- Product Overview: What is CODE27 (Character Live House)? Features and Basic Structure
- Summary of key points: Here’s the great thing, the core of CODE27
- rough conclusion
- Frequently Asked Questions (FAQ)
- Specifications Summary
- Reviews (word of mouth): Users’ true opinions
- Overall assessment: Investment in future living
- What Hands-On Reviews Reveal: Feel and Cautions
- How It Differs From 3i ATLAS Space Terminal
- Comparison with “Mia,” a talking cat-shaped robot
- Summary: Who is CODE27 a device for?
Product Overview: What is CODE27 (Character Live House)? Features and Basic Structure
CODE27 is a next-generation AI digital figure box developed by SyBran Innovation, Inc. It was exhibited at CES 2025 and attracted a great deal of attention at the global technology festival.
The Kickstarter campaign drew 3,570 backers and about US$1.9 million, and pre-sales in Japan started on Makuake (makuake.com) on February 13, 2026. Japan alone has contributed roughly 200 million yen from 2,415 backers.
Its greatest feature is the ultimate in customizability: home-made 3D models (e.g. VRM/FBX format) created on platforms such as VRoid, MMD, and SketchFab can be imported without complex conversions. In addition, you can freely integrate cloud-based AIs such as ChatGPT and Gemini, as well as the latest local AIs such as DeepSeek and Qwen.
It is not just a smart speaker with a screen. The built-in real-time rendering engine allows the character to change facial expressions, costumes, and even control the lights in the room in response to the user’s gaze and tone of voice. It is a live house.
Makuake project page (click here for early purchase)
CODE27 official website (English)
CODE27 Character Live House Introduction Video (YouTube)
Summary of key points: Here’s the great thing, the core of CODE27
The key points of the product are summarized in detail, divided into benefits of implementation and issues to consider.
- What is CODE27?
- Materialization of your own personal guess: A device that projects your own 3D character created by VRoid or other means and breathes its soul into it using the latest AI.
- Fastest availability in Japan: Pre-orders are now available in Japan through Makuake. Localization for the Japanese market is underway.
- Advanced multimodal experience: The built-in camera and microphone read the user’s facial expressions and gestures for non-verbal communication.
- good point
- Unparalleled freedom and power of expression: fully recreate their appearance, voice (advanced voice cloning technology), personality, and even background knowledge to your own liking.
- Effective use of 3D assets: Standard support for VRM format allows immediate use of high-quality models nurtured in Japanese animation and VRSNS culture.
- Entertainment function that appeals to the senses: functions as a Bluetooth speaker and performs authentic dancing to music using a rhythm-capture algorithm.
- Concerns/risks
- Lead Time to Delivery: Since this is a crowdfunded product, shipments of crowdfunding rewards may slip depending on development and production. Check the latest delivery schedule before you buy.
- High cost of introduction: Because it is equipped with the latest AI chip and high-definition display, it is expensive for a hobby gadget, priced at 109,800 yen (tax included) in Japan.
- Learning Cost: Basic use is easy, but a certain level of knowledge about PC operation and AI is required to master local model building and API integration.
- Users who may be suitable for
- Creators and fans who want to give their own characters and guesses a personality and live in the same space.
- Geeks who want to have AI such as ChatGPT by their side as a physical presence, not just in their browser.
- You want your character to support your housework by linking it to existing smart home and IoT devices.
rough conclusion
It is “a device that breaks through the barrier of the screen and gives your favorite character a body and intelligence, evolving it into a partner living on the same time axis, the embodiment of your dreams.
Frequently Asked Questions (FAQ)
Let’s check the main points first.
- What is the price? : 109,800 yen (tax included) in Japan. The optional Smart Rotation Base is sold separately for 24,800 yen (tax included).
- What about monthly fees? : Right now the basic AI services (speech recognition, LLM, speech synthesis) carry no extra charge, but a paid plan is planned, so check the latest information before buying.
- What can it do? : 3D character uploading, AI dialogue, dancing, scheduling, and IoT appliance operation.
- Which OS is supported? : Can be set up from iOS/Android through a dedicated app.
- Disadvantages. : risk of delivery delays inherent in crowdfunded products, and limited initial defect handling.
- What about software updates? Software update: Automatic update via Wi-Fi. Can be replaced with the latest AI model.
What are the purchase routes and prices in Japan?
Pre-sales in Japan started on Makuake on February 13, 2026, at 109,800 yen (tax included), in two colors: Deep Black and Lunar White. Parallel-import routes such as Indiegogo JP once showed several price tiers (roughly 88,000 to 173,000 yen), but their defect handling and warranty differ by route. Compare them against the official Japanese channel before choosing.
