Why I Built Ornix as a Rule-Based System (Not AI-Powered)
Building a file organizer without AI sounds like a strange choice in 2025. When I made that decision for Ornix, I heard the same reaction. But it was not an oversight. It was the conclusion of thinking carefully about what this problem actually requires. Follow the nature of file organization far enough, and the case for a deterministic, rule-based approach becomes clear on its own.
Speed Is Not Negotiable Here
According to McKinsey, employees spend an average of 1.8 hours per day searching for files and gathering information. For a file organizer to reduce that number, the organization process itself cannot be the thing that interrupts work.
Consider what happens when a file moves during organization. If a document currently open in Excel or Word gets relocated, the path changes. The user has to go find it. The workflow breaks. Organization needs to happen as fast as possible, fast enough that users never feel it.
Rule-based file classification handles this in a few milliseconds per file. Checking a file extension and moving it to a designated folder is fundamentally a filesystem operation. An AI-based approach, by contrast, involves a network round trip and model inference time, adding hundreds of milliseconds to several seconds per file. Process a hundred files and that difference becomes immediately apparent.
The Spatial Memory That Consistency Builds
Most AI-based file organizers take a batch processing approach. They organize everything at once, at a specific point in time. There is a fundamental problem with this.
File locations are something users learn through repeated experience. A study published in Nature found that navigating digital folders activates the same brain structures used in real-world spatial navigation, specifically the parahippocampal regions. This is why people continue to prefer browsing folders over search, even as search tools have improved. A file's location becomes a kind of spatial memory.
When files consistently end up in the same places according to the same rules, users build an internal map without thinking about it. They know where to look without opening a folder. Sporadic batch cleanups and irregular organization prevent that spatial memory from forming.
Why Consistency Matters More Than Accuracy
The core limitation of AI-based classification is nondeterminism. Because AI operates probabilistically, the same file can be classified differently depending on context. report_final_v3.xlsx might go to Documents/Work/ one day and Documents/Finance/ another.
Analyzing file contents to find the most accurate destination is an appealing idea. But inconsistent results mean the spatial memory described above never forms. Users end up having to search for files every time. In file organization, consistency is more valuable than accuracy.
Rule-based systems answer this directly. The same input always produces the same output.
Rule-Based vs AI: The Practical Difference for File Organization
| Rule-Based (Ornix) | AI-Based | |
|---|---|---|
| Processing speed | A few ms per file | Hundreds of ms to seconds per file |
| Result consistency | Always identical | Probabilistic, varies by context |
| Offline operation | Yes | Requires API connection |
| Privacy | Files processed locally | File contents sent to external servers |
| Cost | None | Scales with API usage |
The Value of Following Proven Structures
How to organize files is a question that has been answered over decades of practice. Type-based separation, date-based archiving, project-based structures. These approaches have been validated by countless people using them in real workflows.
Ornix currently works by automatically sorting files by type. Simple, predictable, and reliable. The plan is to go further, recommending proven folder structure patterns to users. Following routines with demonstrated results is better suited to this problem than delegating each individual file decision to AI.
Choosing the Right Tool for the Problem
Not every problem calls for AI. Problems that require speed, consistency, and predictability are often better served by clear rules. One of the responsibilities of building tools is choosing the technology that fits the nature of the problem.
The rule-based approach in Ornix is the result of that judgment. Not a rejection of AI, but a choice of what works best here.
Ornix is an automatic file organization app for macOS.