"కొత్త నిర్మాత" — "New Creator" in Telugu
An AI-powered multi-platform content metadata generator that empowers emerging content creators.
🔗 Live Demo · 📄 Documentation · 🚀 Deploy Your Own
Emerging content creators struggle to produce professional, SEO-optimized metadata across multiple social platforms. Crafting the right titles, descriptions, tags, and hashtags for each platform is time-consuming and requires expertise in platform-specific algorithms.
KothaNirmata AI solves this by transforming a single content idea into platform-optimized metadata — instantly.
| Platform | What It Generates |
|---|---|
 |
5 Title options, Video description with timestamps, 30 SEO tags, 10 Hashtags, Thumbnail text suggestions |
 |
5 Caption options, 30 Hashtags (broad/medium/niche), 5 Reel hook lines, Bio link CTA, Posting time |
 |
5 Caption options, 10 Hashtags, 5 Hook lines, Sound/music suggestions, Video length recommendation |
 |
3 Post titles, Professional body copy, 5 Hashtags, Engagement CTA |
 |
5 Tweet options, 5 Thread starters, 10 Hashtags, Posting time suggestion |
┌──────────────────────────────────┐
│ Google Cloud Run │
│ (Serverless, Scale-to-Zero) │
│ │
User ──── HTTP ────▶ │ ┌──────────────────────────┐ │
│ │ 🤖 root_agent (Greeter) │ │
│ │ Collects content idea │ │
│ │ + save_video_idea tool │ │
│ └────────────┬─────────────┘ │
│ │ │
│ ▼ │
│ ┌──────────────────────────┐ │
│ │ ⚡ SequentialAgent │ │
│ │ (metadata_workflow) │ │
│ │ │ │
│ │ Step 1: seo_researcher │ │
│ │ ┌────────────────────┐ │ │
│ │ │ Gemini 2.5 Flash │ │ │
│ │ │ via Vertex AI │ │ │
│ │ └────────┬───────────┘ │ │
│ │ │ │ │
│ │ Step 2: response_formatter│ │
│ │ ┌────────────────────┐ │ │
│ │ │ Clean, copy-paste │ │ │
│ │ │ formatted output │ │ │
│ │ └────────────────────┘ │ │
│ └──────────────────────────┘ │
└──────────────────────────────────┘
│
▼
JSON Response
(Platform-specific metadata)
 Gemini 2.5 Flash LLM Engine |
 Vertex AI AI Platform |
Cloud Run Serverless Hosting |
 Python 3.12 Runtime |
Google ADK Agent Framework |
kotha_nirmata_agent/
├── 🤖 agent.py # Multi-agent pipeline (Greeter → Researcher → Formatter)
├── 📦 __init__.py # Package initializer for ADK discovery
├── 📋 requirements.txt # Python dependencies
├── 🔒 .env.example # Environment variables template
├── 🚫 .gitignore # Git ignore rules
└── 📖 README.md # This file
- Google Cloud Project with billing enabled
gcloudCLI authenticated
# 1. Clone the repo
git clone https://github.com/YOUR_USERNAME/kotha-nirmata-ai.git
cd kotha-nirmata-ai
# 2. Set your project
gcloud config set project YOUR_PROJECT_ID
# 3. Enable APIs
gcloud services enable run.googleapis.com artifactregistry.googleapis.com \
cloudbuild.googleapis.com aiplatform.googleapis.com compute.googleapis.com
# 4. Create .env from template
cp .env.example .env
# Edit .env with your project ID
# 5. Create service account
gcloud iam service-accounts create kotha-nirmata-sa \
--display-name="KothaNirmata AI Service Account"
gcloud projects add-iam-policy-binding YOUR_PROJECT_ID \
--member="serviceAccount:kotha-nirmata-sa@YOUR_PROJECT_ID.iam.gserviceaccount.com" \
--role="roles/aiplatform.user"
# 6. Deploy
uvx --from google-adk==1.14.0 adk deploy cloud_run \
--project=YOUR_PROJECT_ID --region=us-central1 \
--service_name=kotha-nirmata-ai --with_ui .- Open the Live Demo
- Select kotha_nirmata_agent from the dropdown
- Type your content idea, for example:
"I want to create content about AI tools for all platforms""Make a cooking video about South Indian breakfast recipes for YouTube and Instagram""Tech review of the latest iPhone for TikTok"
# Create a session
curl -X POST "${SERVICE_URL}/apps/kotha_nirmata/users/test_user/sessions" \
-H "Content-Type: application/json" \
-d '{"state": {}}'
# Run the agent
curl -X POST "${SERVICE_URL}/run" \
-H "Content-Type: application/json" \
-d '{
"app_name": "kotha_nirmata",
"user_id": "test_user",
"session_id": "SESSION_ID",
"new_message": {
"role": "user",
"parts": [{"text": "Generate metadata for a Python tutorial for beginners on YouTube"}]
}
}'Click to view agent in action
The agent greets the user, saves the video idea using the save_video_idea tool, and transfers to the metadata workflow.
Generates optimized metadata for YouTube, Instagram, TikTok, LinkedIn, and X/Twitter — all from a single prompt.
The State panel shows VIDEO_TOPIC, TARGET_AUDIENCE, PLATFORMS, and CONTENT_STYLE being tracked across the agent pipeline.
| Resource | Cost |
|---|---|
| Cloud Run | $0 when idle (scale-to-zero) |
| Gemini 2.5 Flash | ~$0.0001 per request |
| Artifact Registry | < $0.10/month |
| Total for testing | < $1 |
gcloud run services delete kotha-nirmata-ai --region=us-central1 --quiet
gcloud iam service-accounts delete kotha-nirmata-sa@YOUR_PROJECT_ID.iam.gserviceaccount.com --quiet
gcloud artifacts repositories delete cloud-run-source-deploy --location=us-central1 --quietHotragn Pettugani
Built with ❤️ using Google ADK, Gemini, and Cloud Run
