Voiceprint Identity
Knows whose voice this is — instantly, passively, without asking for a name or birthday. Family-member disambiguation on shared devices, and the same identity across every device.
VoiceQuest is the persistent voice identity, memory, and relationship intelligence layer for Character Physical AI. Voiceprint identity knows whose voice it is. Persistent memory knows what's happened between you. Relationship state knows where you stand — across sessions, across devices.
Conversation is becoming a commodity. Being remembered is not. VoiceQuest combines three capabilities that no pipeline vendor, text-memory tool, or security voiceprint vendor offers together.
Knows whose voice this is — instantly, passively, without asking for a name or birthday. Family-member disambiguation on shared devices, and the same identity across every device.
What's happened between you, remembered across sessions. Layered extraction and multi-signal retrieval, built for voice — not a text tool with a microphone bolted on.
Above facts sits state: familiarity, trust, interaction frequency, milestones. Exposed as a programmable relationship API so products can behave like they actually know their user.
Raw voiceprints never leave the device; the cloud stores only de-identified memory and relationship state. Child-data compliance (COPPA, BIPA, Maryland's Jan-2027 AI-toy deadline) isn't a checkbox — it's the architecture.
Whether you run a front desk or ship a million devices, the path is the same: integrate once, and the relationship asset compounds from day one.
We map your use case — front-desk calls, embedded device, or hybrid — and define what "being remembered" means for your users and your revenue metrics.
Edge SDK for on-device voiceprint and hot memory cache (runs on toy-grade SoCs), Cloud API for the long-term memory graph and relationship state. Model-agnostic — keep your own LLM.
COPPA/BIPA-ready flows, parent data-sovereignty panel, HIPAA options for clinical settings. We hand you the audit trail, not just the tech.
Every session deepens the memory graph. Retention, repeat bookings, and switching costs grow with it — and we stay on to tune routing and cost.
Our AI front desk used to just answer calls. Now it recognizes the patient before they say their name.
Returning patients are identified the moment they speak. It remembers their last treatment, their preferences, where their membership stands — and it books them without the twenty-questions routine. After-hours calls that used to go to voicemail now become appointments.
One engine, delivered three ways — Cloud API, Edge SDK, and hybrid — serving the segments where identity can't be solved by an account login.
MedSpas, clinics, and contractors: recognize returning callers instantly, recall treatment and service history, recover after-hours revenue. Full-stack voice + memory, or memory-only on top of your existing agent.
The "bored in 5 minutes" problem is a missing-relationship problem. Your device runs its own LLM; VoiceQuest adds who-is-this and what-do-we-remember — with edge-first child-data compliance as a product, ahead of Maryland's Jan-2027 deadline.
A robot with no memory has no personality. Persistent identity and memory that survives device swaps — the relationship migrates with the person, not the hardware. Design-in ready for your next product cycle.
Per-minute pricing is a race to the bottom. VoiceQuest bills on Monthly Active Relationships (MAR) and per-device licenses — so our revenue grows only when your users keep coming back.
then from $0.02 / MAR / month
multi-location tiers up to $799
per-device licensing + cloud subscription
Kids got bored with our AI toy in five minutes. With memory, it became their friend.
The toy remembers each child in the family separately — the jokes, the stories, where they left off. Retention stopped being a content problem the moment it became a relationship. And because voiceprints never leave the device, our compliance review for North America got dramatically simpler.
Why voice AI's next act isn't sounding more human — it's remembering like a friend. Analysis on identity, memory, and the physical-AI stack.
The retention gap is a missing relationship layer, not missing content.
Read → ComplianceCOPPA, BIPA, and why raw voiceprints can't go to the cloud.
Read → IndustryPipelines are in a price war; memory is a data asset.
Read → BusinessWhy MAR pricing aligns us with your retention.
Read →We're the layer between voice models and applications: voiceprint identity (whose voice is this), persistent memory (what's happened between us), and relationship state (where we stand) — across sessions and devices. We don't compete with your LLM or your voice pipeline; we make them remember.
No. Keep your pipeline and your model. VoiceQuest is model-agnostic and plugs in as the identity/memory layer — you can swap the underlying LLM anytime and the relationship asset stays intact.
Raw audio and voiceprint features never leave the device. The cloud stores only de-identified memory keyed to an anonymous ID, and parents get a one-tap view/delete/export panel. This maps directly to COPPA, BIPA, and Maryland's 2027 AI-toy requirements.
It's compressed to run on toy-grade SoCs (dollar-level BOM), with Android and embedded Linux support first and RTOS following. Offline fallback included.
The MedSpa vertical pack ships with HIPAA-ready storage, California SB 243 / AB 489 disclosure flows, and VPC / on-prem deployment options for institutional customers.
Speaker disambiguation is native: each family member (or each caller to a shop line) gets a separate memory line under the same device. That's the core problem text-memory tools can't solve.
Front-desk deployments go live in weeks. Device design-ins follow your hardware cycle — we join at product-definition stage and ship with your firmware.
Book a demo, request Edge SDK access, or join the developer API waitlist.