W12-D2 Observations: Front-End Vitality in the AI Wave

Thoughts on business travel and attending D2.

Last week, I clearly felt a growing sense of urgency around AI adoption among people around me. That likely came from two forces. First, the exchange session between the company and the financial services team focused heavily on AI-related topics. Second, after the lobster went viral, social media was flooded with information about AI replacing jobs.

AI trends rotate too quickly and are increasingly starting to feel like fan culture. Most of the content is not worth reading. For example, the hype around topics like “frontend is dead” and “programmers are dead” has faded, replaced by “the internet is dead.” Maybe more and more people are looking down at their own jobs and realizing they are more standardized and easier to replace.

I have always been cautious about claims that something is “dead.” No matter how airtight the logic may seem, that kind of conclusion carries a strong emotional and ideological stance. You can evolve your own field, but why insist on declaring others dead? More often than not, it is about creating a new hierarchy of contempt to grab attention, or pulling out a pile of things from your pocket to sell to you. Serious thinkers about the future, such as the well-known KK, never say “so-and-so is dead.” What they say is “so-and-so is becoming something else.” The gap between those two expressions is the gap between traffic and insight.

I attended D2 in Hangzhou over the weekend. D2 started as a frontend meetup, and this year marks its 20th edition. In recent years, it has gradually expanded into the broader terminal and device space.

This year, I originally wanted to see just how downbeat the industry could feel, but the atmosphere on site surprised me. Developers were still highly engaged and eager to talk, and the technical backgrounds of attendees were becoming more diverse. Looking at the overall agenda, this year’s D2 was already clearly dominated by AI coding. Terminal-related topics were pushed into breakout sessions, and the traditional center of gravity of frontend is giving way. The conference now has a strong chance of evolving further into a technology exchange platform focused on AI. In the future, D2 can be treated as a marker for observing the survival state of SWE. If it successfully transforms from a frontend conference into an AI technology conference, that itself will be a side proof of how much this era is changing.

From a community perspective, frontend remains one of the most active groups in the broader tech world. This has a lot to do with the technical DNA of the web: openness, low barriers to entry, and an emphasis on connection naturally select for developers who are more willing to express themselves and collaborate. This becomes even more obvious when compared with conferences in other technical fields, and it is also the most valuable trait of frontend as a community.

At the event, there were many self-developed tools, platforms, and solutions, which reminded me of the most prosperous years of frontend. That period looked very similar: the whole industry was highly active, but also highly chaotic. Most tools and frameworks eventually die off. What tends to remain are only a few: the ones with the greatest influence, the clearest design philosophy, and enough simplicity and reliability.

I made a brief note of the most worth revisiting items; see below:

Review of Key Points from 2026-D2

Evan You introduced an integrated JS toolchain. Why does he think his work matters? Why does the AI era need a unified and efficient toolchain?

Google introduced A2UI. You can see that a typical web mindset is still extending into new forms. The open, composable paradigm of the web is moving into the AI era.

Zhou Xiao from ByteDance Web Infra shared his views on Harness Engineering. In my view, his talk was among the strongest in technical depth, offering a calm analysis grounded in first principles.