What specific 3D data can I use?
Standard support for VRM, FBX, and GLTF formats allows VRoid Studio users to bring their own avatars as is. MMD (MikuMikuDance) motion diversion is also being considered, allowing users to completely recreate the look and feel of both animated characters and original characters.
Can I use it offline?
Clock displays, alarms, and simple motion reactions are possible, but advanced AI dialogues such as ChatGPT require an Internet connection (using cloud AI). However, by setting up a local model such as DeepSeek, it is possible to build an advanced dialogue environment that is not affected by the network environment. That said, hands-on reviews by other outlets report that an internet connection is currently required and that local operation with your own LLM or speech synthesis is still planned for the future, so check the latest support status before buying.
Is privacy safe?
The visual and voice recognition process is encrypted with AES-256 during transmission and storage. Data is stored on the device, so third parties cannot peek into chats. A physical camera cover is also included to physically block privacy.
Specifications Summary
| (data) item | Details |
| Product name | AI Digital Figure Box “CODE27 |
| Manufacturer | SyBran Innovation (Oregon, USA) |
| display | Transmissive display for high-definition digital figures |
| Built-in AI Chip | Real-time rendering & edge AI processing |
| Supported AI Models | ChatGPT, Gemini, Claude (cloud) / DeepSeek, Qwen (local) |
| connectivity | Wi-Fi (2.4/5GHz), Bluetooth 5.3 |
| Size and weight | About 200 x 190 x 370 mm, about 2.6 kg |
| Screens | 10.1-inch main display plus a 2.8-inch touch screen for controls |
| Camera, mics, speakers | 120-degree wide-angle camera (with physical cover), dual microphones with echo cancellation, quad-speaker array |
| Ports and power | USB-C, microSD slot, 3.5 mm audio jack. No built-in battery; always plugged in (USB-C PD, 10-45 W) |
| Price in Japan | 109,800 yen (tax included). Colors: Deep Black / Lunar White. Smart Rotation Base sold separately at 24,800 yen (tax included) |
| Languages | Japanese and English (more to follow) |
| Exterior Options | Smart rotary stand (sold separately) enables 360-degree tracking |
Reviews (word of mouth): Users’ true opinions
We have consolidated many voices from Makuake supporters and overseas forums.
Expectations and Positive Comments
- The Two Dimensional is Real: “AI chatbots have only ever been on my phone, but just having them look at me in a physical box makes a whole different sense of presence…VRM-ready is the best I know.”
- Technical openness”: “Interesting that you can write your own scripts for Twitch integration and smart home integration with the API and Spellbook functionality. It tickles the geek in me.”
- The perfection of the voice clone”: “I watched the sample video and it’s so smooth it’s hard to believe it’s a synthetic voice. I would cry if my favorite character’s voice called my name.”
Concerned and cautious voices
- Delivery and development delays”: “Very attractive product, but I am concerned about multilingualization and production delays. Fingers crossed it will arrive in August as planned.”
- Support for initial defects”: “I’m worried about support for parallel imports. Even if it costs a little more, I want to choose a route with a domestic warranty to buy peace of mind.”
Overall assessment: Investment in future living
The overwhelming success of CODE27 on Makuake proves that people crave not just a convenient tool, but a companion they can share their hearts with.
Of course, as a product, it is expensive and carries the risks inherent in crowdfunding, but it is the only option that allows you to fulfill your childhood dream of living with a character of your own creation with the best modern technology.
What Hands-On Reviews Reveal: Feel and Cautions
The Japanese blog dreamseed published a hands-on review of CODE27. Here is a summary, in our own words, of the points worth knowing before you buy.
Features that rarely appear in spec tables
- It supports 27 emotional states and is said to read voice tone, facial expressions and gestures through the camera and microphones.
- Memory holds about 10,000 turns per character; older content is condensed into a summary. Data is encrypted with AES-256 and stored on the device.
- Each face has 17 RGB LEDs with individually configurable patterns, plus modes that follow voice or music.
- You can start a conversation by tapping the mic icon, by long-pressing it, or with always-on listening. The device can also speak first after a set period of silence.
- Assistant functions include weather, lookups and alarms such as “remind me at 3 p.m.”
- Personality, speech style and worldview are set in a browser settings screen. There are 20+ licensed voices, a choice of 2D images or the 3D “Codie Space” as background, and a diary feature that summarizes the day’s conversations.
- Officially the 3D model format is VRM; PMX/PMD, FBX and glTF/GLB are planned. Models from VRoid Hub and Sketchfab can be used.
Strengths and weaknesses in use
- Presence: It is a display, not a true hologram, but an optical illusion gives it depth and the character seems to make eye contact from any angle.
