One AI human, assembled from four honest layers.
Gapsmania Humans combines a consented face, a consented voice, grounded knowledge, and permissioned actions into a single talking agent you deploy on infrastructure you control.
System diagram
A layered identity, not a black box
Each layer is independently configurable and independently revocable.
Consent-based portrait-to-avatar identity, rendered live from a verified enrollment.
Cloned voice with revocable consent, streamed over low-latency media transport.
Grounded answers pulled from your connected knowledge, with citations.
Approved workflows and authenticated tool calls, gated by human review.
Getting started
Enroll → Teach → Connect → Publish
Enroll
Capture consented face and voice samples.
Teach
Connect documents, FAQs, and CRM sources.
Connect
Wire up approved tools and webhooks.
Publish
Go live on web, mobile, or kiosk surfaces.
Infrastructure
Self-hosted by design
Separate the SaaS control plane from GPU inference and real-time media transport, and bring your own model where it matters.
Control plane
Manage identities, knowledge, and workflows from one dashboard.
GPU inference
Run rendering and generation on infrastructure you own or lease.
LiveKit media transport
Real-time audio and video streamed with low latency, end to end.
Choose your path
No-code or API, same platform
Customer Studio
Build, preview, and publish an AI human through guided screens — no engineering required.
Developer Docs
Script sessions, manage knowledge, and automate workflows with REST, SDKs, and webhooks.