I embarked on a weekend sprint to build a memory-card app that mimics Anki’s core ideas but adds a few personal twists
. The twist: I did it without writing a single line of code for B4X. Instead, I leaned on AI-powered tools—Cursor to plan and automate the process, Claude Code to generate the practical code scaffolding and UI logic, and a handful of other AI-assisted assets for audio, visuals, and sharing. It sounds wild, but in 72 hours the project came together, ran smoothly, and now ships with all the features I care about: spaced repetition based on the Ebbinghaus forgetting curve, randomized card presentation, audio cues, social sharing, and even a playful poster and micro-interactions. I spent about $140 in total on tools and assets, which felt like a acceptable price for a fully functional prototype.
Why I tackled this project
You can ask in the comments about what you want to know!
. The twist: I did it without writing a single line of code for B4X. Instead, I leaned on AI-powered tools—Cursor to plan and automate the process, Claude Code to generate the practical code scaffolding and UI logic, and a handful of other AI-assisted assets for audio, visuals, and sharing. It sounds wild, but in 72 hours the project came together, ran smoothly, and now ships with all the features I care about: spaced repetition based on the Ebbinghaus forgetting curve, randomized card presentation, audio cues, social sharing, and even a playful poster and micro-interactions. I spent about $140 in total on tools and assets, which felt like a acceptable price for a fully functional prototype.Why I tackled this project
- Proving a concept: Could I ship a usable learning tool with AI-assisted automation in a tight window?
- Learning by doing: Understanding how AI tools can complement product design and lightweight development without traditional coding.
- Personal curiosity: A memory-aid app is a perfect playground for exploring scheduling algorithms, UI/UX polish, and the balance between automation and quality.
- A web-based memory-card app similar to Anki: create and delete cards, study sessions, and a queue that adapts to how well you recall a card.
- Spaced repetition influenced by the Ebbinghaus forgetting curve, with randomized display order to reduce pattern predictability.
- Audio: pronunciation or cue sounds for each card, plus optional text-to-speech playback during review.
- Social sharing: one-click sharing of progress or decks to select social platforms.
- Promotional poster: an AI-generated card-back drop-in poster for sharing or personal branding.
- Micro-interactions and animations: subtle transitions to make study sessions feel responsive and engaging.
- Tools used
- Cursor: for project planning, workflow automation, and assembling the “no-code/low-code” scaffolding.
- Claude Code: to generate practical code templates, UI components, and business logic prompts that guide the app’s behavior.
- AI assets: text-to-speech for audio cues, AI image generation for the poster, and lightweight animation assets.
- Plan and constraints
- Timeframe: 72 hours
- Scope: a complete, runnable prototype with essential features, not a production-grade system
- Budget: around $140, covering AI credits, asset licenses, and hosting costs
- How AI integrated into the build
- Requirement framing: I described the app’s goals, data model, and user flows in natural language and used Cursor to map them into a repeatable plan.
- Code generation and wiring: Claude Code produced UI scaffolds, data schemas, and simple scheduling logic. I refined prompts to ensure the outputs matched the desired behavior (cards, deck management, scheduling, and display randomness).
- Asset creation: Text-to-speech modules generated short audio clips for cards; posters and visuals were created with AI image tools, then refined for readability and accessibility.
- Deployment and iteration: The workflow included quick testing prompts, style adjustments, and small tweaks to the user flow, all guided by AI-generated iterations.
- Create and delete cards: Basic deck management for building a personalized knowledge base.
- Based on the Ebbinghaus forgetting curve: Scheduling that adapts based on recall performance, with intervals influenced by the curves commonly discussed in forgetting theory. Cards due sooner get prioritized, while ones recalled well are shown later, with randomized order to keep recall robust.
- Randomized display: Each study session pulls a shuffled subset of due cards to minimize predictability and maximize retention.
- Audio: Sound cues and optional text-to-speech playback to reinforce memory through auditory channels.
- Share to social media: Simplified sharing of progress or decks to social platforms to celebrate milestones or to invite others to try the deck.
- Promotional poster: An AI-generated poster image that can be used for personal branding, mentoring, or sharing the project publicly.
- Animations and micro-interactions: Gentle motion to convey state changes (e.g., when a card is marked “easy” or “again”) and to reduce cognitive fatigue during sessions.
You can ask in the comments about what you want to know!
