My Journey with Voice AI Hardware #1: The Limitless Pendant That Actually Delivers
It could be really fun and useful!
Hey there! I know I've been a bit quiet here lately, but there's a good reason for that. I've been deep in the trenches, testing what might be one of the most practical AI wearables I've ever encountered. And honestly? I'm kind of obsessed.
Side Note: This Post is NOT sponsored.
This is the first post in what's going to be a series about Voice AI hardware - the good, the bad, and the surprisingly useful. Because let's face it, we've seen enough AI gadget failures (looking at you, Humane AI Pin) that when something actually works... well, it's worth talking about.
The $49 Gamble That Actually Paid Off
A few months back - honestly, time is a blur when you're constantly testing new tech - I came across this Kickstarter-style campaign for something called the Limitless Pendant. Early bird price? $49. The pitch? A simple wearable that records everything, transcribes it, and helps you remember your life better.
No promises of replacing your phone. No holographic projections. No camera trying to see what you're eating. Just... memory augmentation.
I backed it immediately. Why? Because in a world of overhyped AI gadgets promising to revolutionize everything, here was a company that understood something crucial: solve one problem really, really well.
(Fun fact: that early bird price was quite the steal - the regular price is now $399, and they sold 10,000 units in 24 hours when it launched. Sometimes being early really does pay off.)
Who's Behind This Thing Anyway?
Before we dive into my experience, let me give you some context on who built this. The company behind Limitless is led by Dan Siroker - you might know him as the co-founder of Optimizely. But here's what makes his story compelling: his inspiration for Limitless came from his own hearing loss journey.
When he finally tried a hearing aid at 30, he described it as "gaining a superpower." That experience sparked a mission to build technology that could augment human capabilities beyond our biological limitations. And honestly? That personal connection shows in the product.
The company has raised over $33 million from some serious players - Andreessen Horowitz, OpenAI's Sam Altman, First Round Capital. But what's more interesting is their evolution: they started as Scribe (meeting bots), became Rewind (Mac app that captured everything), and finally transformed into Limitless with cross-platform functionality.
Each iteration taught them something. And that learning curve? You can feel it in how polished this thing is.
What Actually Is This Thing?
Picture a small, circular device about the size of a quarter - designed by the same folks who created Beats by Dr. Dre headphones. It clips magnetically to your shirt, or you can wear it as a necklace (my preferred method, because let's be honest, I forget to move clips between shirts).
The battery lasts about 100 hours - roughly four days of continuous use. There's a simple LED that lights up when it's recording, and here's the clever part: it is planned to have a "Consent Mode" that detects new voices and asks permission before recording them(it’s on roadmap). Smart privacy feature that could actually work.
But here's what it actually does that matters: It listens to everything around you, uploads encrypted audio to the cloud, and uses AI to create what they call "lifelogs" - intelligent summaries of your day, conversations, meetings, random thoughts you mumbled while making coffee...
Think of it as having a personal assistant who never forgets anything you've said or heard.
My Real-World Testing Journey
For the first week, I'll admit it - I was just testing accuracy. Speaking in Polish (my native language) and English, seeing how well it captured conversations in noisy environments, checking if it could handle my tendency to mumble during morning calls.
The results? Surprisingly good. English transcription is nearly perfect, Polish is solid (though obviously not as refined), and it handles multiple speakers remarkably well. The voice recognition feature kicks in after about 90 seconds of hearing you speak, then it can distinguish between what you said versus others in future conversations.
But then I discovered something more interesting...
The Self-Reflection Goldmine
After the initial "wow, it actually works" phase, I started using it for something I didn't expect: thinking out loud.
I began deliberately talking through my thoughts, ideas, plans, and tasks while walking, driving, or just sitting at my desk. Not for anyone else - just for me. The pendant captured all of it, and suddenly I had this searchable archive of my actual thinking process.
Want to remember that business idea you had three weeks ago during your morning walk? It's there. That solution to a technical problem you worked through while stuck in traffic? Captured. The pros and cons list you verbally processed about that major decision? All documented.
This became one of my primary use cases - using voice as a way to structure my thoughts, knowing that everything would be preserved and searchable later.
