How carcajo/mtbank-ai-call-analytics works
This is an AI call-analytics platform for contact centers. The principal workflow accepts an audio recording through either the REST API or OpenWebUI pipeline, transcribes it, optionally diarizes speakers, and returns structured analysis. The analysis graph sequences classification, quality scoring, compliance checks, and summarization. The README also documents live pipeline status and Markdown reporting. The graph follows shown calls and documented capabilities; source details for some OpenWebUI behavior and runtime integrations were not sampled.
Main entry points: graph.py Analysis Graph · main.py REST API · __init__.py LLM Client Setup · logger.py JSON Agent Logging
Components 13
API Ingestion
- REST API
api/main.py
Speech Processing
- Whisper Transcription
asr/transcriber.py - Speaker Diarization
asr/transcriber.py
Call Analysis
- LLM Client Setup
agents/__init__.py - Analysis Graph
agents/graph.py - Topic and Priority
agents/classifier.py - Quality Checklist
agents/quality.py - Compliance Check
agents/compliance.py - Summary and Actions
agents/summarizer.py - JSON Agent Logging
utils/logger.py
Chat Integration
- OpenWebUI Pipeline
pipelines/pipeline.py
Other components
- Caller
- OpenRouter LLM
Connections 21
- Caller → REST API: uploads audio
README.md - Caller → OpenWebUI Pipeline: submits audio or URL
README.md - OpenWebUI Pipeline → REST API: posts audio
pipelines/pipeline.py - REST API → Whisper Transcription: transcribes audio
api/main.py - Whisper Transcription → Speaker Diarization: assigns speakers
asr/transcriber.py - REST API → Analysis Graph: analyzes transcript
api/main.py - Analysis Graph → Topic and Priority: dispatches first
agents/graph.py - Analysis Graph → Quality Checklist: dispatches next
agents/graph.py - Analysis Graph → Compliance Check: dispatches next
agents/graph.py - Analysis Graph → Summary and Actions: dispatches last
agents/graph.py - Topic and Priority → LLM Client Setup: gets LLM client
agents/classifier.py - Quality Checklist → LLM Client Setup: gets LLM client
agents/quality.py - Compliance Check → LLM Client Setup: gets LLM client
agents/compliance.py - Summary and Actions → LLM Client Setup: gets LLM client
agents/summarizer.py - LLM Client Setup → OpenRouter LLM: configures endpoint
agents/__init__.py - Topic and Priority → JSON Agent Logging: logs agent I/O
agents/classifier.py - Quality Checklist → JSON Agent Logging: logs agent I/O
agents/quality.py - Compliance Check → JSON Agent Logging: logs agent I/O
agents/compliance.py - Summary and Actions → JSON Agent Logging: logs agent I/O
agents/summarizer.py - REST API → Caller: returns JSON
api/main.py - OpenWebUI Pipeline → Caller: streams report
pipelines/pipeline.py