Build an Internal AI Tool for Product File Verification and Management

### 1. Project Overview I am developing a lightweight internal AI agent designed to run locally within our office network. The goal is to build an automated tool capable of scanning, inspecting, and managing our complete output file system to catch errors prior to product delivery or before moving assets to the next production stage. The system must support rule-based validation, combined with AI/Computer Vision capabilities where appropriate. Practicality, ease of use, extensibility, and the ability to run on-premises on Windows machines or a private local server are top priorities. ### 2. Core Functional Requirements #### A. File System Scanning and Management * Scan designated target directories. * Identify new, missing, duplicate, or misplaced files. * Verify folder structures against standardized company hierarchy guidelines. * Support scheduled automated scans as well as on-demand execution. * Maintain a non-destructive approach, minimizing any need to move or alter original source files. #### B. File Naming and Convention Checks * Validate file names against predefined naming conventions. * Detect errors in product codes, revision tags, suffixes, or missing metadata. * Flag duplicate file names, syntax/formatting mismatches, or files stored in the wrong folders. * Provide clear reporting on non-compliant files along with suggested corrections. * Apply automated renaming only upon explicit user confirmation. #### C. File Specification and Format Validation * Validate file extensions, file types, and required formats. * Check for mandatory deliverables required per product or output package. * Detect missing assets, mismatched versions, or deliverables failing handoff standards. * Allow future extensibility to inspect dimensions, resolution, metadata, and other file-specific technical parameters. * Support customizable rule sets tailored to distinct product categories. #### D. Product Color Defect Detection via AI / Computer Vision * **Key project feature:** Inspect and analyze visual product outputs. * Compare rendered/photographed product colors against reference images or standard digital color swatches. * Flag discrepancies including incorrect hues, color shifts, wrong color codes, or deviations exceeding defined tolerance thresholds ($\Delta E$). * Support region-based or component-level inspections when specific regions of interest (ROIs) are defined. * Output comprehensive inspection results: inspected image, reference image, highlighted discrepancy zones, and variance severity metrics. * Account for environmental variables (lighting, color profiles, backgrounds, and render/capture conditions) to reduce false positives. #### E. Reporting and User Interface * Provide a clean, intuitive UI to trigger scans and review results. * Group detected errors by category: invalid file names, missing files, wrong folder locations, format errors, and color deviations. * Display file paths, error classifications, severity levels, and suggested remediations. * Export audit reports in CSV, Excel, or HTML formats. * Keep the architecture extensible for future integration into an internal dashboard or notification service. ### 3. Working Method and Payment Structure The project should be broken down into discrete, milestone-driven tasks with well-defined acceptance criteria. **Proposed Milestone Breakdown:** * **Task 1:** Workflow assessment, requirements analysis, and system architecture design. * **Task 2:** Core scanning engine and folder structure validation module. * **Task 3:** File naming and convention validation module. * **Task 4:** File format, specification, and missing asset validation module. * **Task 5:** AI/CV color comparison and variance detection module. * **Task 6:** User interface implementation and comprehensive reporting engine. * **Task 7:** End-to-end testing, performance optimization, and handover. **Each task must include:** * A defined scope of work. * Concrete deliverables. * Clear acceptance criteria and corresponding test cases. * A fixed cost per task/milestone. * An estimated timeline for completion. Payments will be released upon meeting agreed-upon acceptance criteria for each milestone. We are open to starting with a smaller pilot task before committing to the full project scope. ### 4. Deliverables and Handover Requirements * Full, standalone source code with no external cloud dependencies required for core operations. * Comprehensive installation, configuration, and user documentation. * Externalized configuration files enabling internal team members to adjust validation rules independently. * Complete test suites and validation reports. * Technical documentation outlining system boundaries and constraints, particularly regarding color inspection tolerances. * Full ownership and intellectual property rights transferred to the client upon project completion. * **Safety constraint:** The system must default to a read-only mode during inspections. It must never delete, overwrite, move, or rename source files without explicit manual confirmation. Please submit your proposal with these information: - Your experience with AI Agents, Automation, Computer Vision - Your previous projects - Implementation for this project (specify tech stacks and hardware requirements) - Your proposed price for each phrase