When I Hit the API Wall (And Broke Through It)
After a few weeks of this, I realized something frustrating: the data was trapped in their app. Sure, I could ask their AI questions about my days, and it worked well... but I wanted more control. I wanted to build my own workflows.
That's when I discovered their developer API - and everything changed.
My first instinct was to build a full app. You know how it goes - big ambitions, fancy interface, lots of features. I started coding something to connect to their API, extract tasks, create analytics dashboards, the whole nine yards.
It was a disaster. Cluttered, inconsistent data, bugs everywhere. Classic over-engineering.
So I pivoted to something smarter: Make.com automations.
The Automation Breakthrough
Instead of building an app, I built workflows. Much faster, much more practical, and I didn't need to maintain any interface.
First automation: Task extraction. Every time the pendant heard me mention something that needed to be done, it would extract that information, send it to OpenAI for proper formatting, and create a task in my to-do app. No more "wait, what was that thing I said I'd do?"
Second automation: Daily journaling. The AI creates a daily journal entry in Google Docs, highlighting key moments, conversations, decisions, and action items. I get a copy emailed to me each evening. It's like having a personal biographer who actually pays attention.
Third automation: Research trigger. When I ask about specific topics or phrase questions in certain ways, it triggers a Perplexity search, runs deep research, and emails me the results. Basically turning random verbal curiosity into structured knowledge.
Latest experiment: I connected my journal data to Claude and created a "Personal Coach" project. Now I have an AI that knows the context of my days and can help me navigate decisions and reflect on patterns. It's like therapy, but with better memory and no appointment scheduling.
Why This Beats the Built-In AI
You might wonder - why build all these automations when Limitless has its own AI built in? Good question.
The in-app AI is solid for basic queries, but it lacks broader context. It doesn't know who I am beyond what I've said recently, doesn't understand my relationships, goals, or personal context that would make its advice actually useful.
By connecting my lifelog data to external tools like Claude with custom instructions, I get AI assistance that actually knows me. It's the difference between talking to a stranger with good memory versus talking to a friend who understands your situation.
The Real Value Proposition
Here's what I've learned after a month of daily use: this isn't really about the technology - it's about changing how you interact with your own thoughts.
When you know everything is being captured, you start thinking out loud more. You process ideas verbally. You talk through problems instead of just worrying about them internally. It's like having a conversation with yourself, but with the confidence that nothing important will be lost.
The transcription accuracy, the AI summaries, the search functionality - those are all table stakes. The real magic is in how it changes your behavior around capturing and processing information.
What's Next in My Experiments?
I'm just getting started. Next up, I'm looking at MCP (Model Context Protocol) integration to connect my lifelogs directly to AI tools. I want to experiment with more sophisticated pattern recognition across my conversations and decisions.
I'm also curious about combining this with other wearables - imagine correlating conversation sentiment with heart rate data from an Oura ring, or understanding how different types of conversations affect your sleep patterns.
The API is still in beta and has some reliability issues (timeouts with large data requests, occasional sync problems), but the potential is massive. When you have a searchable, AI-processable record of everything you've said and heard... the automation possibilities are endless.
The Bottom Line
In a world full of AI gadgets that promise everything and deliver disappointment, the Limitless Pendant does something remarkable: it solves one problem extremely well.
It's not trying to replace your phone or revolutionize how you interact with the world. It's just making sure you never forget anything important again. And in doing that simple thing well, it opens up possibilities I'm still discovering.
This is what practical AI looks like. Not flashy, not overhyped, just genuinely useful in ways that compound over time.
Is it perfect? No. The $399 price point puts it out of reach for many people, there are valid privacy concerns about constant recording, and the learning curve for integration takes some work.
But for someone who lives in the digital space, who's constantly processing information and making decisions based on scattered conversations and half-remembered insights? It's been transformative.
Next up in this series: I'll be diving into some other voice AI hardware I'm testing, including some that... well, let's just say they're not all winners. But that's half the fun of being on the bleeding edge.
Stay tuned, and if you're experimenting with voice AI or memory augmentation tools, I'd love to hear about your experiences.
PS. How do you rate today's email? Leave a comment or "❤️" if you liked the article - I always value your comments and insights, and it also gives me a better position in the Substack network.
Can you share any of your workflows and such?