Deep Research Evidence-Aware Handwriting-to-Report Systems A design argument for handwriting-to-report systems that preserve evidence, provenance, uncertainty, abstention, and human review instead of acting like simple OCR.
Deep Research Agent-First Computing and the Fate of Beginner CLI Literacy A research-grounded argument that beginner programming literacy is shifting from command production toward agent supervision, verification, and recovery.
AI Mental Health Frontier AI 心理健康前沿 - Therabot 之后，真正要验证的是照护流程 Therabot 的随机试验让 AI 辅助心理照护更可信，但下一步不是更像治疗师，而是证明监督、升级和随访流程可靠。
AI Mental Health Frontier AI Mental Health Frontier - Therabot Needs Workflow Proof Therabot's first randomized trial is promising, but the real frontier is whether AI-assisted therapy can prove safety inside clinical workflows.
AI Investment Frontier AI 投资前沿 — SciPhy RL 把组合 AI 拉回成本压力之下 一篇新的 arXiv 论文把组合优化写成成本感知的强化学习问题，提醒我们：投资 AI 的关键不只是预测，更是能否在执行成本和仓位约束下把决策真正落地。
AI Investment Frontier AI Investment Frontier — SciPhy RL Puts Portfolio AI Back Under Cost Pressure A new arXiv paper on science-informed reinforcement learning for portfolio optimization is a useful reminder that the real problem in portfolio AI is not prediction alone, but reaching the target position under execution cost and volatility constraints.
AI Investment Frontier AI Investment Frontier — Portfolio Optimization Is a Search Problem Again Cardinality-constrained portfolio optimization is pushing investment AI toward search, repair, and implementation-aware optimization instead of pure prediction.
AI Signals and Reality Checks On-Device AI Is Becoming a Device Stack Problem The important thing is not that AI is moving to the phone; it is that local inference only works when chips, SDKs, and offline actions line up.
AI Investment Frontier AI Investment Frontier — State-First Microstructure Models Beat Fancy Friction A new crypto futures paper argues that liquidity state should come before order flow, and that latency is a separate frontier from prediction accuracy.
AI Signals and Reality Checks AI 路由正在成为真正的利润杠杆 AI 竞争正在从“谁的模型更强”转向“谁的路由更聪明”。真正决定利润和信任的，是任务被送进快路径、慢路径还是受管路径。
AI Signals and Reality Checks AI Routing Is Becoming the Real Margin Lever The AI battleground is shifting from raw model quality to routing. The winners will route work across fast, slow, and governed paths well enough to protect margin and trust.
AI Investment Frontier AI 投资前沿 — 交易频率才是模型能否落地的分水岭 一篇新的跨资产期货研究再次提醒我们：学出来的策略不怕聪明，怕的是太爱交易。真正能落地的模型必须经得住换手和成本。
AI Investment Frontier AI Investment Frontier — Learned Policies Still Live or Die on Turnover A new cross-asset futures paper shows end-to-end AI policies can beat simple rules, but the edge depends on turnover, costs, and whether the model really trades less.
AI Signals and Reality Checks AI 的竞争正在转向浏览器工作，而不是聊天润色 OpenAI 和 Anthropic 最近都在更强调浏览器任务与 agent 执行。这说明竞争焦点正在从“会不会说”转向“能不能在复杂网页里把事做完”。
AI Signals and Reality Checks The AI Race Is Moving Into Browser Work, Not Chat Polish OpenAI and Anthropic are both pushing harder on browser work and agentic workflows. That is a sign the real competition is shifting from model eloquence to task completion in messy, stateful environments.
AI Investment Frontier AI 投资前沿 — 因果分离让组合 AI 更难“看起来有效” 一篇新的 arXiv 论文把投资组合 AI 从“先预测再优化”推进到“先声明驱动变量，再在其诱导几何里优化”，并给出可检验的因果区分。
AI Investment Frontier AI Investment Frontier — Causal Separation Makes Portfolio AI Harder to Fake A new arXiv paper argues that portfolio AI should separate declared drivers from residual risk, turning Markowitz into a projected problem with clearer estimation and causal tests.
AI Signals and Reality Checks 浏览器智能体真正的门槛，是状态连续性 浏览器智能体不再只是“会不会点”。真正拉开差距的是能否保住会话、恢复上下文、绕开登录摩擦，并在出错后继续完成任务。
AI Signals and Reality Checks Browser Agents Are Becoming State-Management Products Browser agents are no longer judged mainly by how well they click. The real product is state recovery, scoped memory, and clean handoff when the page changes underneath them.
AI Investment Frontier AI 投资前沿：模型真正赢在少交易 本周两篇新论文说明，投资 AI 的边界正在从“更准的预测”转向“更稳的策略、更低的换手率、以及更像真实市场的状态估计”。
AI Investment Frontier AI Models Win by Trading Less, Not Just Forecasting Better Two new July papers suggest the edge in investment AI is shifting from better predictions to better policies, lower turnover, and cleaner state estimation.
