From Language Barriers to Local AI: Rebuilding WACAO in the Age of Browser Intelligence
Sep 19, 2025
We've all been there. You open WhatsApp after a few hours away and find 400 unread messages across various group chats. Your heart sinks as you scroll through memes, random thoughts, and that one important piece of information buried somewhere in the digital noise. What if you could catch up in seconds instead of minutes?
This frustration sparked an idea back in 2017, and eight years later, we've rebuilt it with something that seemed impossible then: AI that runs entirely in your browser.
The Origin Story: Lost in Translation
The journey with WACAO started way back in 2017, when we had just moved to Germany. WhatsApp was already the dominant messaging platform, but we didn't know the language. When talking with service providers, colleagues, and new friends, they naturally preferred to speak in German, and our German was... well, let's just say it needed work.
We found ourselves stuck in this endless loop: write a message in Google Translate, copy-paste the German version, switch back to WhatsApp, type it out, send it, wait for a response, then reverse the entire process. It was exhausting.
But the real challenge wasn't just translation, it was the group chats. Anyone who has ever used WhatsApp knows that group chats can get incredibly noisy, incredibly fast. We needed a way to quickly summarize these conversations so we could identify what actually required our attention.
The First WACAO: When AI Was Still a Dream
This led us to build the original WACAO extension. It got quite popular, even featured by TechCrunch. It was a Chrome extension for regular users, and the fans who discovered it absolutely loved the functionality.
In fact, it became so popular that companies started approaching us. This was functionality that was missing from WhatsApp Business, which had just launched, and they could use these capabilities to respond to customers automatically, provide product catalogs, and conduct smart conversations.
But here's the thing about 2017: the WhatsApp API wasn't readily available, and the technical landscape was very different. For several reasons, including API limitations and some personal circumstances, we had to pause the project. We stayed active in the WhatsApp space, working with customers and startups, but WACAO itself went dormant.
The Spark: What Would WACAO Look Like Today?
Fast forward to today and to NueFunnel. We are an AI collective - a group of explorers and tinkerers who want to make use of advancements in AI and technology to ultimately help people. So we asked ourselves: what would WACAO look like in today's age of AI?
We were quite surprised by the answer.
The Magic Moment: AI in Your Browser
The new version of WACAO provides the same functionality - summarizing group chats and auto-translation - but this time with the help of language models that Chrome has incorporated directly into their browser APIs. When you spin up the Chrome browser with the WACAO extension, it actually downloads a small language model (SLM) directly into your browser. This model handles summarization, language detection, and translation - all locally, without sending your private messages to any server.
This was a pretty cool realization. The same functionality that required multiple Python libraries and complex server setups seven years ago could now be achieved with a small language model running directly in Chrome. And once we got summarization working smoothly, we also wanted to add auto-translate. The language of incoming messages is automatically detected, translated to your preferred language and your response is automatically translated to the other user's language.
The Technical Reality: Making Browser AI Work
Building WACAO 2.0 wasn't without challenges. WhatsApp Web doesn't exactly make it easy to extract messages - there's no simple API like what Telegram offers. The DOM structure changes frequently, and we had to develop robust selectors that work across different layouts.
We implemented a DOM parser that intelligently extracts chat messages using multiple fallback strategies. Chrome adds a layer of APIs on top of the SLM to expose specific functionality like:
Summarizer API for generating key-point summaries in markdown format
Language Detection API for automatically identifying message languages
Translator API for seamless cross-language communication
All of this processing happens client-side using the SLM and you can check for yourself as we have open-sourced the github repository as well.
What We Learned: AI Should Meet People Where They Are
Rebuilding WACAO taught us something important: making AI useful means meeting people where they already are - in the tools they use every day. Rather than asking users to switch platforms or upload their data to cloud services, we brought the intelligence directly into their existing workflow. The barrier to entry became a simple browser extension install, and the privacy concerns disappeared entirely.
This reflects a broader shift we're seeing: AI is becoming more accessible not just because models are getting better, but because they're getting smaller and more local.
Try It Yourself
WACAO is live on the Chrome Web Store and ready to help you tame your WhatsApp chaos. Whether you're dealing with multilingual group chats, trying to catch up on busy conversations, or just want to communicate more freely across language barriers, we built it to work seamlessly in the background.
The extension provides a clean side panel interface with two main features:-
Smart Summaries: Get key points from your last 20 or 50 messages
Instant Translation: Translate incoming messages and send replies in any supported language
Everything processes locally, so your conversations stay private.
The Exploration Continues
At Nuefunnel, we see projects like WACAO as experiments - small steps in our journey to make AI genuinely helpful for real people solving real problems. The fact that we could rebuild and enhance a complex AI application using nothing but browser APIs shows how rapidly the landscape is evolving.
What started as a personal frustration with language barriers has become a window into the future of privacy-preserving, locally-running AI. And we feel like we're just getting started. Try out WACAO and let us know what you think!