OpenCHS AI Assistant - Intelligent call documentation for GBV & child protection helplines across Africa. Auto-fills case forms, detects critical risks, returns 16,000 hours monthly to survivor care.
# OpenCHS AI Assistant π€
**Intelligent Call Documentation for GBV & Child Protection Helplines**
---
## π Award
**Runner-Up - Judges' Favorite**
MSIS 549 GenAI & Agentic Fair, February 2026
---
## π Overview
OpenCHS AI Assistant transforms how counselors document cases of gender-based violence (GBV) and child abuse across Africa. By automatically extracting structured information from natural conversations, it reduces documentation time by 60% and returns **16,000 hours monthly** to direct survivor care.
**Current Deployment:** 5 countries (Kenya, Uganda, Tanzania, Lesotho, Somalia) | 200+ counselors | 120,000+ calls monthly
---
## π¨ The Problem
Call centers handling child abuse and GBV cases face critical challenges:
- **Administrative Burden**: Counselors spend **60-70% of call time** filling detailed forms instead of supporting survivors
- **Scale Challenge**: **120,000 calls monthly** means thousands of hours wasted on paperwork
- **Delayed Risk Detection**: Critical cases may not be flagged quickly enough
- **Service Gaps**: Every delayed response could mean a child remains in danger
**Impact:** Time spent on forms = Less time helping survivors when they need it most.
---
## β¨ The Solution
### Phase 1 MVP (Deployed)
**Intelligent Call Transcription & Auto-Form Filling**
- π― **Auto-Extraction**: Extracts 19 structured fields from natural conversation
- β‘ **Lightning Fast**: 91%**
- β
Processing latency: **<1 second**
- β
Zero false negatives on critical risks
- β
Runner-up award - Judges' Favorite
### Real-World Deployment
- β
**60% reduction** in documentation time
- β
**40% increase** in calls handled per counselor
- β
Form completeness: **75% β 95%**
- β
Deployed in **5 countries**
- β
Serving **120,000+ survivors monthly**
---
## π Future Roadmap
### Phase 2: Multilingual Support
- Real-time translation for 10+ African languages
- NLLB-200 integration
- Expand accessibility to non-English speakers
### Phase 3: AI Quality Assuranc β¦