + summarization)
 - Additional communicator backends (Discord, Teams, Telegram)
 - Stronger observability (structured logs, tracing, evaluation dashboards)
 

architecture.md

@@ -0,0 +1,41 @@
+graph TB
+    subgraph "Client Layer"
+        CLI[Interactive Interface]
+        APP[Python Application]
+    end
+    
+    subgraph "Orchestration Layer"
+        ORCHESTRATOR[AudioProcessingAgent]
+        PARSER[PromptParser]
+    end
+    
+    subgraph "MCP Services Layer"
+        AUDIO_MCP[Audio Processor MCP<br/>Port 9001]
+        TOPIC_MCP[Topic Classifier MCP<br/>Port 9002]
+        SUMMARIZER_MCP[Summarizer MCP<br/>Port 9000]
+        COMMUNICATOR_MCP[Communicator MCP<br/>Port 9003]
+    end
+    
+    subgraph "Communication Backends"
+        EMAIL[Email Service]
+        SLACK[Slack Webhook]
+        PDF[PDF Exporter]
+    end
+    
+    CLI --> ORCHESTRATOR
+    APP --> ORCHESTRATOR
+    ORCHESTRATOR --> PARSER
+    
+    ORCHESTRATOR --> AUDIO_MCP
+    ORCHESTRATOR --> TOPIC_MCP
+    ORCHESTRATOR --> SUMMARIZER_MCP
+    ORCHESTRATOR --> COMMUNICATOR_MCP
+    
+    COMMUNICATOR_MCP --> EMAIL
+    COMMUNICATOR_MCP --> SLACK
+    COMMUNICATOR_MCP --> PDF
+    
+    AUDIO_MCP --> ORCHESTRATOR
+    TOPIC_MCP --> ORCHESTRATOR
+    SUMMARIZER_MCP --> ORCHESTRATOR
+    COMMUNICATOR_MCP --> ORCHESTRATOR