Summary of the Key Points
This article illustrates the impact of AI agents (such as Codex) on traditional utility software through the author's own experience in creating a computer cleanup tool using such an agent. AI agents not only clean up clutter more efficiently and transparently—identifying files like Bilibili cache videos that traditional software misses—but they can also be customized to meet user needs. In contrast, traditional tools lose their advantage due to fixed rules and lack of transparency. The underlying logic is that AI agents are bridging the gap between humans and machines, transforming software from fixed, paid-for tools into flexible resources that can be tailored on demand.
Breakdown and Interpretation
#### 1. AI Agents for Cleaning Up Clutter: More “Considerate” and Transparent than Professional Software
How is the cleanup tool created by the author with an AI agent better than the paid CleanMyMac?
- Identifies hidden clutter that traditional software misses: For example, the author’s computer contained 96GB of Bilibili cache videos in deep directories, which CleanMyMac failed to detect. The AI agent found these files because it understands file content and user behavior (e.g., recognizing unwatched downloaded videos as clutter).
- Transparent information that reassures users at every step: The agent provides a report with three-color labels:
- Green: Pure caches and temporary files that can be deleted without worry (and even moved to the trash can reversibly).
- Yellow: Files that require user judgment (e.g., downloaded installation packages), with just an “open folder” option for users to review.
- Red: Core system files that should not be altered.
In contrast, CleanMyMac only found 15.8GB of clutter after half an hour of scanning, without providing specific details about the files (e.g., 3.8GB of Chrome cache; users had no idea if deleting it would affect login).
#### 2. The Weaknesses of Traditional Utility Software: Fixed Rules Cannot Compete with Flexibility of AI
Traditional tools like CleanMyMac and 360 Security Manager operate based on pre-set rules. Developers define which files are considered clutter, and the software only scans those. However, user needs vary greatly:
- You might want to delete documents downloaded last year but not opened; traditional tools won’t take that into account.
- You might want to keep certain caches (e.g., offline data from frequently used websites); traditional tools cannot make personalized judgments.
AI agents, on the other hand, can understand simple user instructions (e.g., “Delete all old videos larger than 1GB”) and can be updated at any time. The author created a cleanup skill that works on both Mac and Windows, something traditional tools (like CleanMyMac, which is Mac-only) struggle to achieve.
#### 3. The Logic Behind It: AI is Eliminating the “Translation Layer” Between Humans and Machines
Previously, software acted as a translator between humans and machines: users had to use specific commands (e.g., “Scan for clutter”), and the software translated these into machine-readable instructions. But AI agents can directly understand natural language commands (e.g., “Analyze my computer’s storage for optimization”). The author says, “The essence of software is its translation layer; agents are filling this gap,” meaning users no longer need to learn how to use various tools; they can simply talk to the AI, and it will handle everything for them.
#### 4. The Impact on the Utility Software Industry: From “Assets” to “Consumables”
Traditional utility software was often a long-term investment (e.g., CleanMyMac costs $120 and must be used continuously). AI agents, however, are more like consumables:
- You can quickly customize functions using prompts (e.g., the author created a cleanup skill using tokens).
- Functions can be updated at any time (e.g., adding a “find duplicate files” feature by changing the prompt).
- The cost is much lower (token fees are much lower than software subscription fees).
For software companies that rely on single functions (e.g., cleaning, decompression, PDF editing), their competitors are no longer other software companies but user-created prompts. This represents a fatal threat to the traditional utility software industry.
#### 5. Tips for Users Using AI Agents: Be Cautious but Not Panicked
The author advises users to be cautious when using AI tools:
- Always ensure that AI actions are read-only; files are deleted only with explicit user confirmation, and they can be moved to the trash can reversibly.
- Although the error rate of AI is low, it’s important to double-check for sensitive operations like file deletion:
- Clearly understand what files the agent has marked as clutter.
- Prefer moving files to the trash can instead of directly deleting them.
- Back up important files in advance.
Conclusion
AI agents are changing the way we use tools. The fixed functions of traditional software are being replaced by flexible, customizable alternatives provided by AI. For users, this means more affordable and user-friendly tools; for the industry, it means that many traditional tools may become obsolete. In the future, we might not need to download multiple tools; we can simply ask the AI to solve all our problems—a true “tool revolution” in the age of AI.