Varenith
AI-powered visual novels meet on-chain betting and fair storytelling.
Videos
Description
I. Project Overview
1. User Persona
Our core users are young adults aged 18 to 30 who are passionate about anime-style storytelling, interactive fiction, and AI-generated content.
They crave deep narrative immersion, enjoy exploring multiple outcomes based on player choice, and are generally familiar with Web3 concepts. These users are curious, adventurous, and drawn to experiences where “every decision has a cost”—especially when there's a chance to place bets on story outcomes.
They don’t just want to watch the story unfold—they want to shape it, gamble on it, and own their decisions on-chain.
2. Use Case
Our product is an AI-powered visual novel platform where players step into a rich narrative universe driven by multiple AI Agents. These agents collaborate to dynamically generate storylines, control NPC behavior, and evolve the world state in real time based on player input.
At major story nodes, players can do more than just choose—they can place on-chain bets on how the narrative unfolds. Whether it’s wagering on which character survives or which branch wins in popularity, smart contracts ensure transparent and tamper-proof outcomes.
This creates a high-stakes, interactive story environment that is:
Deeply immersive
Player-controlled
Fully verifiable via blockchain
It’s not just a story—it’s a live, evolving game of chance and agency, where every player becomes both a protagonist and a stakeholder.
3. Product Positioning
We’re building a next-generation visual novel platform that combines:
Multi-AI Agent orchestration
Multimodal, immersive story interaction
On-chain narrative wagering and DAO-driven fairness
In short: "AI-powered narrative immersion meets Web3 verifiable gameplay."
Compared to traditional static visual novels like Mystic Messenger or Undertale, or existing GameFi projects focused solely on DeFi mechanics, our platform offers something fundamentally new:
An evolving narrative metaverse where players can wager on decisions, influence characters, and co-create the world in collaboration with AI.
This isn’t just another game—it’s a sandbox for narrative experimentation, a bridge between storytelling and decentralized systems, and a step toward a fairer, co-owned creative future.
II. Product Feature Design
1. Multi-Agent Narrative Engine (Character & Plot Agents)
Our narrative engine is orchestrated by multiple AI Agents, each controlling distinct characters or narrative functions. These agents possess individual memory systems, personalities, and contextual awareness.
All story branches are generated in real-time by AI, based on player input and environmental signals. We support two distinct interaction modes:
Immersive Visual Mode: Character portraits, scene images, dialogues, and branching interactions in visual novel style.
Text-Only Novel Mode: For players who prefer deep-reading and flexible decision trees without visual elements.
Each narrative decision is co-generated through agent collaboration, creating a truly living and evolving story universe.
2. Web3 Narrative Betting System
At key narrative branches, players are able to place on-chain bets on the direction the story will take.
They can wager on outcomes such as:
Which character survives
Which storyline becomes the canonical branch
Community choices or narrative popularity
Smart contracts ensure fair, tamper-proof resolution and automatic reward distribution. This transforms storytelling into a high-stakes, emotionally immersive betting experience backed by Web3 infrastructure.
3. AI-Powered NPCs (Character Plaza & Story Plaza)
We introduce two key NPC systems:
Character Plaza: A public space where AI-controlled NPCs live on-chain. These agents retain memory, interact with players, evolve over time, and offer quest clues or dynamic interactions.
Story Plaza: A sandbox where players can create their own characters and branching stories, attach bounties, and wait for others to play, challenge, or extend the narrative.
Players can essentially build and own their own playable AI characters and plots, contributing directly to the game’s living ecosystem.
4. CARV ID Reputation & Identity Layer
We integrate a modular identity and reputation system powered by CARV ID. Players can log in via MetaMask, Google, or Discord, unifying Web2 and Web3 identities.
User behavior—such as gameplay, decision quality, creation activity, and betting participation—is continuously tracked and scored. This reputation system determines:
Access to advanced narrative branches
Reward multipliers
Voting and governance rights
Reputation evolves as players engage and contribute to the platform meaningfully.