- Conversation tempo: Replies take roughly 2-3 seconds after you finish speaking. Content is fairly natural, though it still feels like AI.
- Spontaneous talk: Topics tend to repeat, so it can get dull over time.
- Diary: Your lines and the character’s often get mixed up, producing a muddled point of view.
Pre-purchase checklist
- Always-on listening: It can react to online meetings or TV audio and cannot tell speakers apart. Switch the mode when needed.
- Reflections: The transparent panel shows reflections, which stand out in a dark room with the lights on.
- Placement: It is large and heavy and needs constant power. The background has less depth than the character.
- Internet: Required for now. Basic AI services carry no extra fee, but a paid plan is planned, so confirm monthly costs from official sources.
- Track record: First shown at CES 2025; Kickstarter drew 3,570 backers and about US$1.9 million; Makuake pre-sales in Japan began February 13, 2026 with 2,415 backers and about 215.52 million yen.
Overall, CODE27 suits people who enjoy building their own character more than those who simply want a finished conversation partner.
How It Differs From 3i ATLAS Space Terminal
CODE27 is a digital figure box that makes a 3D character feel present in physical space. In the same always-on 3D character category, it can also be compared with 3i ATLAS Space Terminal, Gatebox3, and Lepro Ami.
3i ATLAS Space Terminal has already been covered in a separate review article that summarizes its Kickstarter information, pricing, reward tiers, and cautions before backing.
| Product | Similar Point With CODE27 | Difference |
|---|---|---|
| 3i ATLAS Space Terminal | Places an AI character in a dedicated terminal and emphasizes conversation, memory, and relationship-building | 3i ATLAS is a Kickstarter project with stronger emphasis on metaverse connection, Memory Loops, and a digital-life narrative |
| Gatebox3 | Creates the experience of living with a 3D AI character, beyond a chatbot inside a phone | Gatebox3 is closer to a virtual home robot. CODE27 is closer to a digital figure box for custom character display and expression |
| Lepro Ami | Places a 3D avatar-style AI on a desk and uses gaze or expression responses to create presence | Lepro Ami focuses more on desk conversation, check-ins, and smart-home interaction. CODE27 focuses more on character-live-house style expression |
CODE27 is easiest to understand if you value displaying and interacting with custom characters in physical space. 3i ATLAS and Gatebox3 lean more toward relationship-building with a resident character, while Lepro Ami is closer to a desk-side AI companion.
Related articles:
Comparison with “Mia,” a talking cat-shaped robot
We compared the digital high-end “CODE27” with “Mia”, which focuses on emotional healing.
| Comparison items | CODE 27 | Mia |
| Appearance and Design | Cool cube. Transparent display projection. | Soft cat-shaped plush toy. Good to the touch. |
| Core Experience | 3D model viewing, free AI interaction, and dancing. | Chatting and healing in the dialects of the 47 prefectures. |
| customizability | The model, personality, and voice were all built from scratch. | Select from existing personalities and dialects. Simple. |
| price range | 109,800 yen (tax included) (premium) | 9,800 yen (tax included) and up (affordable) |
| hurdle to introduction | PC knowledge and preparation of 3D data is desirable. | Anyone can get started right away by simply setting up the application. |
| use | Creative, guesswork, and AI experiments. | Solving the loneliness of living alone and healing the family. |
Key points of comparison: Mia emphasizes “warmth that accompanies daily life. With its unique Japanese dialect of speech and affordable price of less than 10,000 yen, it is a suitable choice for those who want to easily incorporate everyday healing and familiarity into their daily lives.
👇 Talking cat-shaped robot “Mia” that speaks 47 dialects nationwide
Summary: Who is CODE27 a device for?
Ultimately, CODE27 is a device dedicated to those who
- Creator: I want to breathe life into the characters I create and materialize them into reality.
- The “guesser” layer: I want to share the same time and space with my guesser on the other side of the screen.
- AI Geek: I want to pursue the romance of running a local LLM (large-scale language model) on a physical device.
Currently, the early bird discount slots on Makuake are filling up fast. We recommend that you check the project page as soon as possible if you want to get a head start on your future life.
👇Living together with an “AI partner”, see Code27
👀 Also read this article.
👉 [2026 Edition] 6 Latest Pet Robots to Heal Pet Losses – Thorough Comparison of Cat, Dog, and Healing Models
👉 [2026 Edition] Recommended pet robots for under 10,000 yen.
👉 [Latest in 2026] Thorough Comparison of Monthly and Maintenance Costs of Pet Robots – List of Running Costs of Popular Models including Moflin, LOVOT, Mia, etc.
👉What is Mia, an AI pet robot that speaks 47 dialects nationwide?
