Gallery
Screenshots & Interfaces
Full-size screenshots per role — click card to cycle, hover to enlarge.
V-Fix combines agentic AI diagnostics with enterprise management to eliminate unnecessary technician dispatches — saving time and cost for both users and service companies.
When a home appliance breaks down, diagnosing the problem is time-consuming and costly. Non-expert users rarely know which steps are safe or relevant — leading to unnecessary technician dispatches and waiting times.
"V-Fix combines LLM flexibility with retrieval-based grounding, delivering technician-grade diagnostics through a conversational interface."
Our fully-agentic RAG system, orchestrated by Gemini 2.5 Pro, can conduct adaptive diagnostic conversations, understand images, request specific photos, and either guide users through safe self-repair or book a technician appointment with full diagnostic context pre-filled.
From first symptom to resolution in minutes — no technical expertise required.
Open the app and describe what's wrong. V-Fix understands natural language and error codes.
The agentic system queries the knowledge graph, manufacturer data, and trusted repair resources to pinpoint the fault.
Get step-by-step fix instructions or book a technician with full context auto-filled.
Rate the experience. Your feedback improves the AI model for everyone.
From end-users seeking instant repair guidance to enterprise managers overseeing hundreds of technicians — V-Fix covers the complete service lifecycle.
AI asks intelligent follow-up questions to pinpoint the fault — just like a skilled remote technician.
Upload photos of error codes, components, or damage. The VLM analyzes images to identify faults visually.
Safe, ordered, actionable repair instructions that any non-expert can follow — warranty-compliant.
When self-repair isn't possible, book a technician with full diagnostic context pre-filled — no re-explaining needed.
All chat sessions, uploaded images, and repair steps are securely stored and accessible from any device.
Full-featured iOS & Android app for on-the-go appliance diagnosis, photo uploads, and appointment tracking.
Technicians view assigned appointments, full diagnostic context, and rate AI model performance — feeding the improvement loop.
Monitor technician workload, manage appointments, observe branch performance metrics, and approve leave requests.
Full-system statistics: session outcomes, AI accuracy, user feedback scores, resolution times, and model improvement data.
Technician corrections and ratings automatically feed back into the model training pipeline for continuous improvement.
New service companies integrate into the V-Fix ecosystem easily. Upload brand/model data to train company-specific knowledge.
Strict RBAC ensures customers, technicians, branch managers, and admins each see only their relevant data and actions.
At the heart of V-Fix is a fully-agentic RAG pipeline orchestrated by Gemini 2.5 Pro, grounded with structured knowledge sources for reliable, hallucination-resistant diagnostics.
| Criterion | V-Fix | Manufacturer Assistants | General LLMs (GPT/Gemini) |
|---|---|---|---|
| Diagnostic Adaptability | Fully adaptive conversations | Rule-based, limited paths | Generic, ungrounded |
| Hallucination Risk | Low — RAG-grounded | Low but very limited scope | High in technical scenarios |
| Vision Understanding | Full multimodal VLM | Rarely available | Available but ungrounded |
| Appointment Management | Integrated end-to-end | Call center redirect | Not available |
| Enterprise Dashboard | Full suite (4 roles) | Basic CRM only | Not available |
| AI Improvement Loop | Technician feedback integration | Static rule updates | No domain feedback |
A carefully selected, production-grade stack enabling scalable AI diagnostics, real-time dashboards, and cross-platform mobile support.
Watch end-to-end demo sessions showcasing the diagnostic chatbot, enterprise dashboards, and mobile application.
A short presentation of V-Fix covering the problem, solution, and live demo highlights.
Full-size screenshots per role — click card to cycle, hover to enlarge.
Five computer engineering seniors from METU, advised by a leading researcher in NLP and AI.
Have questions about V-Fix? Reach out to the team or try the application directly.
Middle East Technical University
Dept. of Computer Engineering
Ankara, Turkey 06800
Try V-Fix directly in your browser
METU CENG 491/492
Senior Design Project — Spring 2025/2026
Available on web and mobile. Scan the QR codes below to get started instantly.
