Questions and answers
Plain answers about Tacetto. It is in development, so where something is a goal and not a fact, the answer says so.
What is Tacetto?
An AI that lives in a place instead of a device. Many small nodes, from a single sensor to a room unit with a microphone array, speaker and camera, share what they sense in one memory, so you can ask any room about any other. It is in development.
What does “an AI for places” mean?
Most assistants are one speaker that answers from what it alone heard. Tacetto is meant to belong to a place: a home, an office, a building, a factory. The place has many nodes, and the AI answers from everything they noticed together.
What is a node?
A small device that senses something or speaks. It can be a single microphone, an air-quality sensor, a power meter, a presence radar, a door sensor, a camera, or a room unit with a speaker. Any size can take part, and they can all be different.
What works today?
Working software for one room unit that listens for a wake word. The microphone array and speaker come next, then the speech services, then the part that makes Tacetto different: the shared memory between nodes. That part is a design, not yet an implementation.
Is Tacetto for sale?
No. If you want to hear when there is something real to show, join the quiet list.
Does it need the internet or the cloud?
It is being built to run inside your own network and keep working when the internet is down. The cloud is meant to be optional. Today only the room unit’s software exists, so none of this has been proven yet.
Is it listening all the time?
A small wake-word model in the room listens for one name and nothing else. The design sends audio to the brain only after it hears that name, and a physical mute button on the microphone array is the off switch. These are design goals for what is being built, not measured results.
Does it record me or my guests?
The design keeps raw audio and video on the device that captured them. Nodes share short summaries of what they noticed, such as “the garage door opened”, not the recording. Guests are told. Recording laws differ by place and by whether you are part of the conversation, so this is general information, not legal advice.
Who can see what the nodes remember?
By design, only the people the place belongs to. The memory stays inside the place’s own network. Sharing between places, such as two homes, is a policy each owner sets.
Will it work with my Alexa or Google speakers?
Tacetto is designed to speak through speakers you already own, which can play audio from a local program (Google Cast speakers) or through Amazon’s cloud (Echo). Neither gives a supported way to get microphone audio, so the listening and sensing come from Tacetto’s own nodes. This has not been tested yet.
How is it different from a smart speaker or Home Assistant?
A smart speaker answers from itself and a company’s cloud. Home Assistant is an excellent hub for connecting devices, and its voice satellites send audio to one central pipeline. Tacetto is aiming at something different: a shared memory across all its nodes, so a node in one room can answer about what a node in another room noticed. Tacetto is not built on Home Assistant.
Can it work in an office, a factory or a city?
That is a direction, not a product. The same idea scales, but each step up brings its own laws and safety rules: workplace recording and biometric rules, and strict limits on what software may do to machines.
Does it control things?
By design it reads and reports. Anything that changes a machine or a building would need a person to approve it.
Do you track visitors?
This site counts page views and load times with Cloudflare Web Analytics, which states that it sets no cookies and does not fingerprint visitors. The interest list keeps only what it needs. See Privacy.
Is it open source?
Not decided yet.
When will it be ready?
I do not have a date, and I would rather say so than guess.
Who is building it?
Last updated 2026-10-08. More questions? Use the contact page.