AboutJobsMeetupBlogJoin Slack

Future Frontend 2025 - Recap

Author: Abudumiti Alimujiang
Editor: Juho Vepsäläinen

One important part of running a conference is documenting it. This time around, Abudumiti Alimujiang helped us with documentation as he captured video footage and notes, which he turned into this post. The footage was cut into a short mood video you can find at the end to give you an impression of our event which took place 27-28.5.25 at Finland. Read on to find out what Abudumiti got out of our conference!

Opening#

On the morning of June 27th, Future Frontend 2025 kicked off under a sunny sky. Around 180 developers, designers, and tech professionals gathered at Dipoli, a building once designed in the 60s to represent the future. Today, it’s the home of a conference that looks into what frontend might become.

As speakers and guests arrived gradually, a light breakfast and friendly conversations set a relaxed tone for the day. There was a small hiccup at the beginning as the screen and sound didn’t work. After a brief moment of awkwardness someone in the audience joked, “Have you tried turning it off and on again?” They did. It worked. The room laughed and the conference finally started as Tuuli and Daniel, our MCs, gave a short introduction on what is to come.

Performance#

Performance is one of the core topics in frontend. Ewa Gasperowicz and Barry Pollard from the Chrome team introduced the newest DevTools features. The session focused on performance insights and how AI can help you during development. The updates now support multimodal input. That means developers can describe a problem in plain language and get suggestions. For example, how to fix a styling issue with SCSS.

They also showed a new way of using Workspaces. The teaching materials now include short videos that are easy to follow, step by step. Many junior developers avoid the Performance Panel, but these updates lower the barrier. They analyzed real websites like smartly.io and name.com to demonstrate how the tools work in real life.

Work-life balance#

The work-life balance session reminded us that performance isn’t just about apps. It’s also about people. Georgios Diamantopoulos opened with a talk called “Shifting the Trajectory.” He shared how sitting for long hours caused him physical injuries. He explained how prolonged screen time affects the spine, heart, and energy levels. Then he led a short exercise to show how small movements can help. His message was clear. Your wellbeing matters.

Next, Olavi Haapala brought a different angle. He combined stories from his own life. As a freelance developer and part-time tree cutter, he discovered that switching between mental and physical tasks helped him avoid burnout. Whether it’s climbing trees, running side projects, or playing hockey, adding variety helps you stay sharp and motivated.

Design systems#

Julien Sulpis talked about building framework-independent design systems. His strategy uses Web Components and design tokens. You define tokens in Figma, then bind them to Tailwind CSS. Components can then be shared across projects by publishing them to npm. Since Web Components don’t depend on any framework, that makes them perfect for different kinds of apps, including legacy systems.

The second talk came from m4dz 🎙🥑. He shared how his team modernized a very old Django-based app. Instead of rewriting from scratch, they rebuilt it step by step. Using JinjaX templates and Storybook, they created a UI system that looked like something from 2025. He said even old codebases can evolve if you have the right workflow and tools.

Demonstrations#

Charlie Gerard gave one of the most visually exciting talks of the conference. She explored how LLMs can work with motion input. In the past projects, she used TensorFlow.js to build gaze-controlled keyboards and hand-controlled games. This time, she asked: can we use hand gestures to control light bulbs with LLM? She showed how LLMs interpret video, recognize gestures, and trigger actions.

In the live demo, she used Gemini and pose detection to control two light bulbs on stage. It looked like sci-fi. But it was real. Charlie showed how combining LLMs with motion sensing can create interfaces that feel natural and responsive.

Tero Parviainen’s talk felt like a dream. He built an AI-powered soundscape system that responds to voice. Say “take me to the jungle,” and the browser plays a multi-layered ambient track. All sounds are generated. Nothing is pre-recorded.

The process uses Deepgram for speech-to-text. Then a Gemini model breaks the input into four sound layers. It checks a 20,000-item sound library for matches using CLAP embeddings. If no match is found, it generates new sounds with Stable Audio Open. The final mix is played in the browser with WebAssembly. Tero finished with a 3D visualization of the sound library. The audience was completely immersed. It felt like listening to a hallucination.

Afterparty#

To close the first day, many walked to the seaside sauna, a setting that couldn’t feel more Finnish. The weather was perfect and the sun stayed up late. Attendees from different countries and backgrounds shared drinks, food, and thoughts on everything from frontend tools to Finnish nature.

React#

Day two started with a React deep dive. Aurora Scharff talked about React Router, Next.js, and Server Components. She showed how to migrate to Server Components step by step. Her demo used a task manager app to show performance wins in FCP and LCP.

Then Devlin Duldulao introduced TanStack Start. A full-stack React framework built on the TanStack ecosystem. He showed how routing, layouts, and data are handled. The key message was: no single tool fits all. Choose based on project needs. Know when to stay client-side and when to offload to the server.

Accessibility#

Marianna Österlund opened with a “heavy book” as she talked about the European Accessibility Act that becomes mandatory on June 28th, 2025. She explained what the new law covers, how to check if your work is affected, and why it matters. It turns out understanding Web Content Accessibility Guidelines is the key and the law isn’t just a rule. It’s about people. Over 100 million of them. She shared learning resources and encouraged developers to start now.

Then came Eeva-Jonna 'Eevis' Panula. She spoke from her experience as a disabled developer. Her talk focused on mobile and things like screen readers, font scaling, and orientation locking. She showed how bad WebViews can harm accessibility. She also called need to test, not guess.

Cybersecurity#

Jussi Eronen talked about supply chain risks and dependency overload. For example, installing a package from npm might bring hidden vulnerabilities. His advice: reduce tech debt, track what you use, and keep everything updated. He introduced STRIDE threat modeling. Bring your team together and imagine worst-case scenarios. AI can help, but it doesn’t write secure code. Developers do.

Zak Allal shared his story with a Finnish learning site. His MongoDB was exposed due to a misconfigured deployment. He gave real-world examples of how hackers exploit public endpoints. His solution: reduce the attack surface. Use gRPC, Qwik framework, or single-binary architectures like Bun. Always follow zero-trust principles.

AI and ML#

Rachel-Lee Nabors made a bold prediction. Browsers may fade as AI agents take over tasks like search, shopping, and navigation. She introduced MCP (Model Coordination Protocol). MCP lets LLMs perform actions directly. In this future, developers should build content optimized for AI. She showed tools like Bolt and v0, and asked: what if the agent is the interface? The web still needs us. But it is changing fast.

Stephanie Nemeth and Sebastian Obel closed the event with something personal. They shared a story about finishing a project that was started in 2021 they continued this year to finish it for the conference. The original project was an interactive sound installation with 236 touch points. They built it by hand using Raspberry Pi 4, Trill Craft, Node.js, and Pure Data patches.

The project was more than just tech. It was also about healing. They spoke about burnout and neurodivergence. Sometimes, finishing something can be a form of therapy. The room was quiet and moved. A fitting end.

Closing#

At the very end, the video that failed to play with sound during the opening finally appeared. It was a remix of the traditional Finnish tune Ievan polkka. The visuals matched the iconic metal flower in front of Dipoli. It was a perfect final image.

Juho Vepsäläinen gave a short thank you speech. He thanked everyone who joined — the speakers, the crowd, and the volunteers. And the sponsors who helped make it happen: Smartly, Wunderdog, Alma Media, Backscreen, and Wonna.

Then came the announcement. The next Future Frontend will happen again at Dipoli on June 8–9, 2026 and the workshops follow in 10–12.6. Consider preregistering to the conference to note your interest to the event.

See you in the future.