Your AI product,
shipped in one order.
We build the whole thing — enclosure, board, firmware, the LLM and TTS pipeline, the companion app and the backend. Six AI platforms already in mass production. Put your brand on one, or bring us a product that does not exist yet.
Six platforms you can brand this quarter
Every one of these is tooled, certified and in production. Open a card for the spec sheet and exactly what we can change for you.
Joint development from a sketch: ID, tooling, board, firmware, AI stack, app, certification. You keep the brand and the IP.
Start a specOne supplier for the plastic, the board and the AI
Most factories stop at the hardware and hand you a chip vendor SDK. We write the device app, the parent app and the server, so a change in the conversation flow does not become a three-vendor negotiation.
TV-side app for the 1920×1080 device screen, an H5 parent app for the phone, and our own Node backend. All three are ours, all three get changed for your project.
Cloud or on-premise LLM, your API key or ours, streaming TTS tuned for slow networks and old WebView builds. Voice latency is engineered, not hoped for.
EN, ZH, JA, KO, DE, FR, ES, PT, RU, PL, RO, TH, VI, ID, AR and more already shipped — including right-to-left layouts and local voice options.
Cyber Spirit AMIS, WB-PJ200 and the recorder card are all in production and out the door. Reference units and shipping records available under NDA.
Injection moulding, plush, PCBA and assembly partners we have used across multiple runs. Tooling status and capacity confirmed per project.
We handle test-house scheduling and documentation for your destination market, including battery transport certification. Scope confirmed per order.
Ask the digital human a buyer's question
This is the same conversation layer that runs on the projector, the TV stick and the plush charm — character, voice and knowledge base swapped per brand. Pick a question to see how it answers.
Live digital human — speak to it from this page
Works on desktop and phone. Allow microphone access, press to talk, and you hear the same voice pipeline that ships on the devices.
Three clicks to a requirement sheet
Pick a platform, the features you need and your volume. We generate the brief, you send it, and our engineer replies with a quote and a schedule.
From first call to container
A 30-minute call, then a written scope with pricing by volume. Usually within two working days.
Stock platform samples ship in as little as 72 hours. Branded samples follow the artwork sign-off.
Firmware, character, voice, app and packaging built out in parallel, with a weekly build for you to test.
Test-house submission, pilot run, then mass production. Typically 45–60 days after sign-off.
Book a 30-minute call with the engineer, not a salesperson
Tell us the product, the market and roughly the volume. You get a written scope, a price band and a schedule — and samples can be on a plane while the contract is still being read.
