MarketMuse-AI tool for Brand Taste Profiling

Inspiration

Marketing today is heavily data-driven, yet many small and mid-sized teams lack access to enterprise-grade tools. We were inspired to create MarketMuse to level the playing field—empowering lean teams with AI-powered brand and taste profiling, campaign insights, and persona generation using modular, open APIs.

What it does

MarketMuse is a modular marketing intelligence tool that leverages cultural, behavioral, musical, and visual data to:

Generate detailed brand personas
Provide predictive ROI scores
Suggest tailored marketing content
Visualize search and cultural trends
Perform tone and style analysis from social media and playlists

The result: smarter, more culturally-aligned campaigns without the overhead of traditional enterprise systems.

How we built it

We built MarketMuse using an API-first, modular architecture, enabling smooth integration with best-in-class tools:

🔌 API Integrations

Together AI (Mixtral 8x7B) for tone analysis and persona generation
Qloo API for cultural and taste profiling
Spotify API for music preferences and behavior
Stability AI (Stable Diffusion XL) for generating campaign visuals
Google Trends (via PyTrends) for search volume and market interest
Twitter API for real-time sentiment and topic analysis

🧰 Tech Stack

Flask + Jinja2 for the backend and templating
SQLAlchemy for ORM and database management
Pandas, NumPy, Matplotlib, Seaborn for data processing and visualization
python-pptx for report generation

We implemented retry logic, response caching, and structured logging to ensure a responsive and robust system, even with rate-limited APIs.

Challenges we ran into

Rate limits from various APIs required efficient caching and retry logic.
Balancing speed with modularity while integrating multiple asynchronous services.
Normalizing data from diverse sources (text, images, music, trends).
Designing an interface that communicates complex insights clearly and accessibly.

Accomplishments that we're proud of

Built a working prototype in a short hackathon window.
Seamlessly integrated six external APIs into one cohesive workflow.
Successfully generated dynamic marketing personas and visuals from live data.
Created a user-friendly, insight-rich interface with predictive ROI scoring.

What we learned

The power of modular, API-driven design for rapid prototyping.
The importance of cultural context in marketing data.
How to align LLMs (like Mixtral) with structured business goals (personas, tone, ROI).
How small data signals (tweets, playlists) can guide large branding decisions.