News & Updates

What's Happening

Latest developments from the Creatora team.

July 10, 2026 Milestone

Three Products on a Single AI Core

We validated our founding thesis — one AI core, many products. From a single shared engine we now run three working products: a consumer assistant service, a white-label robot delivered to partners, and a B2B automation platform. The same reasoning, dialogue and safety layer powers all three.

  • 🧩 Shared AI core — reasoning, dialogue, and safety reused across products
  • 🛍️ Consumer service — an AI assistant users talk to directly
  • 🤝 White-label robot — delivered to partners under their own brand
  • 🏢 B2B platform — automation for professional teams

See the full story → products.creatora.app

One core, three products in production — the flywheel that funds what comes next.
June 30, 2026 Milestone

Commercial AI Robot-Lawyer — Delivered

Creatora built and shipped a commercial AI robot-lawyer: it consults clients around the clock, redacts personal data, works in multiple languages, generates documents and issues invoices — running live in production for a real legal practice.

  • ⚖️ 24/7 client consultations — natural dialogue in several languages
  • 🛡️ Personal-data redaction — privacy-compliant by design
  • 📄 Documents & invoices — generated automatically
  • 🚀 Live in production — serving a real legal practice
Proof that our AI core handles a regulated, high-stakes domain end to end.
June 3, 2026 Research

Creatora Competes in Kaggle NeuroGolf 2026

Our team entered the NeuroGolf 2026 challenge on Kaggle (linked to IJCAI–ECAI 2026) — an ARC-style reasoning competition on building compact, efficient neural programs under tight compute and memory budgets. The work sharpened the model-compression and reasoning techniques that feed back into our on-device AI.

Research keeps our engineering edge sharp — smaller, faster models on real devices.
April 10, 2026 Announcement

LiveFace Research Paper Published

Our research paper on LiveFace technology has been published on multiple scientific platforms. The paper describes the architecture, training methodology, and performance benchmarks of our patent-pending system for photorealistic facial animation on mobile devices.

The technology behind your future AI avatar is now open for the scientific community to review and cite. Patent Pending.
April 8, 2026 Announcement

LiveFace Technology — Patent Pending

We filed a provisional patent application with the USPTO for LiveFace — our photorealistic avatar technology that turns a single photo into a talking, blinking, emoting avatar running entirely on your phone.

  • 📸 One photo — just upload a selfie, AI creates your avatar
  • Photorealistic quality — not cartoon, looks like a real person
  • 📱 Runs on your phone — no cloud, no internet, no subscription
  • Real-time — 30+ fps, instant lip sync with voice
  • 😊 Emotions — happy, sad, surprised, calm, thinking
Your personal AI assistant will have YOUR face — speaking, blinking, and reacting naturally. Coming soon.
April 6, 2026 Milestone

LiveFace V2 — All Neural Models Trained

All five LiveFace neural models completed training: mouth (lip sync from speech), eyes (blinking and gaze), hair (natural movement), body (breathing and shoulders), and compositor (final frame assembly).

Next: quality optimization, portrait format (9:16), and Android integration.
April 3, 2026 Feature

On-Device AI (Local LLM)

A development milestone for our flagship assistant: we got a language model running directly on the device — no internet required for basic conversations.

  • 🧠 Local AI model runs directly on the device
  • 📱 Runs on devices with 4GB+ RAM
  • ⚡ ~5–15 tokens/second on mid-range phones
  • 🔀 Smart routing: simple questions → local AI, complex → cloud
  • 🌍 Handles 29 languages out of the box
A step toward an assistant that's fast, works offline, and keeps your data on your device.
April 2, 2026 Feature

Natural Voice (Neural TTS)

We gave the flagship assistant a natural human voice, powered by neural text-to-speech.

  • 🗣️ Near-instant speech generation (~200ms latency)
  • 🔊 Smooth sentence batching — no pauses between sentences
  • 🔄 Automatic fallback to offline TTS if no internet
Toward an assistant that sounds like a real person, not a robot.
April 2, 2026 Feature

Longer, More Detailed Responses

We taught the flagship assistant to give detailed answers on request — stories, explanations, and walkthroughs up to 30 sentences — while keeping short answers for quick questions.

Depth when you want it, brevity when you don't — shaping how the flagship holds a conversation.
April 1, 2026 Fix

Engine Stability — Build Hardening

Hardening the development build: fixed a critical startup crash and improved voice-engine reliability, with better error handling for speech synthesis.

A more stable foundation for the flagship as it takes shape.
March 30, 2026 Feature

Phone Operator — Device Control by Voice

Prototyped device control for the flagship: querying phone status by voice — battery level, device info, network status, Wi-Fi and Bluetooth, location.

"How much battery do I have?" — answered by voice, without touching the screen.
March 30, 2026 Feature

Smart Context — 40–60% Less Cloud Usage

An efficiency milestone: intelligent context compilation sends only relevant data to the model, and a memory-health system auto-cleans stale data.

Faster responses and lower data usage — engineering the flagship to run lean.