Spuric

The problem it solves AI apps have a payments problem: micro-transactions are too expensive, too slow, and too complex for users and developers. Traditional onchain payments require repeated approvals, gas fees, and confirmation delays, which makes “pay-per-action” AI (chat, tools, image/video generation, agents) feel broken. Spuric solves this with ZK-powered AI payments and session-based settlement: Users unlock AI features with a single ZK lock / deposit, then spend instantly off that balance. The app can batch usage and settle onchain efficiently, reducing transaction overhead. Payments become fast, predictable, and UX-friendly, even for frequent small actions. It’s chain-portable: the same UX can switch networks (Base, Kite, etc.) and prompt MetaMask to switch chains when needed. This creates a Stripe-like experience for AI, but with crypto rails — enabling real-time AI usage without forcing users to sign and pay gas for every single action. Instead of paying per request onchain, Spuric uses a ZK lock + aggregated settlement model. That means: far fewer onchain writes lower gas costs faster UX (instant access) better privacy potential (ZK proving/verification patterns can hide granular usage while still proving valid settlement) Use of AI tools and agents Use of AI tools and agents Spuric is a multi-model AI interface with agent-ready workflows that route requests across text, image, and video tools from one unified chat experience, with an onchain-friendly usage gate (lock → metered usage → settlement). Conversation modes (how users interact) Spuric supports multiple conversation modes so users can switch between “what kind of output” they want: Standard Conversation (Chat): default multi-turn LLM chat for Q&A, planning, writing, analysis, and instructions. LLM Conversation Mode: a focused “LLM-only” routing mode used when the user wants strictly text reasoning / code / explanations with minimal extra tool behavior. Initiate / Stop (Run Control): controls agent-style execution loops (start an action, stop an action, prevent runaway generation, and keep the workflow deterministic for demos). Model and tool selection (multi-model routing) Users can choose among multiple AI providers/models from the UI (examples include GPT-style models, Grok-style models, and Gemini-style models), enabling side-by-side capability tradeoffs (speed vs depth vs creativity) without switching apps. AI tools used LLM chat + reasoning: multi-turn reasoning, code generation, structured outputs, and tool planning. Image generation: prompt → image output returned in the same interface. Video generation / imagining: prompt → video or video-ready output displayed in the WebGL UI (where supported). WebGL media utilities: browser-friendly media handling for demo UX (inline/overlay playback and file input when enabled). Agent-style workflow (how tools work together) Spuric follows an agent loop pattern: User intent → Orchestrator: the core chat manager collects context and the selected mode/model. Routing: based on mode + user action (chat/image/video), the request is routed to the right tool. Execution + rendering: outputs return into one timeline (text + media), enabling fast iteration. Iterate: users refine prompts and rerun quickly without leaving the conversation. AI + blockchain integration (usage gating) AI usage is paired with a session-based payment gate: Wallet connect establishes identity/state in the app. A single lock/deposit can unlock high-frequency AI interactions without requiring a transaction for every action. Usage can be metered and later settled efficiently, supporting scalable pay-per-use patterns (optionally compatible with ZK verification approaches for privacy-preserving proofs). Summary Spuric combines multi-model AI tools (chat, image, video) with multiple conversation modes (standard chat, LLM-focused chat, and run control via initiate/stop) to enable reliable, agent-style workflows in a demo-ready WebGL experience.

Spuric

Demo

About

How it's made