Issue #14: Imagine

🎶 You may say I’m a dreamer, but I’m not the only one. I hope someday you’ll join us, and the world will live as one… 🎶

This newsletter, as many others, is a kind of chronicles of what’s happening on the tech scene (and in the metaframeworks ecosystem specifically) currently. But at the same time it is an attempt to analyze and predict what will happen in future based on the development trends and ideas in the air. Technologies evolve constantly, and we are the witnesses of this sweeping evolution. For some this evolution is too fast (like in the old good Fatboy Slim’s video) and overwhelming, for some it’s too slow, and they can’t wait for AI to develop everything for us. Be that as it may, this issue brings some nuggets for both camps (let me know which one are you from personally — just reply to this letter with your thoughts!) so take your popcorn bag and get ready for the ride.

The Good

Malte Ubl from Vercel came with the rather expected disclosure of the company positioning itself as the AI engineering company. The authors of the most popular AI SDK for JS/TS apps, and the most popular (even among the AI agents, one of which is also a product of Vercel’s) metaframework already took the place of trendsetters in the ecosystem. Now they want to reign it and ride the thought leadership wave too balancing between AI shovels selling and being trusted mind mates for developers like you and me.

As developers, we need to find our areas of expertise where we can outperform AI. The technology is a tool to accelerate your work, not replace the thinking part.

Even the recent Next.js Global Hackathon had drawn a lot of good attention from the community and brought some curious results of using Next.js for different creative web app solutions.

At the same time, the team has turned to very active development (and blogging) mode lately, bringing up both the new thoughtful platform features and AI tools, and speeding up development of security and observability improvements. After the recent wave of criticism for Vercel in the context of Next.js openness and portability, these platform improvement enhance the company’s position from one side, but at the same time still make using the Vercel’s products outside the Vercel’s infrastructure “a bit” complicated. Of course, that’s just business and nothing personal and some companies and teams are totally fine with the vendor lock in, but hopefully Vercel will leave us some chance of using their awesome tools in more fine-grained way.

The Bad

On the same AI note, cool guys from AI build tools like Lovable and Bolt (both engaging metaframework-based development actively) came with a lot of updates recently. V2 of the Lovable app declares a lot of awesomeness like better security and multi-user access to real-time development. The latest Bolt capabilities promise extra tokens and rich design options outside the generic shadcn/ui patterns. But while all of that sounds good, more interesting voice to listen to the last week for me personally became the voice of the Stacklok team who had created the LLM Security Leaderboard. What’s interesting in this Huggingface leaderboard is that even though none of the closed source models (like Claude 3.7 Sonnet from Anthropic, or o1 family from OpenAI, or Gemini 2.5 from Google) traditionally used for AI coding assistants and agents is present in the assessment, the results are quite [evenly] pessimistic. That allows to interpolate the stats to the output of tools like Lovable, Bolt, V0, or IDX (which recently was rebranded a bit) and be mindfully cautious about the production readiness of them.

The Noteworthy

While AI largely empowers the new generation of developers, the old generation in the person of Ryan Carniato reflects on the last decade of frontend web development and the path SolidJS had gone during these years. The framework author way was quite harsh.

The average developer had gone through a few years of JavaScript fatigue which was very much still a fresh topic at the time, so a new framework was the last thing anyone wanted. People were angry.

Nevertheless, I’m quite happy we now are in the stable stage where Solid and SolidStart are here and thriving proceeding setting new high bars for client-side performance among JS/TS libraries and frameworks. It’s exciting to see how many efforts are on the backstage of the tooling development work. As a developer tooling craftsman myself, I love to explore the technologies and approaches behind the proven ecosystem gems.

There are still ways to make Web Development better and that energizes me. If what we accomplished as a small force in a hostile environment is impressive, imagine what we can accomplish with that many more people living and working in this space. This is only the beginning of the journey.

Other tool that is near and dear to my heart (and my hands which literally hold it right now developing a new component for a design library), Storybook, went beta with its new major version 9 the last week.

Storybook 9 is our most ambitious release yet.

I anticipate the updates this version brings a lot, especially the testing aspects of it. Accessibility testing, visual testing, and testing coverage are the things that will empower a lot of dev teams with better UI development flows (and better accountability for their bosses too, heh). You can give it a try right now and provide your feedback which will help the guys build the greatest tool for UI libraries grinding.

The Nuxt team had brought the new minor version the last week but as usual, there are plenty of precious deliverables in it. Including the awesome thing even the demanding Ryan Carniato would love — the reactive keys. Other goodies include some new useful components and the packaged documentation — the feature I’d love to see from all other metaframework vendors so much.

This issue mentions a lot of good teams and people, not only the Nuxt guys who are just the elves of the metaframeworks ecosystem bringing such a lot of innovation to the space. I truly believe that it is people that will move this technological evolution forward, with or without AI. And I hope these examples will draw new curious talents to open source communities working on the tooling ecosystem. We all need that to evolve and thrive.

👋