What I Built
MarketMitra-AI is a social media post generator for small businesses—such as local shops, food businesses, and independent service providers—that need promotional content but may not have time or budget for a copywriter.
Enter a business and product description, price, target audience, preferred platform, language, tone, and optional call to action. The app generates a caption, short description, hashtags, a call to action, and a WhatsApp message. It supports Instagram, Facebook, and WhatsApp, with English, Hindi, and Hinglish output.
The project is designed to run locally. Instead of sending a business’s information to a hosted AI API, the Flask app sends the prompt to an Ollama model running on the user’s computer. There’s also a sample-output demo mode for trying the interface without Ollama.
It’s a prototype: it doesn’t currently process payments, and I haven’t verified external customers or revenue.
Demo
There isn’t a public deployment or video demo available yet. The app currently runs locally:
Install the project dependencies and Ollama.
Pull the model: ollama pull llama3.2:3b Start the app with python app.py.
Open http://127.0.0.1:5000. The demo mode works without calling the AI model. Before publishing, add a public demo link or a short screen-recording URL here if you have one.
Code
Repository: https://github.com/anubhav9236/MarketMitra-AI.git The current README still contains a your-username placeholder, and I couldn’t confirm a Git remote from the project folder. Don’t leave a placeholder or link readers to a nonexistent repo.
How I Built It
The frontend uses HTML, CSS, and JavaScript. A Python Flask backend serves the interface, validates requests, calls Ollama, and returns generated content. The model is Llama 3.2 3B via Ollama, running locally; the app requests structured JSON so it can display the generated fields separately.
The output includes:
A platform-oriented caption
A short description
Hashtags
A call to action
A WhatsApp promotional message
I tested the live Flask app, Ollama health check, demo generation, download endpoint, and real model generation on my local Windows setup. One real generation completed in about 44 seconds; that’s a single local test, not a general performance benchmark.
Why Does Open Innovation Matter?
Open innovation made it possible to build and test an AI-powered tool without depending on a paid hosted inference API. Running an openly available model locally also lets users keep their business details on their own machine during normal use, while making the app’s model choice and implementation easier to inspect and adapt.
The project is not a claim that every component or model has an OSI-approved open-source license. Users should check the relevant model and dependency licenses for their intended use.
Prize Categories
I haven’t integrated a partner product or verified a qualifying sponsor track, so I’m not claiming a prize category. Remove this section unless you confirm a track that fits and meet its requirements.