This post explores the continued development of Tapeify Cassettes, a project that combines AI technology with retro music products. Using Midjourney and OpenAI, I developed an interactive website with an AI ChatBot by employing my rapid prototyping of ChatGPT HTML code, and Midjourney UI design. Tapeify caters to music enthusiasts who appreciate retro technology. Our AI-powered chatbot allows users to input artist and mood preferences to generate personalised Spotify playlists. These playlists can be transformed into laser-cut wooden cassettes with integrated NFC cards. Tapping a phone against the cassette launches each song.
A Unified Web Presence
We faced an issue with multiple websites for our services: TapeifyBot, the Online store, and the new LIVECassette player each had their own URLs. Each with interdependence on one another. This led to user redirection and difficulty returning home. Load times were affected due to different hosting servers, potentially causing impatience among customers. As audience attention is extremely scarce, I decided to combine all services into a single website using WordPress under our domain tapeify.co. From audience feedback regarding the small bot window, I embedded both the bot and the LIVECassette player on the homepage. Initially, I encountered difficulties with HTTPS compatibility and embedding, but learnt that HTTP sources cannot mix with HTTPS. After reading about online privacy, I converted all the HTML sources to be secure. Now everything works smoothly, and user info is protected via HTTPS. Overall this change has enhanced Tapeify's position by capturing users' attention with a modern website which combines all services in one place.
Switching to a single website informed my next action by cleaning up our flow for our Source Code page, which endeavours to improve our project sustainability by sharing all the code and resources used, inviting new people to remix this project.
TapeifyBot Unlocked
We received lots of feedback about TapeifyBot, requesting more artists and moods. While it's a great idea, there are challenges with implementation. Ideally, users would specify their favourite artists and mood in text, but this made parameters too broad given our small artist database. Due to ChatGPT's limited knowledge after 2021, we relied on third-party integrations and manual artist updates. Fortunately, after deep research, I found PlaylistAI, which converts text inputs into Spotify playlists. It's an advantageous solution, but it takes a few minutes to generate a playlist due to its extensive 80 million+ song library. Informed by attention economy, to manage user expectations, I added a step to capture their email and send the playlist both in chat and via email. Guided by online privacy, all emails are purged after the playlist is sent. Additionally, I fully integrated OpenAI's API into chat to enable users to ask the bot questions by saying something like "I have a question", making it more capable at having natural conversations. This change broadens the bot's music options and maintains engagement whilst managing a longtail product library. Inspired by Snapchat's 'My AI', I designed a friendly persona for the bot. Complex systems thinking gain insight into systems composed of numerous interconnected components. Tapeify's web operations have interdependence on OpenAI, uChat, Google Sheets, Zapier, Zoho Mail, Wordpress and Spotify. As the semester concludes, we're focusing on project sustainability. Integrating PlaylistAI makes our project more appealing and relevant for future remixers.
Playlist generated from above conversation
Project Sustainability
As we draw closer to the end of semester, the sustainability of our project was pointing towards a potential dead end. To overcome this, I drew inspiration from remix culture and collective intelligence by developing a source code page. It is available on our website, blog, and shared through our social media channels as it's a natural environment for processes of remix and bricolage. This participatory approach allows us to provide open-source materials for others to create their own versions. Understanding copyright and content control, I have carefully indicated which of Tapeify's code and assets are copyright-free and what require attribution. This ensures compliance with copyright law and clarifies the extent of resource utilisation. Embracing collective intelligence, we aim to share the inner workings of our project, hoping others can benefit and learn from our work.
Key Takeaways
Overall, Tapeify has prompted new ways to create music playlists. This project has greatly enhanced my development and experience with virtual and physical emergent media forms. It's my first opportunity to creatively prototype with OpenAI's API, combining a scary, messy backend into a user friendly, customer-focused chatbot. This project has defined numerous new use-cases for me with A.I technology. I've learned that AI tools are only as good as their user's control, and they sometimes require boundaries due to their vast possibilities. ChatGPT is a time-saving tool rather than a replacement for tasks. I've also realised that humans are currently safe, as AI is good at lots of jobs, but not truly perfect at anything yet. Above all, this project has refreshed my perspective on group work. Disconnecting from traditional stereotypes. Mixing liquid workers with those on-site. It has promoted flexibility, exploration of non-bureaucratic workflows, and allowed everyone to contribute based on their strengths, creating a less intimidating and more enjoyable learning environment.
