This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle
What I Built
I built Social Studio, an AI-assisted social media manager for freelancers and small business owners who do not have the time or budget for a dedicated marketing team.
The idea came from my mother, who works as a freelance makeup artist. Alongside serving customers and managing appointments, she must regularly turn her portfolio photographs into social media content. Writing captions, finding hashtags, adapting posts for different platforms, and maintaining a consistent schedule takes time away from her actual work.
Social Studio brings that workflow into one simple application. A user can:
- Upload a product or portfolio image
- Generate an editable caption and relevant hashtags
- Adapt content for different social platforms
- Preview a post before publishing
- Organize media in one library
- Schedule upcoming content through a calendar
- Review engagement and performance information
The platform is designed for non-technical users. Instead of presenting a complicated marketing dashboard, it guides users through a clear workflow: select an image, generate the post, review it, and schedule it.
The initial target users are freelance makeup artists, home-based beauty professionals, and other visual service businesses. These users could pay for Social Studio because it saves time every week while helping them maintain a more professional and consistent online presence.
Demo
How I Built It
I designed Social Studio as a full-stack web application centered around an image-to-content workflow.
The user interface contains separate areas for creating posts from images, editing generated content, scheduling posts, managing uploaded media, reviewing engagement, and viewing analytics. I kept the design intentionally simple because the target audience includes business owners who may not be comfortable with complex marketing software.
For the AI workflow, Social Studio uses Gemma 3 4B, an open-weight multimodal model from Google, running locally through Ollama. Gemma analyzes uploaded portfolio images and generates editable, platform-specific captions, calls to action, and hashtags without sending customer photographs to a commercial AI API.
When a user uploads an image, the application sends it to the local multimodal model with structured instructions containing:
- The business type
- The intended social platform
- The preferred writing tone
- The desired call to action
- Relevant caption and hashtag requirements
The model returns structured content that the application converts into an editable social media post. The user remains in control and can modify everything before scheduling it.
Why Does Open Innovation Matter?
Open innovation makes Social Studio practical for the freelancers it is intended to support.
A makeup artist's portfolio can contain identifiable photographs of customers. By running an open-weight model locally, Social Studio can analyze those images without sending them to a third-party AI service. This gives the business owner greater control over client data and reduces the number of external systems that handle sensitive photographs.
Local inference also changes the economics of the product. Freelancers may generate many drafts before selecting what to publish. With a commercial API, every attempt introduces another operating cost. A locally hosted open-weight model removes the per-generation charge and makes predictable, affordable pricing possible for small businesses.
The open approach also prevents the application from being tied to one model provider. I can test different models, change the prompting workflow, and specialize the application for industries such as makeup, hairstyling, photography, food, or handmade products.
Most importantly, the open model is not an optional feature added for the challenge. Image understanding and content generation are the central operations that make Social Studio useful.
Prize Categories
- Best Use of Gemma — Social Studio uses Gemma 3 4B as the open-weight model at the center of its image-to-content workflow.
Top comments (0)