review items - nov 2025 - part2

[kimtth]

Azure

• Magentic-Marketplace✨: An Market Simulation using Agents [27 Oct 2025] 📑
• AGDebugger✨: Built on top of AutoGen — a visual tool for agent debugging. [Jan 2025]
• PIKE-RAG✨: sPecIalized KnowledgE and Rationale Augmentation Generation (PIKE-RAG) [20 Jan 2025]
• AsyncThink📑: An asynchronous thinking protocol between an Organizer and Workers using fork–join actions, enhanced from the general sequential and parallel thinking approaches. [30 Oct 2025]
• Call center AI✨: Send a phone call from AI agent, in an API call. Or, directly. [Jan 2024]
• Agent 365✍️: The control plane for AI agents. Unified observability with Registry, Access Control, Visualization, Interoperability, and Security. [18 Nov 2025]
• Work IQ✍️: AI agents smarter with enterprise context. The intelligence layer behind Microsoft 365 Copilot and agents that helps Copilot understand you, your job, and your company. | Fabric IQ | Foundry IQ [18 Nov 2025]
• Microsoft Agent Factory✍️: a new program designed for building agents with Work IQ, Fabric IQ, and Foundry IQ using Microsoft Foundry and Copilot Studio [19 Nov 2025]
• Azure HorizonDB✍️: a New PostgreSQL Service. Advanced DiskANN vector indexing. Scaling (15 replicas with 192 vCores each, Auto-Scaling 128TB). [19 Nov 2025]
• Word, Excel, and PowerPoint Agents in Microsoft 365✍️: Dedicated agents that help users create documents, spreadsheets, and presentations [19 Nov 2025]
• BuildYourOwnCopilot✨: RAG demo using CosmosDB, Azure Kubernetes Service, and Semantic Kernel. [Dec 2023]

Items

• Kimi K2 Thinking✍️: The first open-source model beats GPT-5 in Agent benchmark. [7 Nov 2025]
• GPT 5.1✍️: GPT-5.1 Auto, GPT-5.1 Instant, and GPT-5.1 Thinking. Better instruction-following, More customization for tone and style. [12 Nov 2025]
• Nested Learning: A new ML paradigm for continual learning✍️: A self-modifying architecture. Nested Learning (HOPE) views a model and its training as multiple nested, multi-level optimization problems, each with its own “context flow,” pairing deep optimizers + continuum memory systems for continual, human-like learning. [7 Nov 2025]
• PageIndex✨: a vectorless, reasoning-based RAG system that builds a hierarchical tree index [Apr 2025]
• LEANN✨: The smallest vector database. 97% less storage. [Jun 2025]
• Memori✨: a SQL native memory engine (SQLite, PostgreSQL, MySQL) [Jul 2025]
• Grok 4.1✍️ [17 Nov 2025]
• Gemini 3 Pro✍️: Deep Think reasoning, Advanced multimodal understanding, spatial reasoning, and agentic capabilities up 30% from 2.5 Pro — reaching 37.5% on Humanity’s Last Exam (41% in Deep Think mode). [18 Nov 2025]
• Vibe Hacking✍️: Anthropic reports vibe-hacking attempts. [14 Nov 2025]
• Gemini Memory✍️: Gemini uses a structured, typed “user_context” summary with timestamps, accessed only when you explicitly ask. simpler and more unified than ChatGPT or Claude, and it rarely uses data from the Google ecosystem. [19 Nov 2025]
• Google Antigravity✍️: A VSCode‑forked IDE with an artifacts concept, similar to Claude. [18 Nov 2025]
• ModelScope-Agent✨: Lightweight Framework for Agents with Autonomous Exploration [Aug 2023]
• Solving a Million-Step LLM Task with Zero Errors📑:
• mgrep
• OLMo 3
• GPT-5.1 Codex Max
• Google CodeWiki
• Agent-R1
• Tinker Cookbook, verl, RLinf
• MineContext