Sento AI is an intelligent social media automation platform that empowers creators and businesses to manage their online presence effortlessly. Using AI agents, Sento generates engaging content, creates reels, and posts directly to Instagram and Facebook — all from a simple text prompt.
Small and medium businesses (MSMEs) often lack access to expensive marketing teams and professional social media expertise. Creating consistent, high-quality content across multiple platforms is time-consuming and requires skills that many entrepreneurs simply don't have.
Sento AI bridges this gap by providing an AI-powered solution that automates content creation and publishing, allowing business owners to focus on what they do best — running their business.
| Technology | Purpose |
|---|---|
| Next.js | Frontend framework for the web application |
| Kestra | Workflow orchestration engine for AI pipelines |
| Google Gemini API | AI-powered content generation and prompt enhancement |
| ImageKit | Media storage and content delivery (containerization) |
| Vercel | Hosting the frontend |
| Meta Graph API | API to interact with Meta App Suite |
Sento AI uses Kestra to orchestrate complex AI workflows. Here's how each flow works:
Flow: agentic_video_director.yml
This flow transforms a simple text prompt into a fully produced Instagram Reel:
- Receives the user's prompt via webhook
- Enhances the prompt using Gemini AI for better video generation
- Generates a video using Google Veo 3.1
- Adds background audio to make the reel engaging
- Creates a catchy caption with relevant hashtags
- Uploads the video to ImageKit for hosting
- Publishes directly to Instagram as a Reel
Flow: agentic_image_poster.yml
This flow creates stunning image posts from text descriptions:
- Receives the user's creative prompt
- Enhances the prompt with Gemini for photorealistic results
- Generates a high-quality image using Google Imagen
- Creates an engaging caption with trending hashtags
- Uploads the image to ImageKit
- Posts to Instagram feed automatically
Flow: instagram_auto_reply.yaml
This flow handles audience engagement automatically:
- Listens for new comments via Meta webhook
- Verifies the webhook signature for security
- Analyzes the comment sentiment using AI
- Generates a contextual, on-brand reply
- Posts the reply to the comment thread
- Logs the interaction for analytics
The platform follows a modern microservices architecture:
- Frontend: Next.js application handling user interactions
- Orchestration: Kestra manages all AI workflows and API integrations
- AI Layer: Google Gemini for text, Imagen for images, Veo for videos
- Storage: ImageKit for media hosting and CDN delivery
- Publishing: Direct integration with Meta Graph API for Instagram/Facebook
- **To run the project, you must first setup Kestra using Docker Compose and the Next app locally. Deployed version is just for frontend.
- Clone the repository
- Copy
docker-compose.ymlfrom this repo - Add your API keys (Base64 encoded) to the environment
- Run
docker compose up -d - Access Kestra at
http://localhost:8080 - Run the Next.js app with
pnpm dev
For detailed setup instructions, visit the Documentation.
Built with ❤️ by Khushal Agarwal