5. AI Agent-Governed DAO (Anti-Sybil, Scalable Governance)
To ensure fairness and resist manipulation, AI Agents support DAO governance throughout the platform.
When community votes are initiated (e.g., to determine which branch becomes canonical), the system generates AI-curated questionnaires that scale with reputation weight and complexity. This approach:
Deters bots and sybil attacks
Surfaces high-quality input from real users
Keeps the governance process transparent and fair
Outcomes are validated via a reputation-weighted consensus, combining the intelligence of the community with the scalability of AI.
III. Track Relevance & Evaluation Fit
🏆 A. Tech Fairness Exploration Awards
Evaluation Criteria | Varenith Implementation | Differentiation |
|---|---|---|
Originality | AI-driven visual novel with real-time branching, multi-Agent control, and on-chain wagering | While many projects focus on static storytelling or isolated chain interactions, Varenith integrates dynamic AI storytelling, agent-based emotional progression, and user-driven bet-based narratives in one system. |
Algorithm Transparency | All key decisions—narrative outcomes, bets, and reward flows—are executed on-chain via smart contracts | Players can fully trace how a story evolved and how winnings were calculated. Unlike other projects relying on off-chain verdicts, our logic is open, auditable, and deterministic. |
Data Sovereignty | Unified CARV ID system binds Web2/Web3 identity; all user data is exportable | Players retain full visibility into their behavioral data. No third-party holds full control; instead, the platform uses hashed logs with opt-in data mapping. |
Sustainable Income | Users can earn by uploading stories or characters, setting bounties; AI assists long-tail creators in generating content | Unlike reward systems that favor influencers or core teams, we empower small creators with modular monetization—every narrative branch can earn via interaction. |
Design & Clarity | Dual interaction modes (visual/text), structured UIs (story plaza, task board, governance panel) | Most GameFi products have complex UIs and learning curves; our design allows new users to join, interact, and create within minutes, no tutorials needed. |
🏛 B. FAIR3 Public Infrastructure Awards
Evaluation Criteria | Varenith Implementation | Differentiation |
|---|---|---|
Modularity | All core systems (narrative engine, NPCs, betting, reputation) are modular and exportable as SDKs | Unlike projects bundling all logic into single contracts, our components are plug-and-play—ideal for external DAOs or games to adopt specific layers. |
Interoperability | Full integration with MetaMask, Google, Discord, and CARV ID | Other infra projects typically support only wallet login; Varenith bridges mainstream accessibility with decentralized tech. |
User-Facing Infrastructure | Includes fully designed frontends: NPC plaza, governance dashboard, story creation panel | Frontend + contract + API layers are all provided, reducing integration overhead for partner platforms. |
🧠 C. SVM On-Chain AI Agent Infrastructure
Evaluation Criteria | Varenith Implementation | Differentiation |
|---|---|---|
On-chain Execution | NPCs run as SVM-executed agents with persistent memory and logs stored on-chain | Unlike IPFS-based memory or off-chain AI calls, NPC actions and learning histories are provable and globally replayable. |
Dynamic GameFi NPCs | Characters react and evolve based on user decisions; emotions and alliances change in real time | Most NPCs are either scripted or static; ours exhibit true behavioral variance and can reflect user reputation and prior interactions. |
Shared Memory & Coordination | Multi-agent story generation with contextual handover and synchronized dialogue states | Enables story arcs to span multiple agents and sessions, building long-term consistency not achievable in single-agent setups. |
📡 D. D.A.T.A. Decentralized Data Coordination
Evaluation Criteria | Varenith Implementation | Differentiation |
|---|---|---|
Data Lifecycle | Full loop from user input → on-chain logs → cleaned behavior → AI inference → applied rewards or branch changes | Other projects may only collect data without application; Varenith completes the circle by pushing outputs directly into gameplay and governance. |
AI Reasoning Layer | Structured behavioral logs drive prediction (e.g., branch preference, player trust level) | Instead of text-only embeddings, our engine relies on structured schemas—lowering semantic drift and improving narrative precision. |
Tooling Integration | Native use of CARV D.A.T.A. pipeline for schema standardization and reputation feedback | Our schema maps every player action to a game event and score, allowing reuse in other AI/DAO settings. |
