Loading video...

Video Failed to Load

Go Home

This guy fit a barista AI inside an $8 ESP32. Ask it an espresso question over USB. It answers on a tiny OLED. Fully local. No cloud. No GPU. Tiny specialized AI is getting interesting. Repo below.

minchoi's profile picture

Min Choi

381,730 subscribers

51,718 views • 25 days ago •via X (Twitter)

Science & TechnologyEducation
LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

just generated an ai video on my mac no cloud, no server bill last year this needed a gpu cluster
0:36

just generated an ai video on my mac no cloud, no server bill last year this needed a gpu cluster

Igor Tarasenko

427,139 views • 1 year ago

An 8-year-old turned an $8 microcontroller into a fully offline, voice-controlled AI storyteller. No Wi-Fi. No cloud. No API. Using an ESP32-S3, INMP 441 mic, i2s speaker and Max 98357A amplifier, Laksh built a bot that generates stories, creates animations, reads them aloud and responds to voice commands: play, pause, resume and next story. He built it by forking the brilliant open-source TinyStories LLM for ESP32 created by Slava S.. It had text. Laksh added voice, interaction, controls and animation. Git repo: Current drawback: bad English TTS on the ESP32. So that is what we are building next.
1:32

An 8-year-old turned an $8 microcontroller into a fully offline, voice-controlled AI storyteller. No Wi-Fi. No cloud. No API. Using an ESP32-S3, INMP 441 mic, i2s speaker and Max 98357A amplifier, Laksh built a bot that generates stories, creates animations, reads them aloud and responds to voice commands: play, pause, resume and next story. He built it by forking the brilliant open-source TinyStories LLM for ESP32 created by Slava S.. It had text. Laksh added voice, interaction, controls and animation. Git repo: Current drawback: bad English TTS on the ESP32. So that is what we are building next.

Capt Venkat 🇮🇳

165,766 views • 1 month ago

meet WALL-E no, not the one from the cartoon :) this is a fully 3D-printed, autonomous desk companion running on an ESP32. it uses a local AI voice assistant to process commands in real-time and control its own motors building intelligent robotics just dropped to zero cost...
0:25

meet WALL-E no, not the one from the cartoon :) this is a fully 3D-printed, autonomous desk companion running on an ESP32. it uses a local AI voice assistant to process commands in real-time and control its own motors building intelligent robotics just dropped to zero cost...

ard

65,781 views • 26 days ago

OSINT + AI 🧠 Vyntral's God’s Eye integrates local AI via Ollama for vuln analysis and reporting. No cloud. No APIs. Fully private. Free. AI is becoming a default layer in open source tooling, not a premium feature. Try it here:
0:35

OSINT + AI 🧠 Vyntral's God’s Eye integrates local AI via Ollama for vuln analysis and reporting. No cloud. No APIs. Fully private. Free. AI is becoming a default layer in open source tooling, not a premium feature. Try it here:

The OSINT Newsletter

53,194 views • 7 months ago

btw this robot runs on a fully open source AI framework called Xiaozhi it turns a $8 ESP32 chip into a fully autonomous voice assistant NO expensive cloud servers or high latency wake-word detection and audio streaming happen locally on the chip: > requires only an ESP32-S3, a microphone, and a basic speaker > supports real-time conversational voice interaction > completely open-source and customizable for any DIY robotics project anyone can now build a personal AI companion for the price of a cup of coffee. (github link in the replies)
0:25

btw this robot runs on a fully open source AI framework called Xiaozhi it turns a $8 ESP32 chip into a fully autonomous voice assistant NO expensive cloud servers or high latency wake-word detection and audio streaming happen locally on the chip: > requires only an ESP32-S3, a microphone, and a basic speaker > supports real-time conversational voice interaction > completely open-source and customizable for any DIY robotics project anyone can now build a personal AI companion for the price of a cup of coffee. (github link in the replies)

ard

42,210 views • 26 days ago

If, as everyone is claiming, this motion to bar Jeremy Corbyn from standing as a candidate is about antisemitism, why does it only mention the election loss? Would be an interesting question, with some uncomfortable answers, but no one wants to ask it.
0:27

If, as everyone is claiming, this motion to bar Jeremy Corbyn from standing as a candidate is about antisemitism, why does it only mention the election loss? Would be an interesting question, with some uncomfortable answers, but no one wants to ask it.

Ash Sarkar

296,256 views • 3 years ago

Voice request to an OpenHome speaker. A local AI books a real Delta flight end to end on a fresh Minisforum X1 iGPU. No human clicks. No cloud agent. 73 seconds. The video is the demo. Numbers below. 🧵
1:12

Voice request to an OpenHome speaker. A local AI books a real Delta flight end to end on a fresh Minisforum X1 iGPU. No human clicks. No cloud agent. 73 seconds. The video is the demo. Numbers below. 🧵

cCross

524,633 views • 3 months ago

Merge it. Taylor demos the full SDLC of 2026: Nightwatch catching an error, AI fixing it and opening a PR, Cloud creating a preview environment, OpenClaw🦞 calling to ask if he wants to merge the PR. No code touched. Fully self-healing app.
1:02

Merge it. Taylor demos the full SDLC of 2026: Nightwatch catching an error, AI fixing it and opening a PR, Cloud creating a preview environment, OpenClaw🦞 calling to ask if he wants to merge the PR. No code touched. Fully self-healing app.

