This is a submission for the Google AI Studio Multimodal Challenge
What I Built
HabitForge AI is a web application designed to solve a fundamental challenge in personal development: the motivation gap.
It’s difficult to stay committed to new habits, whether good or bad, because their long-term effects are invisible in the short term. HabitForge AI bridges this gap by providing an instant, powerful, and personalized visualization of your future self.
It transforms the abstract goal of “getting healthier” into a concrete, emotionally resonant visual, serving as a daily motivator.
Core Capabilities
- Create a Privacy-First Avatar
Users securely upload a photo, which is instantly transformed by AI into a stylized avatar (Cinematic, Anime, Ghibli-style, etc.). This creates a personal yet private canvas for their transformation, making it safe, fun, and share-worthy.
- Visualize Habit Impact
Users can select up to three habits at a time - all positive, or all negative. The app visualizes their cumulative physical, mental, and environmental effects across timeframes (21 days → 5 yrs).
👉 Positive habits reveal the glow-up. Negative ones show the hidden cost of choices like poor sleep, endless scrolling, or inactivity.
- Custom Habit Creation
Go beyond the curated list. Users can add up to two custom habits each (positive & negative). With the Gemini API, the app instantly analyzes the habit, generates an insightful description, and produces tailored visualization prompts on the fly.
- Explore a “Path to Recovery”
Every negative habit has a potential counterbalance. With a single click, users can swap harmful habits for healthier ones and see the rebound effect visualized instantly.
- AI-Powered Coaching
More than visuals, the app doubles as a pocket wellness coach. For any habit, it provides key benefits, common pitfalls, and actionable pro-tips to help users stay consistent.
- Save & Share the Journey
Transformations can be:
- Saved in a personal in-app gallery
- Downloaded as dramatic Before & After composites
- Shared directly to X to inspire others 🌍
User Flow
The experience is designed to be intuitive, inspiring, and seamless:
-
Landing page
- It explains the mission: “Visualize Your Future Self. Today.”
- Clear CTA: Start Your Transformation.
-
Step 1: Avatar Creation
- Upload a photo → choose an art style (Anime, Cinematic, Ghibli, etc.).
- Gemini generates a stylized privacy-first avatar.
-
Step 2: Habit Selection & Visualization
- Avatar displayed in the main visualizer.
- Select up to 3 habits (positive or negative).
- Choose timeframe (21 days → 5 years).
- Click: Show Me The Glow Up.
-
Step 3: The Reveal
- A before-and-after slider reveals transformation.
- Visible effects: healthier skin, energy, or toll of negative habits.
- Background adapts to habits (e.g., meditation → serene room).
-
Engage & Deepen
- Adjust timeframes.
- Get AI Pro Tips.
- Try Path to Recovery swaps.
- Save, download, or share results.
HabitForge AI doesn’t just track habits, it lets you face your future self, today.
Demo
🚀 Try it yourself:
Visualize your habits combo, save your before/after, and share the results with the world 🌍✨ :- HabitForge AI
📽️ Project Demo
Here’s my video walkthrough showcasing all the key features of HabitForge AI:
📸 Project Snapshots
Here are screenshots showcasing the different sections and core functionalities of HabitForge AI:
How I Used Google AI Studio
Google AI Studio isn't just a helper in HabitForge AI, it's the engine that brought the entire applet to life. Every aspect of this project - from designing features, refining UI, coding functionality, and shaping visualization logic, has been built through iterative collaboration with AI Studio. It wasn’t just “code generation”; it became my co-creator, art director, and habit coach, all in one.
🔑 Key Ways I Used Google AI Studio
- End-to-End Coding Partner
I built the entire app through AI Studio prompts - specifying features, constraints, functionality, and UI elements. It generated the code, which I directly integrated into my applet.
- Iterative Suggestions & Testing
AI Studio suggested improvements (functionality, UI, visualization flow). I implemented these one by one, tested them in the live app, and only kept the ones that truly added value. This iterative loop ensured the app grew stronger without unnecessary complexity.
-
AI as an “Art Director” 🎨
For visual transformations, I engineered detailed prompts for thegemini-2.5-flash-image-previewmodel. AI Studio acted as an art director, guiding:- Physiological Manifestations: How habits reflect on skin tone, posture, expressions.
- Intensity Scaling: Subtle vs. dramatic changes depending on timeframe (21 days vs. 1 year).
