AI video production infrastructure for teams

At Tuba Intelligence Limited, headquartered in Hong Kong, we build a production workspace where teams can plan scripts, construct AI workflows, manage assets, collaborate, review, and deliver video projects with control.

1

Workflow project model

4K

Seedance 2.0 output

Team

Collaboration tier

Enterprise

Task review tier

What Tuba Means

To us, Tuba represents a place where ideas come alive.

Tuba Intelligence is building a next-generation AI content creation platform. We believe AI should not replace creativity, but amplify it.

TubaGen is pronounced “TOO-buh-jen” (/ˈtuːbəˌdʒɛn/).

Our Story

TubaGen was founded to solve a practical production gap: AI video teams had to jump across disconnected chat tools, model dashboards, asset folders, review documents, and editing systems. We are building a single workspace where a project is an executable AI workflow that keeps scripts, scenes, shots, assets, tasks, generated candidates, and review records together.

What We Ship Today

These platform modules are already represented in the product and workflow editor.

Workflow Project Console

Manage workflow projects, workflow templates, tags, trash, workspace switching, billing, members, roles, SSO settings, and quota context from the console.

AI Planning Agent

Use the planning agent to understand briefs, parse scripts, split episodes, extract assets, generate storyboard structure, and prepare workflow-ready material.

Structured Asset Library

Create and manage characters, scenes, props, products, sounds, effects, styles, references, and real-person assets with tags and media.

Workflow Canvas

Build image, video, audio, text, edit, frame extraction, compose, and audio overlay workflows with node-level execution and result history.

Model Nodes

Run connected model nodes such as Seedance 2.0 with 4K output, Veo 3.1, Kling 2.6, Wan 2.7, HappyHorse, Nano Banana, GPT Image 2, Seedream, Flux, Recraft, Topaz, and audio generation models.

Enterprise Task Flow

Support producer and reviewer roles, task assignment, submissions, review comments, storyboard approval, and locked handoff into video generation.

Built for Production Control

The product is organized around operational clarity rather than one-off prompt generation.

Workspace Boundary

Workspace is the tenant, project, asset, template, tag, task, member, billing, and permission boundary.

Manual Asset Curation

Generated outputs stay as node results until a user intentionally saves selected results into the workspace asset library.

Reviewable AI Changes

The planning agent should propose changes that users can inspect and apply, keeping critical scripts and production structure under human control.

One Workflow, End to End

Teams can go from idea to reviewed video through a continuous production path.

Step 1

Plan

Create or import scripts, parse scenes and shots, identify asset candidates, and confirm the structure.

Step 2

Generate

Build a workflow canvas, run model nodes, compare candidates, and save selected outputs.

Step 3

Review

Assign tasks, collect comments, lock approved storyboard frames, generate video, and deliver final outputs.

Why Teams Choose TUBAGEN

One workflow project keeps scripts, scenes, shots, assets, generated candidates, tasks, review records, and outputs together.

The canvas supports image, video, audio, text, edit, compose, and AI assistant nodes instead of isolated prompt runs.

Workspace assets and templates make repeatable production possible.

Enterprise roles and task review match real producer/reviewer handoff.

In Development

Deeper Script Module

Script objects will continue to evolve as independent, versioned production objects that can generate and update workflow projects.

Expanded Asset Review

Asset suitability checks will help teams identify resolution, format, real-person, model compatibility, and risk issues earlier.

Our Vision

From Hong Kong to global markets, Tuba Intelligence is building production infrastructure for AI-native video teams.

Our mission is to turn AI generation from a collection of isolated model calls into a managed, reviewable, collaborative production process.