🆔 E. Modular Identity & Reputation with CARV ID (ERC-7231)
Evaluation Criteria | Varenith Implementation | Differentiation |
|---|---|---|
Multi-Identity Binding | Supports MetaMask, Discord, Google; merges into ERC-7231-compliant CARV ID | Most solutions treat Web2 and Web3 identity separately—Varenith unifies them into a verifiable profile reusable across platforms. |
Reputation Calculation | Uses player behavior (branch creation, voting accuracy, governance participation) with weighted scores | Reputation logic is transparent and configurable. Players can trace their score sources and influence proposals through score weight. |
Extensibility & Compliance | ERC-7231 compatible; supports ID-bound NFTs and Quest systems | Enables cross-project features such as governance badges, gated content, or ID-tied access layers—beyond static credentials. |
🌍 F. Open Innovation: AI × Web3 for Real-World Use Cases
Evaluation Criteria | Varenith Implementation | Differentiation |
|---|---|---|
Real-World Viability | Creators can monetize via story uploads and interactive NPCs; long-tail users get distribution tools (e.g., TikTok auto video exporter) | Helps underserved creators break through visibility barriers without requiring prior fame; actual use in content economies, not just theoretical. |
Multi-Agent Collaboration | Orchestration of story agents, proposal agents, governance evaluators | Not just multiple bots—agents coordinate to generate narrative arcs, vote logic, and player prompts with continuity. |
User & Data Protection | All user data is exportable, opt-in logged, and bound to CARV ID | Players control what data is stored; logs are hashed and auditable, reducing data abuse risk. |
Progress During Hackathon
(1) AI Narrative Engine We’ve completed a multi-agent narrative system where distinct AI Agents control characters with memory and dynamic reactions. The platform supports real-time, multimodal visual-novel interaction and adaptive story branching based on player input. A narrative path-tracing system is in progress to improve continuity and replayability. (2) CARV ID Modular Identity & Reputation System Login via MetaMask, Google, and Discord is fully integrated, enabling seamless Web2 and Web3 access. CARV ID binding and behavior tracking are operational, with a dynamic reputation scoring visualization currently in development. (3) Data Governance & Pipeline We’ve implemented on-chain logging for all key user actions, including bets and decisions. A structured data pipeline is active, and a standardized schema has been finalized for AI training. Data export and self-sovereign control are under integration. (4) AI Agent DAO Governance AI Agents can now generate branching proposals and governance prompts automatically. Questionnaire-driven anti-Sybil mechanisms are functional, and voting is now weighted by user reputation. An interactive task board and proposal UI are in active development. (5) Web3 Narrative Betting System Smart contracts for narrative outcome betting are live, allowing players to wager on branching paths. Reward distribution logic is deployed, and reputation-based betting limits and feedback loops are currently being finalized. (6) Automated Video Content Generation We’ve built a story-to-video scripting system that assembles narrative content into visual scenes. Summarization and AE scripting pipelines are under active development, alongside a one-click social video publishing workflow.
Tech Stack
Fundraising Status
We have secured $50,000 in early-stage funding and are currently incubated by both Beta University and the Founder Institute. Project timeline and key pivots: (1) February 2025 – We launched an AI-powered children's storybook product focused on generative storytelling and interactive education for kids. While technically solid, the project encountered commercial challenges due to a long and inefficient user conversion path—from parents to children to payment—and was ultimately discontinued. (2) April 2025 – The team pivoted to building a custom AI bidding Agent SaaS for enterprise clients, assisting with RFP generation and tender automation. This enterprise product successfully went live and is now operated by a domestic partner. (3) Late May 2025 – We began prototyping the current project as an AI-driven interactive narrative engine, featuring early-stage multi-Agent story generation and visual storytelling systems. (4) Late June 2025 – We started integrating Web3 infrastructure into the narrative engine, including CARV-based identity and reputation, on-chain betting mechanisms, and DAO governance tools—transforming the product into a next-generation AI × Web3 immersive storytelling platform.