← Portfolio

Turbonice · ELVTR · 2026

Turbonice

A design system that turns ELVTR’s business context into finished materials.

An optical design instrument receives a holographic context stream and produces colourful design materials
Drag to view
Role
Product design
Development & testing
Scope
Brand system
Generation workflows
Impact
61% of routine
design work eliminated
Reach
Three markets
One scalable identity

Turn recurring design workinto a repeatable process.

Turbonice is the design production system I built at ELVTR, an online education company. As ELVTR expanded across three markets, every course needed brochures, emails and visual materials adapted for each locale. Rebuilding them by hand could not keep pace.

I connected design templates and tools to an AI agent layer that gathers course and business context, prepares the content and runs production workflows. Early versions imported instructor photos from Figma; we later moved that step to an internal photo service. Depending on the workflow, finished materials are delivered to services such as HubSpot and GitHub.

BUSINESS CONTEXTTEMPLATE SYSTEMOUTPUTDELIVERY
Google Drive
Asana
Internal ERP
Slack
Context agent
CourseMarketInstructorLaunch
PDF
Google DriveFiles
HTML
HubSpotDraft
PNG
MP4
Google DriveFiles
Google DriveFiles
Design system
AI models & skills

Google Drive → Context agent → Brochure → PDF → Google Drive

Business context, design tools and delivery belong to one workflow. Figma was part of the early setup; the wider system delivers materials to Google Drive, HubSpot and GitHub, depending on the task.

Programmatic
design agent.

Brand primitives guide how assets are made: their colour, form and visual language. Describe an icon, generate it and add it to a shared library that other templates can use.

01 / Icon libraryCreate an icon and save it to the shared library. Interactive demos use sample data and simulated generation.

Design once.
Adapt every course.

Start with a template, then work directly with its pages, layers and brand palettes. The course brochure keeps its content bindings and layout rules as the design changes.

02 / Templates & design editorRobert Rabin · Game Design with Unreal Engine 5. Demo layout and sample schedule.

Follow the work,
step by step.

The workflow gathers course context, checks copy, resolves photography and icons, then populates the template. The canvas follows the active node so each stage stays visible as the run progresses.

03 / Generation workflowEarlier workflow: instructor photos imported from Figma, before our internal photo service.

Keep production moving.

Requests from different templates enter one queue. Each run moves from waiting to execution and completion, giving the team one place to see what is happening.

04 / Production queueNew requests enter the queue, start running and finish.

See what every AI call costs.

The AI cost ledger records the provider, model, usage and cost of each call. Text and image generation appear in one list, with totals that update as work completes.

05 / AI Cost LedgerNew calls appear in the ledger and update the usage totals.

The run ends.
The files are ready.

Each workflow turns course context into files the team can use: course documents, classroom images, lesson intros and supporting graphics.

The final steps render the material and deliver the output. Here, a selection of real template previews shows what comes out of Turbonice.

61% less routine
design work.

Across three markets, Turbonice eliminated 61% of routine design work by moving recurring material production into templates and workflows. The brand guidelines became working software, with enough structure to scale and enough variation to serve different courses.

Designers can focus on new ideas and exceptions while the system handles repeatable production.

More selected work