To understand the current museum app ecosystem, I conducted a competitive analysis of existing museum guides, art discovery platforms, audio tour experiences, and AI-based interaction patterns. I also reviewed app store feedback to identify recurring usability gaps, especially around information overload, static descriptions, navigation friction, and lack of personalization.

I organized the findings through affinity mapping to compare how current products support artwork discovery, guided learning, audio interpretation, saving, and recommendation behavior.

MUSEUM GUIDE APPS: Products that help visitors access museum information, browse collections, or follow guided experiences.

AUDIO TOUR EXPERIENCES: Tools focused on listening, narration, and structured artwork interpretation.

AI CHAT + VOICE INTERFACES: Interaction patterns that show how conversational AI, voice input, and adaptive responses could support more flexible learning.

Visitors enjoy the museum experience, but the meaning doesn’t always stay with them. After visiting museums, participants could remember enjoying the artworks, but struggled to recall what they had learned or why certain pieces stood out.

Visitors wanted a way to explore artworks instead of only reading one-way static descriptions. Participants wanted space to ask questions, clarify meaning, and explore artworks through a more conversational experience.

Visitors preferred explanations that felt personal, quick, and easier to follow than static wall text. Instead of reading long or generic descriptions, visitors wanted explanations that could match their pace, interests, and level of understanding.

User starts without account friction: When visitors open MUSE, they can sign in, create an account, or continue as a guest. Since museums are often visited in the moment, the goal was to let users enter the experience quickly without forcing account creation before they can explore.

User selects interests to personalize the experience: Before landing on the home page, visitors can select artworks and museums they are interested in. This helps MUSE understand what kind of experiences they are drawn to, while still giving users the option to skip and start exploring right away.

User scans an artwork to learn in context: From the home page, visitors can scan an artwork while standing in front of it. Once the artwork is recognized, MUSE brings them to an artwork detail page where they can choose how they want to learn about the piece.

User chooses between voice-guided audio and AI chat: Visitors can either listen to a voice-guided audio description or continue through AI chat. The audio experience is designed for moments when users want to keep looking at the artwork, while chat gives them space to ask questions, clarify meaning, or explore the artwork in a more conversational way.

User adjusts the pace of the explanation: On the artwork experience, visitors can choose how detailed they want the explanation to be. A slower pace uses simpler vocabulary and shorter descriptions, while a faster pace gives more context without over-explaining the artwork. The same explanation is also supported through a transcript, so users can listen, read, or switch between both.

User saves artworks to revisit later: Visitors can save artworks and museums they like, making it easier to return to them after the visit. These saved items also help MUSE understand the user’s interests and recommend more relevant artworks and museums over time.

Reducing friction before personalization: One of the main challenges was designing personalization without forcing users through a heavy setup. MUSE gives visitors a guest mode and a skippable interest selection flow, so they can start exploring quickly while still having the option to shape the experience.

Making personalization visible through pace: Personalization can feel vague when it only happens in the background. The pace selector gives users a clear way to control how they want the artwork explained, whether they prefer a shorter and simpler description or a more detailed one.

Using audio and chat for different museum moments: Visitors do not always want to interact with an app in the same way. Audio supports moments where they want to keep looking at the artwork, while chat supports moments where they want to ask questions or go deeper. Designing both made the experience more flexible without adding unnecessary complexity.

Personalization should not depend on perfect user input: One of the hardest parts of this project was figuring out how MUSE could feel personal for first-time users, especially if they skipped onboarding, did not save artworks, or had no previous activity. At first, the experience felt too similar to other recommendation-based products: ask users what they like, then recommend more of it. I realized that personalization had to work beyond saved items. It needed to come from smaller moments in the experience, like the user’s chosen explanation pace, the way they interact with chat, and the type of content they choose to explore.

Designing with AI requires balance between usefulness and control: AI made it possible to imagine a more adaptive museum experience, but it also raised new design questions around trust, clarity, and over-personalization. I had to think carefully about how much the app should infer about the user and how much control should stay visible. Features like guest mode, skip options, adjustable explanation pace, and saved artworks helped keep the experience flexible without making the app feel overwhelming or intrusive.