A private conversation deserves a private record.
Meet Kinig, a meeting pod designed to keep AI local.
Kinig / kee·nig
Filipino for listen.
Client details. Unfinished strategy. The decision that changes everything. Meetings hold information that deserves deliberate control.
Taking notes divides your attention. A recording leaves another hour to replay. And sending sensitive discussions to a cloud service is not always an option.
We envision a better way: one pod on the table, useful notes after the meeting, and your organization in control of the record.
One shared transcriber. A visible recording state. A screen for local access. Every part designed around the room.
A quiet presence at the center of the table. Inside, every layer has a role in turning a conversation into a record your company controls.
Scroll to open the pod, follow the signal, and see how the parts could evolve.
A perimeter microphone array gives the room one shared pickup point. The highlighted ring is where the conversation enters the pod.
The ring lifts away as a serviceable assembly. Optional satellite microphones could extend capture across a larger table.
The compute module is intended to turn audio into a transcript, decisions, and action items on company controlled hardware.
The board connects to the pod through a removable interface. Upgrade the computing capability while keeping the enclosure and other modules.
The screen makes the session visible and displays a QR invitation to request the notes. Scanning is not permission by itself.
The host reviews the draft and approves access. A removable display module keeps a damaged screen from retiring the whole pod.
A separate power and storage layer sits beneath the compute board. Keeping these parts accessible supports maintenance over the life of the device.
A battery service or storage repair should not require replacing every component. Retention, deletion, and encryption still need to be engineered.
Three recessed sockets follow the curve of the aluminum shell: USB C, an audio connection, and a dedicated accessory connector.
A replaceable interface board would connect power and peripherals without redesigning the entire pod. Final standards and electrical specifications remain open.
The compute module slides out. Microphone and speaker extensions connect around the pod. The display, pickup ring, power layer, and interface board remain distinct assemblies.
Further ahead: a cartridge with a trained model embodied in specialized silicon. A model upgrade could mean a module upgrade, not a new pod.
Designed to give everyone their attention back, with a clear path from a spoken thought to an approved record.
The host announces the session. The pod makes its recording state visible.
Local AI creates a transcript, key decisions, and action items for the host to review.
Scan the screen to request local access. The host approves who receives the notes.
September 10, 2026 · Meeting room A
The team aligned on a smaller discovery phase before expanding the client engagement. The revised proposal will focus on two priority workflows.
Key decisionsPrivate by intention. Local by design.
A shared device for teams whose conversations carry business value.
For consulting and advisory teams discussing sensitive briefs, client details, and next steps.
For leadership and product teams shaping plans that are not ready to leave the room.
For internal training and handovers where the spoken detail matters as much as the slides.
We are exploring a model where companies own their meeting infrastructure and upgrade the parts that need to grow.
A focused first product. An ambitious modular future.
Build an offline proof of concept. Evaluate English and Taglish, note quality, and local delivery.
Co design a business pilot. Test useful output, host controls, processing time, and willingness to pay.
Explore replaceable compute, microphones, speakers, screens, and ports in a serviceable enclosure.
Imagine a cartridge with a trained model embodied in specialized silicon. A new generation of intelligence could mean a new module, while the rest of the pod stays.
Inspired by research into models implemented directly in hardware, including Taalas, with no affiliation. Pocket scale performance, power, cost, and availability remain open engineering questions.
Explore the technical precedentUseful technology should give people more room to think. We are a team exploring what that looks like, together.
Turning a clear problem into a product worth building. Product direction, customer discovery, and partnerships.
Making local intelligence useful and accountable. Model evaluation, inference, and security architecture.
Bringing intelligence into the physical world. Electronics, embedded compute, and power systems.
Connecting the pod to the people around it. Device software, local access, and user experience.
Giving the idea its physical form. Industrial design, modular mechanisms, and prototyping.
Proposed team roles.
What we intend to build.
What we still need to prove.
Kinig is a product concept. We are seeking development and pilot partners. The renders and meeting walkthrough show our intended experience, not working hardware. No audio is recorded or processed, and no notes are actually transferred by this page.
The core workflow is designed to capture audio, transcribe, and generate notes on the device. Approved notes would transfer over a local connection. Demonstrating this complete offline workflow is our first engineering milestone.
No. In the proposed workflow, the QR code starts a local access request. The host approves participants before sharing reviewed notes. The code itself is not a substitute for authorization.
Our intended design gives the organization control over access, retention, and deletion on its hardware. Those controls, including handling a lost device, must be engineered and tested before a pilot.
Both are evaluation targets. We plan to measure transcription accuracy separately and test whether generated decisions and action items reflect the discussion. We have no measured accuracy to report yet.
A shared pod could combine deliberate microphone placement, company controlled processing, and a consistent meeting workflow. A pilot must establish whether that combination offers enough value over existing apps and local software.
Our roadmap starts with serviceable hardware and a replaceable compute module. Longer term, we want to explore cartridges containing models embodied in specialized silicon. Microphones, speakers, screens, and ports are further extension concepts.
Pricing is open. We are exploring device sales or leases, support, and modular upgrades. Component costing and conversations with prospective buyers will inform the offer.
We are looking for businesses willing to help define what private meeting intelligence should become.
Download the pilot brief