Introduction
Over the past few years, the landscape of artificial intelligence has significantly transformed. Initially seen as a set of tools that were “easy to use,” AI is now becoming an indispensable part of our everyday lives. This evolution unfolded in three distinct phases, each marking a pivotal shift in how we interact with the technology.
Phase 1: AI as a “New Species”
Three years ago, AI applications were largely centered around a few standout products:
- ChatGPT: A platform for conversation and Q&A.
- Midjourney: A tool for generating images.
- Character.AI: A service for virtual character dialogues.
These applications are categorized as “AI native applications,” primarily designed to demonstrate the capabilities of AI.
During this initial phase, user interactions typically involved:
- Asking questions
- Generating images
- Engaging in casual conversations and entertainment
At this stage, users were more focused on experiencing AI rather than relying on it. Consequently, AI served more as a showcase of potential rather than a resource for productivity.
II. Second Phase: AI Embedded in Everyday Products
Significant advancements have unfolded over the last two years.
Today’s leading AI applications are not standalone products but established tools that have been enhanced with AI capabilities:
- CapCut: With 736 million monthly active users, almost all functionalities are now AI-driven.
- Canva: Integrating AI tools into its design processes.
- Notion: Increasing AI functionality from 20% to over 50%.
A crucial trend has emerged:
AI is starting to contribute nearly half of annual recurring revenue (ARR).
This shift underscores one key point:
AI has evolved from merely a function to a foundational infrastructure.
Differentiation Among Platforms
As AI capabilities become ingrained, the role of large models is changing:
They are transitioning from “chat tools” to “user portals.” Two discernible paths are emerging:
1) Super Entry Point (Consumer Level)
ChatGPT is pursuing several initiatives:
- Introduction of GPTs coupled with an App Store
- “Log in with ChatGPT” account system
- Integration into daily activities such as shopping, travel, and health.
The ultimate aim is to establish itself as the primary gateway to the internet.
2) Professional Work Platform (Productivity Side)
Claude’s approach is markedly different, focusing on:
- MCP (Model Context Protocol)
- Integrating development tools and data systems
- Building intricate workflows
This functions more like an AI operating system designed for knowledge workers.
The Rise of the Platform Flywheel
As users begin to incorporate AI into their daily routines:
- calendar
- CRM systems
- workflows
The barriers to switching platforms will increase rapidly, establishing a sense of stickiness to these platforms.
This leads to the classic flywheel effect:
- More users → More developers
- More developers → More features
- More features lead to greater user reliance.
This dynamic suggests that competition will not yield a single dominant company but rather two ecosystems that will coexist over the long run.
Third Stage: AI Transitioning to Active Task Completion
The real turning point has arisen over the past year.
AI is evolving from simply “generating content” to actively “performing tasks.” This represents a shift from “content generation” to the entirety of “task completion.”
Early AI initiatives, such as Midjourney and DALL·E, achieved the following:
- Content writing
- Image generation
Today’s advanced products have begun to:
- Break down tasks
- Execute automatically
- Deliver the completed work
Emergence of AI Agents
Taking products like OpenClaw as a reference, key changes are evident:
- Moving beyond mere question-answering
- Involving task disassembly
- Automating the entire process
For example, an entire process can look like this:
- Set a target
- Gather information
- Analyze and process the data
- Output the results
- Send automatically
At this juncture, AI shifts from being a mere tool to becoming a “software entity capable of action.”
A Further Development: AI Assisting in Product Creation
Emerging concepts like Vibe Coding are gaining traction, marked by products that essentially allow AI to directly “create” products for users. This transformative shift moves us from “humans writing code” to
IV. AI’s Shift Towards Web3
As AI progresses from “answering questions” to “carrying out tasks,” an essential question arises: how are these tasks completed in terms of transactions and settlements? Traditionally, these functions depend on platforms and middlemen—structures designed for humans that may not be suitable for independent machine operation.
Web3 offers a more fitting architecture for AI, characterized by:
- 24/7 operation: AI can operate continuously and respond at any time.
- Machine-native interface: Contracts serve as APIs that machines can directly call.
- Programmable assets: Fund transfers can occur automatically and without manual intervention.
This leads to a significant evolution: AI will not only “perform tasks” but also automatically handle payments and settlements.
Moreover, blockchain technology ensures transactions are immutable and verifiable, facilitating AI collaboration without the need for intermediaries. As a result, the fabric of trust on the internet is shifting—from “trust platforms” to “trust rules.”
Thus, the relationship between AI and Web3 represents a natural division of labor: AI manages actions while Web3 oversees settlements. This synergy may pave the way for the foundational structure of the next generation of the Internet.
