We Used AI to Build Book and Music Recommendations Tools

Designing and Building Customer Experiences with LLMs

‍

It has never been easier to design, code, and deploy digital experiences using AI-enabled tools. ChatGPT, DALL-E, and Gemini. To prove the point, Brighter Side leveraged these tools to concept, design, and build two digital experiences that leverage LLMs to provide recommendations for books and music. The primary takeaway: Prompt design is essential to getting results.

What is prompt design? LLMs are only as effective as the questions they are asked. Thoughtful, detailed, and narrow prompts are the key to getting valuable results. This means that commanding language is essential. At the moment, it is clear that prompt engineering, or prompt design, currently relies upon the ability to use very precise language to deliver useful and usable results. Could it be the momentary rise of the English major? And we all thought the classics were dead.

For BotBookPix.com, our hypothesis was simple: How might we leverage LLMs to make thoughtful book recommendations? What we found was equally fascinating and discouraging. On the good side, a very well engineered prompt delivered relevant results that users have found valuable. On the down side, the code is middling, at best. As a former web developer, I could recognize the intent of the Ai-generated code, but it required some human context to deploy.

We experimented extensively with prompts to deliver valuable results. We queried ChatGPT to explore the appropriate content and design for a website that delivered book recommendations based on a user's favorite book. We also used ChatGPT to discover appropriate color palettes, typography and content to support the user experience. But the key was designing a prompt that delivers effective results based on the user's inputs to the two questions that drive the results: Your favorite book and who wrote it.

‍

‍

Aside from using LLMs to explore design, which is itself quite impactful, the most important function for BotBookPix is to deliver insightful results. To do so, we explored numerous prompts. We found that the most useful prompt asked the LLM to combine literary criticism perspectives from the New York Times, the Wall Street Journal, and Kirkus Reviews, along with ratings from GoodReads.com. This blended approach yields results that, so far, users have found valuable.

After building BotBookPix, we were curious to see if a similar approach could help deliver music recommendations. After exploring prompts and designs, we emerged with BotMusicPix.com. This tool allows users to select a favorite artist and then a genre of music. The ah-ha moment came to use when we realized we could select our favorite artist and then cross genres of music to get recommendations.

‍

Takeaways

Brighter Side is using LLMs daily to help get to great faster. That's the value that LLMs offer in 2025: They're not the most inventive, but they help get to great faster. Interested in building customer experiences with AI? So are we. Contact us to get started.

‍