Flowback: AI-Powered Work-Context Recovery System

Flowback is an AI-powered Work-Context Recovery System built by HackMantra to solve a problem that traditional browser history does not address: losing the mental context of work after an interruption. Modern work is fragmented across documentation, code, research, AI tools, GitHub, and communication tabs. When an interruption happens, users often remember the website they were on but not what they were doing, what they already tried, where they stopped, or what they should do next. Flowback observes meaningful browser context and detects interruptions, then reconstructs a structured recovery capsule containing what the user was working on, what they tried or referenced, where they left off, and a recommended next step. The user can resume the original tab and scroll position with one action. The current MVP is a Chrome Manifest V3 extension with background monitoring, context extraction, deterministic recovery logic, a recovery UI, resume functionality, and optional AI enhancement. AI is treated as an enhancement rather than a single point of failure, with deterministic fallback when AI or backend services are unavailable. Our architecture is designed for accelerated AI inference and future deployment on AMD infrastructure. During the hackathon, HackMantra aims to experiment with AMD GPU acceleration, optimized local/hosted inference, and benchmarking to determine where AMD compute provides the greatest benefit to Flowback's recovery intelligence. Flowback's long-term direction is to evolve from browser recovery into cross-device work continuity, helping users preserve work context across desktop and mobile environments. Team: HackMantra.