Brand Sonar

Inspiration

My inspiration came from studying the success stories of how major companies like Netflix and JCDecaux use Qloo's data to drive multi-million dollar decisions. I realized the incredible power wasn't just in the data itself, but in making it accessible. I saw a major gap in the market: brand strategists were stuck between slow, expensive traditional market research and fast but unreliable generative AI that often "hallucinates" facts. My goal was to build a bridge between these two worlds—a tool that combines the conversational ease of an LLM with the verifiable, real-world truth of Qloo's cultural intelligence.

What it does

Brand Sonar acts as an on-demand strategic counsel for brands. It's a market intelligence and market research platform that empowers a brand strategist to ask complex, high-level questions in plain English and get back comprehensive, data-driven answers in seconds. For example, a user can ask, "Identify three emerging fashion brands that resonate with fans of Taylor Swift who don't currently follow my brand," or "Find me three US cities where our direct competitors are weak, but where there is a high affinity for sustainable products and outdoor activities." The platform deconstructs the question, queries Qloo's massive repository of consumer taste data, and delivers a full strategic brief complete with narrative analysis, actionable recommendations, and interactive visualizations.

How I built it

I built Brand Sonar using a Graph-Agentic Retrieval-Augmented Generation (GA-RAG) architecture. The frontend is a professional dashboard built with a modern stack including Next.js 15, React 19, and Tailwind v4. The backend services run on the cloud.

The real core of my work was the agentic intelligence layer, which uses the OpenAI GPT API. First, I ingested Qloo's cultural entity data into a Neo4j knowledge graph. This allows the system to understand the deep, multi-hop relationships between tastes and preferences. Then, I developed an AI agent that acts as an expert "business analyst." This agent takes a user's question, breaks it down into a logical plan, and then executes that plan. It autonomously decides which tools to use, whether that's querying the Qloo knowledge graph for deep relational data or performing a corrective web search for breaking news. Finally, the agent synthesizes all the verified information into a polished, human-readable report with charts and maps created using Recharts and Google Maps.

Challenges I ran into

The biggest challenge was mapping the ambiguity of human language to the precision of API calls and graph queries. An early query like "find audiences who like cool music" was impossible for the system to act on. To solve this, I had to engineer a self-correction mechanism into the agent. Now, if a query is too vague, the agent is trained to prompt the user for clarification, asking things like, "Can you give me an example of an artist or brand you consider 'cool'?" This allows it to effectively ground the subsequent Qloo queries in concrete entities.

Accomplishments that I’m proud of

My proudest accomplishment is the seamless synergy between the GPT-generated narrative and the Qloo-powered data visualizations. The platform doesn't just show you a chart of market data; it tells you the story behind the numbers, explaining why a specific market is a prime opportunity based on a complex web of cultural affinities. I'm also incredibly proud of building a system designed for trust. Every key insight is transparently sourced to Qloo's data, creating a verifiable and auditable system that a C-suite executive can confidently rely on to make high-stakes decisions.

What I learned

I learned that for B2B AI applications, grounding isn't just about factual accuracy; it's about building institutional trust. Simply providing a correct answer isn't enough. You have to show your work. By making the data source transparent and verifiable, I transformed the tool from a "black box" into a trusted advisor. I also learned that the true power of an agentic architecture is teaching an AI how to think like a strategist—how to break down a problem, gather evidence, and build a case—rather than just answering a simple question.

What's next for Brand Sonar

The future of Brand Sonar is proactive intelligence. The next version will include an alerting system that uses persistent monitoring agents. These agents will constantly analyze real-time data streams from the Qloo API to detect significant shifts in cultural tastes, emerging trends, or competitive threats relevant to a user's brand. Instead of waiting for a user to ask a question, Brand Sonar will proactively send strategic alerts and opportunity briefs directly to them, transforming it from a powerful reactive tool into an indispensable, always-on strategic partner.

Built With

• cloud-services
• google-maps
• javascript
• mapbox
• neo4j
• openai-gpt-api
• python
• qloo-api
• react
• recharts