“Omamori.” — A Shrine-Support Prototype that Uses IO intelligence to Trial a “Favorite Deity AI × Behavior Analytics AI” Approach and Test Whether Donations Can Become a Daily Habit

Testing a “Habit-Forming UX” Hypothesis Built in Two Weeks

“Omamori.” by Team Pooh-san is an app developed over roughly two weeks to test the hypothesis:

“Can we create an experience where supporting shrines continues as part of everyday life?”

In response to the financial challenges many shrines face, the team built a working prototype to explore how to design both the entry point to donating and an experience that encourages continued support.

This article focuses on:

• Why they chose to combine a character-based AI (Favorite Deity AI) with a behavior analytics AI
• Where and how IO intelligence was used in that setup
• Why the cost and scaling properties of distributed GPUs matter for this kind of idea

1. Framing the Problem: Wanting to Donate Isn’t Enough to Sustain It

Shrines serve as cultural and local community anchors, yet many struggle to secure the funds needed for upkeep, putting their long-term survival at risk. From a supporter’s perspective, however, the key issue isn’t only the need for donations — it’s that the donation experience often doesn’t continue.

• Donations easily become a “one-time thing when you remember”
• Actions end after a single step
• The impact feels distant, making it hard to internalize as part of daily life

To change this structure, the team’s starting point was the idea that donations shouldn’t remain a “special event,” but instead should be brought closer to an everyday habit that can naturally continue.

2. Core Hypothesis of the Solution: A Two-Layer AI Structure

The experience design of “Omamori.” uses AI in two layers, each with a different role.

2–1. Character AI (Favorite Deity AI): Supporting Continuation on the Emotional Side

• The app centers on daily conversations with a personified deity (Favorite Deity AI).
• The team aims to test whether this dialogue can function as emotional support and a point of connection similar to visiting a shrine — giving users a reason to keep logging in and engaging each day.
• Rather than being an AI for casual chat, it is envisioned as a companion that supports people emotionally before the act of donating.

2–2. Behavior Analytics AI: Fine-Tuning Continuation on the Behavioral Side

• The app periodically analyzes behavior logs such as logins, chats, and fortune draws.
• By detecting trends in continued use and changes in donation-related actions, the goal is to connect those insights to encouragement or suggestions tailored to each user.
• At this stage, the focus is simply on verifying whether this loop can work in practice through a prototype.

The core of “Omamori.” is the idea that the Character AI can serve as emotional support, while the Analytics AI can act as a habit-adjustment mechanism — and that this hypothesis can be tested in a short build cycle with a working prototype.

3. Where IO intelligence Fits In: Implementing AI as the Center of the Experience

As a prototype, “Omamori.” needed to place AI at the heart of the user experience. IO intelligence APIs were used to implement that setup.

• Real-time responses for the Favorite Deity AI chat AI powers the dialogue layer that anchors the habit-forming experience.
• Periodic analysis of user behavior logs AI is used in daily/weekly cycles to observe habit trends and feed improvements.

In this project, IO intelligence was a practical choice to quickly shape these components into a working prototype and bring them to a testable state within a short period.

4. The Value of Distributed GPUs: Making Always-On AI Economically Feasible

“Omamori.” is based on the assumption that AI runs continuously as part of everyday use. That makes operational cost realism just as important as user experience.

With IO intelligence’s distributed GPU approach, the team expects:

• It’s easier to plan for parallel AI conversations across many users
• Costs are less likely to spike as continued use grows
• Operational expenses can be significantly lower than traditional centralized clouds (roughly ~70% reduction at the estimation level)

Having this scaling and cost perspective even at the prototype stage is important, because it allows the team to realistically consider whether the system would remain sustainable if the habit-forming UX proves effective.

5. What the MVP Reached: A Minimal Form to Test the Experience Hypothesis

Within the two-week constraint, the team built an MVP centered on Next.js and Supabase, with IO intelligence as the core AI integration.

Technical Stack (MVP)

• Frontend: Next.js 14 / TypeScript / Tailwind / shadcn/ui
• Backend: Supabase (DB / Storage / Edge Functions)
• AI Integration: IO intelligence API
• Hosting: Vercel + Supabase Cloud
• Auth: guest_id-based anonymous usage (no login required)

Implemented Features (Minimum Set for Validation)

• Favorite Deity AI chat
• Daily “rituals/tasks” (entry point for habit-building actions)
• Donation dashboard (visualization)
• Shrine fortune feature
• Point management / shrine listings / rankings, etc.

6. Conclusion: What This Prototype Suggests

“Omamori.” is a prototype that explores a big theme — supporting shrines — through an AI-centered attempt to see whether donations can move toward a habit-like experience.

The team built a working form to test whether:

• The Favorite Deity AI can drive attachment and continuation
• Feedback from behavior analysis can help shape habits
• Operational costs remain realistic even under always-on AI assumptions

Bringing these questions into a testable, functioning prototype within a short timeframe has value in itself.

From the IO intelligence perspective, the most notable takeaway is that in prototypes assuming everyday, continuous AI use, distributed GPU cost and scaling properties provide tangible realism — making this a meaningful early use case.

[embed]

[embed]おまモリ。 - 神社×クリック募金×推し活で、「あなた」と「神社」をつなぐ - Manus Manus is the action engine that goes beyond answers to execute tasks, automate workflows, and extend your human reach.manus.im