Inspiration ARRAI was inspired by a simple but persistent problem we observed in real-world equipment inspections: constant interruption. Inspectors working around heavy machinery frequently have to stop what they are doing to pull out their phones, take photos, type descriptions, and manually fill out inspection forms. This disrupts workflow, increases cognitive load, and creates inconsistencies in documentation. In high-risk industrial environments, even small inefficiencies compound into lost time and missed issues. We realized that the true bottleneck was not inspection itself, but documentation friction. Traditional inspections can be modeled as a multi-step function: Inspection = Capture + Describe + Classify + Document. Each stage introduces manual effort, delay, and variability. We asked ourselves: What if inspectors did not have to stop inspecting in order to document? Instead of juggling tools, ARRAI allows inspectors to record a continuous walkthrough using Meta smart glasses. With a single click, they capture visuals, environmental context, and verbal commentary — completely hands-free. No stopping. No typing. No switching applications. The process becomes: Inspection = Record → AI Analysis → Structured Report. Once uploaded to the dashboard, ARRAI's AI analyzes both video and audio data. It detects potential equipment faults, classifies severity levels, and automatically fills out structured inspection templates. Because AI systems are probabilistic rather than perfectly deterministic, we acknowledge that accuracy is less than 100%. For this reason, ARRAI keeps the human in the loop. Inspectors can review the uploaded footage, verify detected issues, and manually confirm or override AI-generated findings. This hybrid model combines automation with accountability: Human Expertise + AI Assistance = Reliable Inspections. Ultimately, ARRAI was inspired by the desire to remove friction from inspections without removing human expertise. We are not replacing inspectors — we are augmenting them with tools that allow them to focus on safety, precision, and preventative maintenance.
Built With
- bash
- celery
- clip
- docker
- docker-compose
- fastapi
- ffmpeg
- framer-motion
- google-cloud
- google-cloud-pub/sub
- javascript
- jsonschema
- minio
- next.js
- numpy
- openai-api
- opencv
- postgresql
- pydantic
- python
- react
- sql
- sqlalchemy
- tailwind-css
- typescript
- zod
Log in or sign up for Devpost to join the conversation.