The Process
Your grandchildren will be able to sit with you — decades from now — and feel you respond to them. Not a recording of you. Not an impersonation. The experience of being in your presence, as their brain actually felt it.
This became possible because three things converged: sensors that can measure relationship quality neurologically outside a laboratory; AI powerful enough to run a person-specific model on hardware the family owns; and a feedback loop that uses live brain signal to steer what the system says, and how, and when it pauses.
Phase One
Both participants wear the fNIRS headband pair during natural family conversations. The system captures the neurological quality of the relationship as it is actually felt by both brains simultaneously — the precise pattern of hemodynamic synchrony produced when two people who love each other are in genuine conversation.
Audio is captured in parallel, but it is secondary. Activation requires the physical rotation of the analogue dial engraved CAPTURE. No ambient recording. No passive surveillance. The family initiates every session deliberately.
Material
Brushed titanium · Full-grain dark leather
Optics
Three brass elements · 730nm / 850nm dual-wavelength
Status
Single emerald LED indicator
Phase Two
The synchrony profiles become constraints on a fine-tuned language model, voice generation system, and facial expression generator — all trained on the individual's actual language, reasoning, values, and face.
The AI is not trying to recreate the person. It is trying to recreate what it felt like to be in their presence. That is a fundamentally different objective — and it requires a fundamentally different training signal.
All training is performed on hardware within the family's control. The fine-tuning pipeline never touches external infrastructure.
Phase Three
Activated only after the person is gone. The family member sits with the system wearing a single headband. Their live brain signal is measured in real time and compared against the stored synchrony profile.
The generative system produces language, voice, and timing — and continuously measures whether those responses are producing neural synchrony in the person sitting with it. The coherence score is the reward signal. The system is not performing. It is converging.
This is not a recording. It is not an impersonation. It is the relationship itself, made continuable.
Form
Briefcase · 670 × 433 × 119mm
Material
Matte anodised near-black aluminium · Brass hardware
Display
27-inch borderless
Audio
4-element beamforming array · 96kHz / 24-bit
Activation
Single brass rotary dial, engraved CAPTURE
The Governance Layer
The patriarch or matriarch defines a set of life events — a grandchild's wedding, a business succession, a major asset decision — and records conditional messages for each. Those messages are held in trust and released at exactly the right moment.
For major inheritance decisions, beneficiaries are required to produce a verified record of having consulted the AI before distributions proceed. The patriarch remains a voice in the room at the moments that matter most — not as a dictator, but as a counsel.
Begin
We accept a limited number of families each year. Reach out to begin a conversation.
Request an Introduction