The Allegorical Illustrator

Shyam Kumar V N — Thu, 30 Oct 2025 16:05:32 +0000

This post is my submission for DEV Education Track: Build Apps with Google AI Studio.

What I Built

I built The Allegorical Illustrator, an application that transforms abstract philosophical concepts into stunning, allegorical art pieces. The app allows users to select a philosophical dilemma (like "The Ship of Theseus") and an artistic style (like "Japanese Woodblock Print") and uses AI to generate a unique visual representation along with a detailed explanation.

This app was created iteratively using a series of prompts. The initial prompt was simply to build the application based on a descriptive image. From there, I added features through conversational requests like:

"Implement a 'Share' button... that allows users to share the generated image and its explanation together as an image."
"Add a guardrail that first verifies if a given concept is actually representing a philosophical concept."
"Add a feature to display a history of previously generated allegories... stored client-side using localStorage."

A key feature of this app is its multi-step, AI-powered generation and validation pipeline, which uses different models for different tasks. It first validates user input, then generates a detailed image prompt and a philosophical explanation, creates the image using Imagen, and finally performs a multimodal check to ensure the image is a relevant allegory for the concept.

Demo

https://the-allegorical-illustrator-789616276303.us-west1.run.app/

My Experience

Working on this project was a fascinating exercise in AI-driven development. My key takeaway is the power of a multi-step, "chain-of-thought" approach to prompting for complex tasks. Instead of trying to do everything in one massive prompt, breaking the problem down into sequential, validated steps—input validation, content generation, image creation, and image validation—produced a far more robust and reliable application.

I learned a great deal about leveraging the specific strengths of different models. I used the powerful gemini-2.5-pro for the nuanced tasks of interpreting philosophy and crafting creative prompts, imagen-4.0-generate-001 for high-quality image generation, and the fast, multimodal gemini-2.5-flash for the validation steps. Forcing the models to return structured JSON using responseSchema was a game-changer, making the integration between the AI and the application's logic seamless and error-free.

What surprised me most was how effectively the AI could be used to build its own guardrails. Prompting the AI to act as a "validation expert" to check the user's input, and later as an "art critic" to validate its own image output, felt like a glimpse into the future of building safe and high-quality AI applications. The entire development process felt less like traditional coding and more like a creative collaboration, guiding an incredibly capable assistant to bring a complex idea to life.

AI Meal Planner App

Shyam Kumar V N — Thu, 30 Oct 2025 07:07:49 +0000

This post is my submission for DEV Education Track: Build Apps with Google AI Studio.

What I Built

I built the AI Healthy Meal Planner, an intelligent application that generates comprehensive, 7-day diabetes-friendly and heart-healthy vegetarian meal plans. It specializes in North/South Indian and Chinese cuisines, generating meal plans from a user-managed dish database. The core functionality relies on a powerful prompt engineering approach with the Gemini API to request a structured JSON output that includes the entire weekly plan, ensuring it adheres to strict health rules like low glycemic index, low saturated fats, and low sodium. I also utilized the AI for dynamic features like generating recipes for new custom dishes and providing instant, health-focused feedback when a user tries to swap a meal.

Demo

https://ai-healthy-meal-planner-554719291717.us-west1.run.app/

My Experience

Working through the track and building the AI Healthy Meal Planner was a highly valuable and surprisingly smooth experience.

What I Learned and Key Takeaways

My biggest takeaway was the power and reliability of structured output when working with the Gemini API. By defining a strict JSON schema for the meal plan, I was able to get a complex, multi-day data structure in a predictable format, which drastically simplified the frontend integration and logic. It taught me that for complex applications, the model isn't just a text generator; it's a powerful, self-correcting data engine if given the right instructions.

I also learned to appreciate the efficiency of using AI for database management logic. Instead of writing complex, rule-based code to check if a meal plan adhered to "diabetes-friendly, heart-healthy" rules, I simply included these constraints directly in the system prompt. The Gemini model handles the complex filtering and balancing of dishes from the database, effectively acting as an intelligent planner and rule-checker in one go.

What Was Surprising

The most surprising aspect was how effective the model was at handling layered constraints and contextual feedback. For instance, when a user attempts a meal swap, the AI not only replaces the meal but provides an immediate, nutritional-based justification for the swap (e.g., "This new dish adds more fiber but slightly increases the sodium for the day"). This level of subtle, context-aware reasoning felt like a huge leap from simpler API calls and opened my eyes to the potential for creating truly interactive and personalized AI features within an application.

DEV Community: Shyam Kumar V N

The Allegorical Illustrator

What I Built

Demo

My Experience

AI Meal Planner App

What I Built

Demo

My Experience

What I Learned and Key Takeaways

What Was Surprising