Remember when asking your smart speaker a question actually worked? When it didn’t try to sell you something first, or misunderstand your request, or just say “Sorry, I can’t help with that yet”? Those days are fading fast.
Google‘s Nest Audio and Amazon‘s Echo speakers have quietly been getting worse. Not by accident — by design. Features disappear. Ads creep in. Accuracy drops. It’s a pattern familiar to anyone who’s watched Big Tech slowly enshittify its own products.
But there’s a countertrend. A small, open-source voice assistant called Mycroft — now reborn as Neon AI — keeps improving. No ads. No data mining. No forced updates that break what worked yesterday. And it runs on hardware you can actually control.
What’s going wrong with Google and Amazon smart speakers
Let’s be specific. In 2023, Google removed several features from its smart displays, including the ability to set a default music service on the Nest Hub Max. Amazon, meanwhile, has been pushing more advertising into Alexa — from sponsored answers to audible ads before your weather report.
Both companies have also quietly cut back on third-party integrations. The “Skills” ecosystem on Alexa is littered with broken or abandoned apps. Google Assistant’s once-impressive natural language understanding has, by many users’ accounts, regressed. Simple queries like “What’s the capital of France?” sometimes trigger a web search instead of a direct answer.
Why? Because neither company makes money selling speakers. They make money collecting your data and serving you ads. Once you’ve bought the hardware, the incentive is to extract more value from you — not to make the device better for you.
Meet the open-source alternative that’s actually improving
Enter Neon AI, the successor to the Mycroft project. It’s a fully open-source voice assistant that runs on a Raspberry Pi or a dedicated device like the Mark II. And unlike its corporate cousins, it’s getting better with each update.
Recent improvements include:
- Better wake word detection — it actually hears you from across the room now
- Offline speech recognition for basic commands (no cloud required)
- Integration with home automation platforms like Home Assistant and openHAB
- Customizable skills that anyone can write, not just approved developers
- No ads, no data collection, no lock-in
It’s not perfect. The voice still sounds a bit robotic. It can’t answer every random trivia question. But for core tasks — controlling lights, setting timers, playing music from your own library — it works. And it respects your privacy.
Why open source matters for smart speakers
The fundamental problem with Google and Amazon speakers is that you don’t own them. You’re renting a listening device that happens to be shaped like a speaker. The companies can change the terms anytime. And they do.
With an open-source voice assistant, you control the software. If the community doesn’t like a change, it can fork the code. If a feature disappears, someone can rebuild it. The device serves you, not a corporate bottom line.
This isn’t just idealism. It’s practical. When Amazon announced in 2023 that it would lay off thousands of employees, including many working on Alexa, the writing was on the wall. The platform’s future is uncertain. Open-source projects, by contrast, don’t depend on quarterly earnings reports.
What you need to get started
If you’re curious, the barrier to entry is low. You need:
- A Raspberry Pi 4 or 5 (about $35–$75)
- A USB microphone array (around $20–$50)
- A speaker with a 3.5mm input
- An SD card with the Neon AI image flashed onto it
Total cost: roughly $100–$150. That’s comparable to a single Echo Studio. And you can build multiple units for different rooms.
Is it ready for everyone?
Let’s be honest: no. The open-source smart speaker is still a project for tinkerers. Setting it up requires basic Linux knowledge. You won’t find it at Best Buy. And it won’t order pizza for you (yet).
But for people who care about privacy, who are tired of ads in their kitchen, or who just want a device that doesn’t get worse every year, it’s a genuine alternative. And it’s only getting better — which is more than you can say for the big-brand options.
The real question is: how much longer are you willing to put up with a smart speaker that works for its manufacturer, not for you?
