Executive Summary
Shinkai is an open-source platform designed to make artificial intelligence (AI) more accessible and useful, similar to how spreadsheets made computing practical for everyday people. It allows users, without needing to code, to create, manage, and share AI agents that can handle complex tasks together, remember past interactions, and work with local or decentralized setups. As of v1 in July 2025, we've seen over 40,000 installations confirmed through web analytics and unique identities from three CoinList testnets, with thousands using the desktop app on macOS, Windows, and Linux, plus more than 330 GitHub stars. Supported by dcSpark, a team experienced in building blockchain tools for chains like Ethereum, Solana, Cardano, and Midnight, and reviewed by auditors PeckShield and Halborn Security, Shinkai works to overcome common AI challenges like limited scope, short-term memory, and integration hurdles.
Existing AI tools are effective for simple things like answering questions or summarizing text, but they often fall short on more comprehensive, ongoing workflows—much like basic calculators before spreadsheets allowed for flexible, repeatable processes. Shinkai approaches this by blending AI planning with peer-to-peer agent communication and targeted blockchain elements for reliable, decentralized interactions. Compared to options like ChatGPT Agents, which are web-based but limited to active sessions without true persistence or scheduling, or Cursor Agents, which excel at code tweaks but cap at 25 API calls per session and lack automation, Shinkai provides features like unlimited scheduled agents, support for heavy tasks (such as video processing), and full Model Context Protocol (MCP) compatibility to integrate with tools like Claude or Cursor.
Building agents is straightforward: users set goals, incorporate tools (for instance, auto-generated code for APIs like YouTube or Jupiter DEX), and set them to run on schedules. Practical examples include an agent that pulls YouTube transcripts and provides summaries, one that applies negotiation principles from established sources, or a Solana-focused bot that identifies price differences (like USDC at $1.01 versus Tether at $0.99), handles trades, and aims to create value—always with crypto keys staying local for better security.
Core aspects of Shinkai include:
- Agent Creation: A no-code setup with memory retention, options to explore alternatives, and the ability to run generated code in Python or Deno. It works with local models through Ollama (over 300 GGUF choices) or remote/cloud options.
- Network and Marketplace: Peer-to-peer connections using blockchain-based identities (e.g., @@alice.shinkai) for secure messaging and privacy via onion routing, plus a community marketplace to share and earn from agents through pay-per-use or reselling API keys.
- Local-First Setup: Runs mainly on-device for privacy and efficiency, with adjustable permissions, file handling, and a sandboxed framework for tools—keeping computations local, which sets it apart from purely web-based systems.
Blockchain use is focused and efficient: The KAI token (Shinkai's utility token) on Base Layer 2 supports one-way lock-ups for registering identities (transferable like NFTs), small payments, and resource access. With a total supply of 1 billion tokens and no ongoing inflation, distributions encourage participation—such as 20% for rewards and 20% for airdrops—while burning fees helps manage supply. This enables creators to monetize by scaling agents, reselling unused credits (like from OpenAI or Gemini), or charging for services, helping build a collaborative ecosystem.
Our user base shows steady interest, with people creating agents for things like tutoring or coaching and sharing them securely. This version of the whitepaper covers the current AI environment, the issues we're addressing, our technical setup (including planning, network, and nodes), comparisons (where Shinkai offers strengths in persistence, earning potential, and local control), token details, governance, and potential risks (like tech vulnerabilities addressed through audits, or market changes we can't control).
Moving forward, Shinkai seeks to make AI a reliable tool for more people, inspired by past tech shifts toward simplicity and adaptability. While future developments come with typical challenges, we're encouraged by the progress and community input so far. Explore and contribute at https://github.com/dcSpark/shinkai-local-ai-agents.