ZJT
Open-source, AI-agent-driven platform that turns a marketing brief into finished video — script, storyboard, and synthesis, with video cloning, brand-IP consist
What is ZJT?
Open-source, AI-agent-driven platform that turns a marketing brief into finished video — script, storyboard, and synthesis, with video cloning, brand-IP consistency, and batch A/B creative
Overview of ZJT
What it is
ZhiJuTong (ZJT) is an AI-powered, open-source platform specifically designed for creating professional short dramas. It automates the entire production pipeline, from script and storyboard creation to final video synthesis.
Key features
- 8 expert AI agents collaborate (Story Writer, Character Creator, Location Creator, Plot Analyzer, Content Compliance, Novel Splitter, Character Designer, Location Designer) with a PM Agent.
- Real-time team collaboration in the browser (LAN and public network), with SSE push and two-level permission management.
- User-level independent compute accounts: each user has separate quotas, can choose LLM and image/video providers, and administrators can hot-update configurations.
- Supports 10+ LLMs, multiple image/video generation APIs (Nano, Banana, Duomi, ComfyUI, etc.), and automatic failover.
- Infinite canvas editor with multi-grid storyboard layout, consistent character design, and intelligent workflow optimization.
- Built-in resources: 500+ prompt templates, free image hosting with CDN, TTS service (10+ languages).
Use cases
- Online education: allocate compute quotas to students for short drama creation.
- Enterprise training: department-level compute management and cost control.
- Creative studios: team members manage their own budgets transparently.
- Startup incubators: provide compute support for startups.
Getting started
See the repository README for detailed instructions. Quick start options include:
- Online experience at ailive.perseids.cn (no installation required).
- Windows: download the latest executable from GitHub Releases and run it.- macOS: download and unzip the release, then run
open start.command. - Linux: clone the repo, configure database, install dependencies with
uv sync, and runpython3 scripts/running/run_prod.py. - Docker:
cd docker && docker-compose up -d.
License and pricing
The license is not stated in the repository metadata (NOASSERTION). The platform is open-source (pricing_type: open_source).
Based on the project's public README and repository metadata.