- Environmental Storytelling: changing the background (e.g., minimalist calm for meditation).
I also iteratively tested different instructional styles, to see how the model would interpret nuanced, abstract concepts like “the inner glow of hydration” or the “subtle mental aura of meditation.” This rapid feedback loop fine-tuned the prompts into acting as a sophisticated art director, ensuring the visualizations weren’t generic, but believable transformations of each user’s specific avatar.
Structured & Reliable Data ⚙️
Using gemini-2.5-flash, I enforced strict JSON schemas for habit analysis and wellness insights. AI Studio helped refine and test the responseSchema, eliminating fragile string parsing and ensuring consistent, clean outputs for the UI.
For text-based features, I also structured prompts and tested the JSON schema for analyzing custom habits and generating coaching insights. This guaranteed reliable, structured outputs that the UI could display without errors.
-
Persona Engineering 👥
I embedded roles directly into prompts:
- Habit Coach Persona: a globally renowned coach, analyzing habits and providing deep, motivational insights.
- Artist Persona: an internationally renowned artist, visualizing the impact of chosen habits with creative precision. Both personas work in tandem - the artist shaping visuals with feedback from the coach, making transformations more believable and relatable.
🔥 With this dual role of engineer + creative partner, Google AI Studio helped me transform HabitForge AI from an idea into a living, interactive experience that people can see, feel, and connect with.
Multimodal Features
HabitForge AI is fundamentally powered by Google’s Gemini multimodal capabilities, combining "image + text inputs" to generate "image + structured data outputs"". This synergy creates a deeply personal, private, and emotionally impactful user experience. All visual and textual processing is handled through specific Gemini models:
-
gemini-2.5-flash-image-preview→ Image generation and avatar/habit visualization -
gemini-2.5-flash→ Text analysis and structured JSON outputs
🌐 Core Multimodal Pipelines
-
Privacy-First Avatar Creation (Image + Text → Image)
- User Action: Upload a personal photo.
-
App Action: Sends the image along with a text prompt (e.g., “Transform this photo into a stylized avatar in the style of Anime/Cinematic/Ghibli…”) to
gemini-2.5-flash-image-preview. - AI Output: Returns a stylized avatar preserving the user’s key features while maintaining privacy.
-
Why It Matters:
- Ensures privacy and trust (no raw photos stored).
- Creates a personalized and shareable experience.
- Encourages user engagement with visually appealing avatars.
Dynamic Habit Transformation Visualization (Image + Text → Image)
This is the heart of HabitForge AI.
- User Action: Select up to 3 habits (positive or negative) and choose a timeframe (21 days → 5 years).
-
App Action: Combines the user’s avatar with a dynamically generated text prompt that encodes:
- Physiological Effects: skin tone, posture, energy, expressions
- Intensity Scaling: subtle for short durations, pronounced for long-term
- Environmental Storytelling: Background changes (e.g., serene space for meditation, chaotic glow for late-night scrolling)
-
AI Output: Transformed “after” image, generated via
gemini-2.5-flash-image-preview, displayed alongside the original in a before-and-after slider. -
Why It Matters:
- Converts abstract habits into tangible, visual outcomes.
- Generates emotional resonance, users can see the effects of their habits.
- More motivating and immersive than stock images or text-only advice.
⚙️ Supporting Multimodal Feature (Text → Structured Data / JSON)
- User Action: Input a custom habit (e.g., “cutting sugar”, “learning to code”).
-
App Action: Sends the text to
gemini-2.5-flashwith a strictresponseSchemafor JSON output. -
AI Output: Returns a complete habit profile:
- Habit name
- Positive/Negative classification
- Descriptive impact summary
- Visualization prompts for the avatar editor
-
Why It Matters:
- Allows user-defined custom habits, beyond the pre-defined library.
- Ensures structured, reliable integration with the UI.
- Adds intelligence and personalization to each user journey.
🙏 Thanks for checking out HabitForge AI!
✨ Try it yourself, share your visualization journey, or drop any questions/thoughts in the comments below, I’d love to hear from you! 💬
Top comments (2)
This is such a wild idea—most habit trackers just nag you with streaks, but yours basically pulls out a crystal ball and says “here’s future you if you keep skipping leg day.” I kinda love (and fear) it.
Thank you 😁
I'm glad you like it.
I just wanted a motivator for me as well yk, something to kinda terrify me and at the same time, motivate me to finish my tasks, hence i got this idea.