AI Manual Assistant

Record once. ZASSHA builds the manual for you.

Upload a screen recording, click Analyze, and ZASSHA extracts clean, reproducible steps with thumbnails, rich context, and exports for Word, PowerPoint, or Excel in minutes.

Download on GitHub

Automatic step extraction

ZASSHA analyzes your workflow video, detects each operation, and turns it into a timeline of steps complete with timestamps, tools, and AI-generated explanations.

Share-ready exports

Ship polished documentation instantly with Word, PowerPoint, or Excel exports that include screenshots, captions, and speaker notes.

Private by default

Analysis runs locally on your machine—videos aren’t uploaded or stored in any external database. Only requests you initiate to the Gemini API leave your device, and you control exactly what’s sent.

Five-minute publishing workflow

Capture

Record at 1080p+ using macOS or Windows screen capture tools for best accuracy.

Upload

Drop the file into ZASSHA and pick Summary or Detail mode depending on the guidance you need.

Hint

Add a short goal statement so AI can focus on the outcome you expect.

Review

Inspect the overview, business inference, and operation table with hoverable thumbnails.

Export

Send the manual to DOCX, PPTX, or XLSX with consistent naming that mirrors your source video.

For your first try, we recommend using a video under 5 minutes and 100 MB.

Built for ops, enablement, and QA teams

Anyone tasked with documenting processes can move from raw capture to polished manual without rewriting steps by hand.

Operations

Standardize cross-team procedures and keep audit logs fresh with every process change.

Customer success

Produce up-to-date walkthroughs for onboarding, support runbooks, and help center refreshes.

Quality assurance

Record a test session and ship reproducible bug reports complete with evidence.

Learning & development

Convert training recordings into step-by-step playbooks for new hires and reskilling projects.

Product management

Document feature walk-throughs so engineers, designers, and stakeholders align on expected behavior.

IT support

Capture troubleshooting sessions and publish ready-to-follow guides for service desks.

New-hire onboarding

Welcome teammates with walkthroughs that mirror real workflows and keep institutional knowledge fresh.

Technical writers

Draft knowledge base articles faster by reusing AI-generated copy and screenshots.

BPO teams

Keep offshore partners aligned with continuously updated, visual instructions.

Frequently asked questions

Clone the repo and enter the folder:

git clone https://github.com/co-r-e/Zassha
cd Zassha

Use Node 18.18+ and install dependencies:
```
npm install
```

Create your env file and add your Gemini key:

cp .env.example .env.local
# then edit .env.local and set
GEMINI_API_KEY=YOUR_KEY_HERE

Start the app and open it:
```
npm start
```
Visit http://localhost:3000.

Optional — enable server-side video chunking (ffmpeg required):

# macOS (Homebrew)
brew install ffmpeg

# in .env.local
ZASSHA_SEGMENT_LEN=45

ZASSHA analyzes videos locally on your machine. Videos are not uploaded to or stored in any external database. Your Gemini API key lives in .env.local and never gets committed to git. Only requests you initiate to the Gemini API leave your machine, and you control what is sent. If you enable optional ffmpeg chunking, processing still happens locally.

Yes. It is open source. You only pay Google Gemini API usage when you analyze videos.

English and Japanese manuals are supported out of the box, and you can swap between them anytime from the header.

Nope. ZASSHA generates thumbnails alongside each detected operation so you can ship documentation with minimal edits.

Ready to turn recordings into living manuals?

Clone the repository, connect your Gemini API key, and publish your next standard operating procedure in minutes.

Download on GitHub