Honestly, this isn't my first attempt at building myself a voice assistant. The first attempt was 4 months ago. I spent about a week and couldn't build anything decent - I didn't have the experience. Back then I was building everything in OpenClaw straight through Telegram. It ended in a flop 😅
But time passed, and I got a client who asked me to build him an agent like Siri, only smarter. So you could talk to it by voice like in the movies and it would do what you ask. Before building anything for a client, I built one for myself
And... Meet Jes!
This is NOT a chatbot in the "send a message and wait three days" style. It's a live voice I talk to. In Russian. Like with a real person - one you can interrupt mid-word. I talk - she answers instantly. You can rattle on, you can mumble, you can cut in in the middle of her sentence - she'll get it and pick it up
But the voice is just the wrapper. Now comes the juiciest part 👇
🎛 She runs my whole system
I've got a whole system grinding away - bots develop, write texts, draw images, crunch numbers, monitor. I used to dig around in it by hand, now I just say:
- "Jes, give the designer a task to draw an image for the voice assistant presentation"
- "Setting up the task now"
And she went and did it. Gave the task to the bot, found out which one and reported back on the result - by voice, like in a freaking movie!
🧠 And she REMEMBERS
Memory on several levels:
▪️ Remembers important facts and decisions - across conversations, doesn't lose them
▪️ At the start of every conversation she already knows what we talked about yesterday - makes a short recap herself, doesn't make you retell it
▪️ Searches all our past conversations by MEANING, not by exact words ("where did we decide that payment thing?" - finds it, even if you don't remember it word for word)
▪️ Digs through my knowledge base like it's her own pocket
I say: "Jes, remember that..." - and next time she already knows it
📲 And she's EVERYWHERE
On my phone - I put the icon on the home screen, it opens like a real app. On my MacBook - she lives in the Dock as a separate window: click - and you're talking. Pull out the phone - Jes is right there. No installs, no app stores, none of that fucking hassle with "update to version X"
⚡️ And the small stuff:
🌐 Searches the internet, GitHub, documentation - by voice, and reads out the gist, not links
⏰ Sets reminders and timers - at the right moment says it out loud herself + duplicates it in Telegram
📬 Reads email and calendar, sends emails
🌦 Tells the weather, news and a whole bunch of other stuff
Under the hood - a live realtime engine from OpenAI, my own multi-level memory and a commander-orchestrator that pulls the bots' strings
For ten years we were promised "smart assistants", but in reality we got sold push-button answering machines that "didn't catch that request" 🤡 And here - a live voice that runs a whole system, remembers everything, doesn't get confused and is always at hand. Not a demo, not a concept, not "coming soon". It works. Right now
The future isn't "someday later". It's already in my pocket. And her name is Jes 🔴