Guide
ai-application-roadmap tracks how applied AI engineering practices evolve over time.
A Small Story
Around age 17, I got obsessed with philosophy. I read both Chinese and Western works, including Caigentan, Principles of Human Knowledge, The Origin of the Family, Private Property and the State, One-Dimensional Man, and On Contradiction.
As I read more, I realized philosophy was huge, but I had no timeline or map to tell me where I was. Then I thought: philosophy is taught in undergraduate programs, so I bought a philosophy textbook to see how they learn it.
That textbook offered one idea that stayed with me: learning philosophy means learning the history of philosophy. At that moment, I knew I had bought the right book.
That is also why I started this project. Applied AI engineering is moving fast, but we still lack a clear timeline map that tells people where they are and what comes next.
If you only master one or two hot technologies, you may get strong short-term returns. But that alone will not make you an expert. To become an expert, you need to understand the history.
If you are serious about building AI systems, you are invited to contribute. The bar is intentionally low: add one markdown file under docs/zh/timeline/*.md (or docs/en/timeline/*.md for English), document a node clearly, and open a PR.
What You Will Find
This site has two types of content:
Timeline
Answers: which technical nodes matter, what stage they are in, and where they are heading.
Each node includes engineering significance and adoption judgment, not just hype.Guide tutorials
Answers: how to actually do the work. We keep updating three blocks:- Core ideas and hands-on methods for vibe coding
- Practical onboarding paths for key timeline technologies (how to use, when to use, common pitfalls)
- Concrete AI productivity workflows (browser operation, PPT generation, data handling, and more)
How to Use This Site
Recommended flow:
1. Start with the timeline to decide what to learn
Use homepage filter sets to narrow by year and adoption effort, then check recommended and key nodes.
2. Use evolution labels to decide whether to adopt now
Treat phase / trend / signal as three decision questions:
- Is this a mature approach or an early one?
- Is it rising, stable, or being replaced/shrinking?
- Is community attention aligned with real engineering value?
3. Go to Guide for implementation details
Open the matching tutorial and focus on: minimum runnable path, recommended setup, common mistakes, and migration patterns.
Example: everything-claude-code is popular. It was also the setup used by the Anthropic x Forum Ventures hackathon winner, and the README includes a Quick Start. But real onboarding is still heavier than it first appears: beyond installing plugins, you often need to handle rules, install dependencies, switch setups across languages, and configure extra capabilities. Many people follow the steps and still feel Claude Code ends up getting worse, not better.
You should not have to repeat that kind of detour.
Before AGI fully arrives, hopefully this site helps you waste less time.