Summary of Key Points
AI technology has made software development incredibly easy (reducing the time from weeks/months to hours/minutes), but making software user-friendly has become the biggest challenge. The traditional App Store model, which functions like a “showcase,” is outdated. The future of software distribution lies in the “intent routing layer”—where users don’t need to search for software; they simply need to specify their tasks using commonly used tools like ChatGPT, Slack, or Notion, and the system will automatically match, verify, and invoke the most suitable tool to complete the task. The key to new distribution models is “machines that understand user needs” combined with verifiable trust,” rather than relying on more apps or additional distribution channels.
1. The AI Era: Making Software Easier, but Using It More Difficult
In the past, creating a small tool required a team’s weeks of effort; now, with AI tools like Cursor and Replit, one person can complete it in just a few hours. However, the problem is that there are so many apps that users don’t have time to try them all.
For example, there are tens of thousands of tools for organizing meeting minutes, but you won’t download and compare each one manually. Users trust familiar sources: recommendations from influential bloggers they follow, default options in their communities, or tools automatically suggested by platforms like Slack. Therefore, AI hasn’t solved the distribution problem; instead, gaining user trust and attention has become more expensive—only those with capital (VC support), a reputation (recommendations from influencers), or marketing skills can succeed.
2. Why isn’t an “AI App Store” the Future Trend?
Many think that since AI generates so many apps, creating a new App Store would be the solution. But this approach is flawed.
Traditional App Stores are like showcases, designed to help users find software. In the AI era, with an overwhelming number of apps, users won’t have the time to browse through them. What’s needed are intelligent navigators that can understand user intentions. For instance, after you say “convert the meeting minutes into actionable items and send them to attendees,” the system should:
1. Understand your needs: Know that you need to organize the minutes, generate action items, and sync them with relevant tools.
2. Choose the best option: Pick the safest, most affordable, and most suitable tool for the task.
3. Use it directly: The system should automatically invoke the tool to complete the task without requiring users to download or register (handling permissions and payments as well).
In the future, software won’t be found by users; instead, it will appear automatically at the right time.
3. The Three Layers of Future Software Distribution: From “ Selling Apps” to “Selling Abilities”
The new distribution system consists of three layers, similar to building with blocks:
1. Generation and Hosting Layer: Tools like Replit and Cursor help you create software quickly, but they don’t handle how users use it.
2. Ability Registration Layer: These tools are registered as “callable modules” with clear information about their functionality (e.g., what they can do, required permissions, pricing model, success rate).
3. Intent Routing Layer: The most valuable layer—users specify their needs, and the system matches them with the most suitable module and invokes it directly. Examples include ChatGPT plugins and Slack Agents.
The key change is that apps are no longer the focus; instead, it’s about providing specific abilities that users can utilize without downloading the entire app.
4. Seven Channels for New Software Distribution
Software will be distributed across various user scenarios:
1. AI Assistants: Users can specify tasks directly within AI platforms like ChatGPT or Claude, and the assistants will automatically invoke relevant tools (e.g., booking flights through flight APIs).
2. Workflow Integration: Systems within platforms like Salesforce will suggest tools for writing follow-up emails; IDEs will offer code-generation tools.
3. Social Sharing: Users can share customizable tools on platforms like TikTok, which can spread quickly.
4. Creator Channels: Influencers can recommend tools, and some may even use equity as a form of investment (similar to VC).
5. Vertical Services: Solutions are first provided to solve specific problems (e.g., for dental clinics), and then repeated usage patterns are turned into software.
5. Five Questions Every Entrepreneur Should Ask
In the AI era, being able to create software is no longer enough; the real question is whether it can be used by users. Consider these:
1. In what scenarios do users need your product? (e.g., automatically organizing invoices for expense reports, not just providing an AI tool.)
2. On which platforms will users use your product? (Slack, Notion, etc.) Be where your users are, not make them search for you.
3. In what form should it be packaged? It doesn’t have to be an app; it could be a ChatGPT plugin, a Slack Agent, or an API.
4. Do you have signals of trust? Clear pricing, security certifications, user reviews are essential.
5. Does your product have a mechanism for sharing? Can users modify and share the results?
In summary, AI has increased the supply of software, but the real opportunity lies in turning it into **usable capabilities that machines can discover, trust, and invoke automatically.* Those who master this will control the new distribution landscape.