Products

Four products.
One operating system.

Each company runs independently, with its own founding team, roadmap and P&L. They share what's worth sharing — infra, brand gravity, hiring network — and nothing else.

01 / LIVE sha.chat 02 / BETA ZVM 03 / IN DEV algobeat.ai 04 / EARLY alienway.fun
sha.chat
01 / sha.chat · messaging

Secure communications,
built to connect.

Private messaging and direct secure calling with Push ID, nearby discovery, and a calm interface built for fast, secure communication.

Visit sha.chat White paper
Push IDDirect secure identity
CallsPhone or Push ID mode
NearbyFast local discovery
sha.chat secure calling screen showing Push ID calling, direct secure calling status, and a start secure call action
ID

Push ID calling

Start direct secure calls using a Push ID instead of relying only on a phone number.

Multiple contact paths

Reach people by handle, email, UID or push://, with a clean flow for fast connection.

Nearby and quick actions

Paste an identity, discover nearby peers, and start a secure call without extra friction.

02 / ZVM · infrastructure

The Zero-trust
Virtual Machine.

Safely run untrusted code — model outputs, third-party plugins, customer uploads — inside provable sandboxes with policy-first egress. Built in Rust, with post-quantum primitives (ML-DSA, ML-KEM) at the trust boundary.

Join beta Read the whitepaper
WASMRuntime core
RustImplementation
ML-DSA · ML-KEMPost-quantum
zvm · policy.yaml ● Sandbox · ISOLATED
# untrusted model output
runtime: wasm
egress:
  - api.stripe.com:443
  - deny *
memory: 128MB
timeout: 2s
attest: true
# ↪ proof published to ledger

Isolated execution

Every call runs in a fresh micro-VM. Zero shared state, zero blast radius.

Policy-first egress

Network access is denied by default. Allow-list in YAML, enforced in kernel.

Attestation

Cryptographic proof of what ran, on what code, against what policy. Auditor-ready.

03 / algobeat.ai · voice & sound

Voice and sound,
engineered.

AI voice and sound tooling: retrieval-based voice conversion, speech generation, and prosody control — with consent and provenance built in from the start.

Join waitlist How it works
RVC + TTSCore stack
Consent-firstVoice policy
In developmentStatus
VOICE · TAKE 014 RVC 48 kHz · prosody lock

Voice conversion

Retrieval-based voice conversion that preserves timing, emotion, and natural prosody.

Speech generation

Expressive text-to-speech with fine-grained prosody control for narration and dialogue.

Consent & provenance

Voices are onboarded with consent, and generated audio carries provenance signals.

04 / alienway.fun · film

Unleashing fun
in film.

An early-stage entertainment venture exploring playful, film-centered experiences — where deeztik's AI voice, video, and sound tooling meets storytelling.

Visit alienway.fun
FilmFocus
EarlyStage
2025Announced

Want a custom solution?

We're actively scouting real work-life pain points. If you need custom agents, new financial primitives, or secure compute — talk to us.

Reach us