Laravel

1,844,480 views • 5 months ago

AMD on local AI: "this is the smallest AI development system in the world, capable of running models up to 200 billion parameters locally, not connected to anything" 128GB of unified memory shared by CPU, GPU, and NPU fits in your hand. Runs a model larger than GPT-3 with no internet $3,999 once. roughly $16 a month in power Cloud equivalent: up to $750 a month The question was never whether local AI could compete on power It was whether it could compete on price. Now it does
1:00

AMD on local AI: "this is the smallest AI development system in the world, capable of running models up to 200 billion parameters locally, not connected to anything" 128GB of unified memory shared by CPU, GPU, and NPU fits in your hand. Runs a model larger than GPT-3 with no internet $3,999 once. roughly $16 a month in power Cloud equivalent: up to $750 a month The question was never whether local AI could compete on power It was whether it could compete on price. Now it does

plutos

152,392 views • 2 months ago

I tested Gemini 3.1 Flash Live and one thing stood out immediately: the speed. You ask a question. It answers almost instantly. No awkward delay. No clunky back-and-forth. Then it gets more interesting. You can: use voice mode, use webcam mode, share your screen, connect tools, add system instructions, and publish the app to the cloud. So this is not just “voice chat with AI.” It is the beginning of real-time AI agents you can actually ship.
18:11

I tested Gemini 3.1 Flash Live and one thing stood out immediately: the speed. You ask a question. It answers almost instantly. No awkward delay. No clunky back-and-forth. Then it gets more interesting. You can: use voice mode, use webcam mode, share your screen, connect tools, add system instructions, and publish the app to the cloud. So this is not just “voice chat with AI.” It is the beginning of real-time AI agents you can actually ship.

Julian Goldie SEO

21,000 views • 5 months ago

most people don’t have a content idea problem they have an attention loop problem: - one-off videos - no recurring character - no emotional formula - no comment bait - no promo angle AI brainrot works when it stops being random and starts becoming a tiny media system full guide below
1:12

most people don’t have a content idea problem they have an attention loop problem: - one-off videos - no recurring character - no emotional formula - no comment bait - no promo angle AI brainrot works when it stops being random and starts becoming a tiny media system full guide below

Penguin

13,322 views • 3 months ago

“BRIAN, IT CANT BE DONE” Done: A low resolution holographic of the entire corpus of online information. A fully local AI with no Internet Connection running on 4 models back, iPhones. Next up YOUR AI trained on YOUR local data, to become an active part of your life.
0:45

“BRIAN, IT CANT BE DONE” Done: A low resolution holographic of the entire corpus of online information. A fully local AI with no Internet Connection running on 4 models back, iPhones. Next up YOUR AI trained on YOUR local data, to become an active part of your life.

Brian Roemmele

269,226 views • 10 months ago

Advice for AI engineers 💡 A local AI assistant is just a while loop, an LLM and a set of tools. Here's a step-by-step lesson on how to build one from scratch with Liquid AI LFM2-24B-A2B running on llama.cpp. No cloud APIs. Enjoy ↓
57:09

Advice for AI engineers 💡 A local AI assistant is just a while loop, an LLM and a set of tools. Here's a step-by-step lesson on how to build one from scratch with Liquid AI LFM2-24B-A2B running on llama.cpp. No cloud APIs. Enjoy ↓

Pau Labarta Bajo

15,463 views • 3 months ago

So this guy plugged an Ai to a robot and ask the Ai to shoot him, but the Ai refused... Until it actually shoot him 😂 We are so fucked
1:05

So this guy plugged an Ai to a robot and ask the Ai to shoot him, but the Ai refused... Until it actually shoot him 😂 We are so fucked

Jon Hernandez

1,393,102 views • 8 months ago

A 19-year-old student built a little robot that goes to lectures and takes notes so they can sleep in class It is a tiny screen with an animated smiley face, sitting on a 3D-printed stickman body You just tap its face at the start of class It listens through a built-in mic, records the professor, and as soon as class ends, turns the entire lecture into a structured summary right on the display Under the hood, it runs on an ESP32 microcontroller, an OLED screen, a small battery, and an SD card The best part - the student did not even code it line by line. The entire codebase was written using Cursor AI Audio goes in, gets transcribed, AI structures the notes, and it is ready before you even leave the room Most students are still struggling to take notes while half asleep. This little guy just sits on the desk and handles it all
0:23

