On October 8, Likeness House was just a document. By October 9, it was a live, bilingual product at likenesshouse.com. Founder ilinmaks shipped a digital twin platform featuring a custom WebGL portrait engine, client accounts, and an admin panel in under 24 hours. The launch demonstrates how AI coding assistants and a minimal infrastructure stack can compress development timelines from weeks to hours.

The Zero-Dependency Architecture

The site runs on a small Node.js server with no frameworks and no external dependencies. This choice was deliberate, driven by past lessons learned on a robot catalog where Google struggled to index multiple languages. Server-side rendering ensures both English and Russian pages are fully readable by search engines. Security is handled with scrypt password hashing, HttpOnly cookies for sessions, and origin checks on forms. The entire application runs on Railway with a persistent volume, keeping the operational overhead low.

AI-Built WebGL Portrait Engine

The homepage features a 2.5D relief of the founder’s face, rendered as thousands of points in WebGL. This isn't a video; it’s a live, interactive engine that gathers, blinks, and follows the cursor. The code was split between two AI assistants: Codex built the portrait engine, while Claude built the site structure. The engine originated in Pulsara, a voice assistant project, and was tuned for this launch by adjusting point density and colors to ensure it runs smoothly on mobile devices without draining battery life.

Public Pricing and Minimalist UX

Likeness House offers three tiers with transparent pricing: Portrait (€15,000 + €290/mo), Legacy (€30,000 + €490/mo), and Private (from €60,000 + €990/mo). The site avoids typical startup fluffβ€”no testimonials, no 'coming soon' blocks, and no placeholder text. The admin panel, written in Russian, allows the founder to manage requests, change statuses, and reply to clients directly. Telegram notifications handle new requests, while email delivery via Resend is queued for future integration, currently relying on persistent storage for outgoing messages.

Key Takeaways

  • Speed via AI: Using Codex for graphics and Claude for logic allowed a full-stack launch in one day.
  • No Frameworks: A vanilla Node.js server reduced bloat and solved SEO issues for multi-language content.
  • Transparency: Public pricing and a lack of marketing fluff signal a focus on early adopters over mass-market hype.
  • Infrastructure Simplicity: Railway and a persistent volume provided enough backend capability for MVP without complex orchestration.

The Bottom Line

This launch proves that for niche, high-value services, you don't need a massive engineering team or a complex microservices architecture. A single developer with clear direction, AI assistants, and a zero-dependency mindset can ship a polished, functional product in a day. The real innovation here isn't the digital twin concept, but the ruthless efficiency of the build process.