NEOLAF builds persistent AI companion infrastructure — where one human brain + one AI brain grow together through shared memory, creating a cognitive partnership more capable than either alone. NEOLAF 构建持久化的 AI 伴侣基础设施——一个人脑 + 一个 AI 脑，通过共享记忆共同成长，形成超越任何一方的认知伙伴关系。
Learn More 了解更多For 150 years, education and enterprise optimized for what humans can memorize and recall. AI has commoditized that. What remains scarce: judgment, question framing, orchestration, and continuous learning. 150年来，教育和企业一直在优化人类的记忆与回忆能力。AI 已将其商品化。真正稀缺的是：判断力、问题构建、流程编排和持续学习能力。
Knowledge locked in one head. Learning = memorize, recall, test. Value = what you know. Scaling = more people. 知识锁在一个脑袋里。学习 = 记忆、回忆、考试。价值 = 你知道什么。扩展 = 更多人力。
Human + AI companion that grows with you. Learning = collaborate, produce, reflect. Value = what you can do. Scaling = agent leverage. 人类 + 与你共同成长的 AI 伴侣。学习 = 协作、产出、反思。价值 = 你能做什么。扩展 = 智能体杠杆。
Not an upgrade — a recomposition. Like horses to automobiles: a change in the fundamental mechanism, not an optimization of the old one. 不是升级——是重组。如同从马车到汽车：根本机制的改变，而非旧模式的优化。
NEOLAF replaces the industrial-era learning model with three pillars designed for the dual-brain era. NEOLAF 用三大支柱替代工业时代的教学模式，为双脑时代而设计。
AI companion adapts to each learner through Socratic dialogue — not passive content delivery. AI 伴侣通过苏格拉底式对话适配每位学习者——而非被动内容推送。
Students produce real work artifacts using AI as collaborator. Capability is built through doing. 学生以 AI 为协作者产出真实工作成果。能力在实践中构建。
Structured after-action review turns experience into expertise. Portfolio replaces transcript. 结构化复盘将经验转化为专长。作品集替代成绩单。
A neuro-symbolic cognitive architecture that gives every AI companion persistent memory and a self-improving learning loop. 一种神经符号认知架构，赋予每个 AI 伴侣持久记忆和自我改进的学习循环。
Every interaction is structured and retained. Knowledge compounds across sessions, semesters, and careers. 每次交互都被结构化保留。知识在会话、学期和职业生涯中持续积累。
Combines LLM fluency with symbolic precision. Creative problem-solving meets rigorous validation. 融合大模型的流畅性与符号系统的精确性。创造性解决问题与严格验证并存。
The agent learns from the mismatch between predicted and actual confidence — self-supervised, always on. 智能体从预测置信度与实际置信度的偏差中学习——自监督，持续运行。
| ChatGPT / Generic AI ChatGPT / 通用 AI | NEOLAF + KSTAR | |
|---|---|---|
| Memory记忆 | Stateless (per session)无状态（按会话） | Persistent (longitudinal)持久化（纵向） |
| Architecture架构 | Purely neural纯神经网络 | Neuro-symbolic神经符号融合 |
| Scope范围 | Individual sessions单次会话 | Individual + institutional个人 + 机构 |
| Role角色 | Tool工具 | Cognitive partner认知伙伴 |
| Learning学习 | None (frozen after training)无（训练后冻结） | Continuous (KSTAR delta)持续改进（KSTAR 增量） |
When generating a tool costs less than finding one, the entire logic of application-centric software inverts. NEOLAF provides the architecture for this transition. 当生成工具的成本低于寻找工具的成本时，以应用为中心的软件逻辑将全面反转。NEOLAF 为这一转型提供架构支撑。
Composable, versioned, dependency-aware capability units. The new atomic unit of software. 可组合、可版本化、感知依赖的能力单元。软件的新原子单位。
The human directs; the agent translates intent into action. Judgment replaces coding as the core competency. 人类指引方向，智能体将意图转化为行动。判断力取代编码成为核心能力。
Sensor, Actuator, Skill, Context, Authorization — all agentic systems compose from these five primitives. 感知器、执行器、技能、上下文、授权——所有智能体系统由这五种原语组合而成。
Realized Action = Plan ⊗ Token ⊗ Energy. An action plan without authorization is just a wish. 实现的行动 = 计划 ⊗ 令牌 ⊗ 能量。没有授权的行动计划只是一个愿望。
Semantic → Planning → Capability → Authorization → Execution. The UI is optional; the interface is primary. 语义层 → 规划层 → 能力层 → 授权层 → 执行层。UI 可选；接口为先。
Understanding = structure-preserving transformation. Agents transfer patterns across domains, not just match symbols. 理解 = 保结构变换。智能体跨领域迁移模式，而非简单匹配符号。
NEOLAF leads the development of open standards for agent communication, learning records, and knowledge representation. NEOLAF 引领智能体通信、学习记录和知识表示的开放标准制定。
IEEE P3394 enables agent-to-agent interoperability across any channel: MCP, HTTP, Slack, WeChat, WebSocket. IEEE P3394 实现跨任意通道的智能体互操作：MCP、HTTP、Slack、微信、WebSocket。
Every module only needs to understand UMF. Arbitrary composition becomes possible — the agent equivalent of the Unix pipe. 每个模块只需理解 UMF。任意组合成为可能——相当于智能体世界的 Unix 管道。
KSTAR accumulates knowledge that compounds with every interaction — at individual and institutional level. KSTAR 积累的知识随每次交互复合增长——覆盖个人和机构层面。
Deep integration with university workflows creates high switching costs and long-term partnerships. 与高校工作流的深度集成创造高转换成本和长期合作关系。
Years of domain expertise in learning science, instructional design, and AI-augmented pedagogy. 多年学习科学、教学设计和 AI 增强教学法领域的积累。
Shaping IEEE standards for agent communication and educational technology interoperability. 主导 IEEE 智能体通信和教育技术互操作标准的制定。
Verified capability portfolios replace traditional transcripts as employers adopt new talent signals. 经验证的能力作品集取代传统成绩单，成为雇主认可的新人才信号。
Each institution's deployment enriches the platform. The data moat grows more valuable with every cohort. 每个机构的部署丰富平台。数据护城河随每届学生变得更有价值。
The future of intelligence is not AI replacing humans. It is one biological brain + one AI brain, coordinated by shared memory, growing together. 智能的未来不是 AI 取代人类。而是一个生物大脑 + 一个 AI 大脑，通过共享记忆协同合作，共同成长。
Intent, values, judgment, taste, domain expertise, and the questions worth asking. 意图、价值观、判断力、品味、领域专长，以及值得提出的问题。
Retrieval, planning, execution, scaling, and tireless consistency. 检索、规划、执行、扩展，以及不知疲倦的一致性。
A cognitive partnership more capable than either alone. Knowledge that compounds over a lifetime. 超越任何一方的认知伙伴关系。跨越一生持续积累的知识。
NEOLAF is building the cognitive infrastructure layer between humans and AI — in education, enterprise, and beyond. NEOLAF 正在构建人与 AI 之间的认知基础设施层——覆盖教育、企业及更广泛的领域。
Richard Tong | Founder & CEO | richard.tong@ieee.org