A 19-year-old student built a little robot that goes to lectures and takes notes so they can sleep in class It is a tiny screen with an animated smiley face, sitting on a 3D-printed stickman body You just tap its face at the start of class It listens through a built-in mic, records the professor, and as soon as class ends, turns the entire lecture into a structured summary right on the display Under the hood, it runs on an ESP32 microcontroller, an OLED screen, a small battery, and an SD card The best part - the student did not even code it line by line. The entire codebase was written using Cursor AI Audio goes in, gets transcribed, AI structures the notes, and it is ready before you even leave the room Most students are still struggling to take notes while half asleep. This little guy just sits on the desk and handles it all

pulmencr

56,123 views • 28 days ago

What Lisa Su actually held on stage: A mini PC the size of a lunchbox running Qwen3-235B locally, with no cloud and no discrete GPU Inside: the Ryzen AI Max+ 395, 128GB unified memory, 110GB usable as VRAM on Linux The first x86 chip that handles 200+ billion parameters on a single die AMD claims it beats the RTX 5080 by several times on memory-bound models — because the 5080 simply cannot fit them $1,400 to $2,500 once. cloud bills run $200 to $400 a month It pays for itself in a few months, then costs nothing per request This is not a faster GPU. it is the first real argument that your AI does not belong in someone else's data center
0:31

What Lisa Su actually held on stage: A mini PC the size of a lunchbox running Qwen3-235B locally, with no cloud and no discrete GPU Inside: the Ryzen AI Max+ 395, 128GB unified memory, 110GB usable as VRAM on Linux The first x86 chip that handles 200+ billion parameters on a single die AMD claims it beats the RTX 5080 by several times on memory-bound models — because the 5080 simply cannot fit them $1,400 to $2,500 once. cloud bills run $200 to $400 a month It pays for itself in a few months, then costs nothing per request This is not a faster GPU. it is the first real argument that your AI does not belong in someone else's data center

plutos

1,237,247 views • 2 months ago

Researchers are using ChatGPT to generate content for their papers...and are getting caught. Want to avoid this kind of embarrassement? Use Consensus, it has an AI-powered Copilot integrated with academic databses. Ask it to generate an outline for a literature review. It will give you an outline with citations to published research. No nonsense like "As an AI language model..."
6:01

Researchers are using ChatGPT to generate content for their papers...and are getting caught. Want to avoid this kind of embarrassement? Use Consensus, it has an AI-powered Copilot integrated with academic databses. Ask it to generate an outline for a literature review. It will give you an outline with citations to published research. No nonsense like "As an AI language model..."

Mushtaq Bilal, PhD

122,389 views • 2 years ago

🚨🇺🇸 VP VANCE: I'M A GROK GUY “I’m a Grok guy. I think it’s the best. It’s also the least woke. I’m always asking myself, if I ask an AI a question, is it gonna give me an objective answer or is it gonna give me a woke answer.” Source: Fox Vice President JD Vance
0:46

🚨🇺🇸 VP VANCE: I'M A GROK GUY “I’m a Grok guy. I think it’s the best. It’s also the least woke. I’m always asking myself, if I ask an AI a question, is it gonna give me an objective answer or is it gonna give me a woke answer.” Source: Fox Vice President JD Vance

Mario Nawfal

59,248 views • 9 months ago

YOUR OLD ANDROID PHONE IS A FREE AI VISION SERVER AND YOU DON’T KNOW IT one dev built an app that turns any android into a local openai-compatible LLM server with vision support, analyzes camera feeds entirely on-device, no cloud, no subscriptions, nothing leaves your network you already own the hardware, it’s just sitting in a drawer full breakdown below ↓
1:03

YOUR OLD ANDROID PHONE IS A FREE AI VISION SERVER AND YOU DON’T KNOW IT one dev built an app that turns any android into a local openai-compatible LLM server with vision support, analyzes camera feeds entirely on-device, no cloud, no subscriptions, nothing leaves your network you already own the hardware, it’s just sitting in a drawer full breakdown below ↓

leopardracer

21,844 views • 2 months ago

another example: a developer used the exact same $8 ESP32-S3 chip and local AI setup to build a real-time object recognition camera the fact that this is running entirely offline, on an eight-dollar microcontroller, without any cloud API calls, is wild... it captures the video feed, processes it locally through the model, and identifies objects in real-time. we are entering an era where you can just glue intelligence onto any physical object for less than $10. the hardware barrier is completely shattered.
0:39

another example: a developer used the exact same $8 ESP32-S3 chip and local AI setup to build a real-time object recognition camera the fact that this is running entirely offline, on an eight-dollar microcontroller, without any cloud API calls, is wild... it captures the video feed, processes it locally through the model, and identifies objects in real-time. we are entering an era where you can just glue intelligence onto any physical object for less than $10. the hardware barrier is completely shattered.

ard

224,247 views • 1 month ago