Hark Pro (Hark)
Brett Adcock's Hark Pro, a free personal agent
AI Year in Review · Oct 10, 2025 – Oct 10, 2026
AI agents & coding agents
146 stories in this category.
Grok Bot model routing (xAI)
Grok Bot routes tasks to Claude Opus 5.5, Midjourney and Suno
Photocraft (ArtCraft)
Photocraft and the ArtCraft suite: open-source Adobe replacements in Rust
Hermes Index (Nous Research)
Nous Research launches the Hermes Index; Claude Opus 5.5 leads at 63.31
Personal Agent Protocol (Meta AI)
Meta and Sierra announce the Personal Agent Protocol
Agent Lens (CoreWeave)
CoreWeave Agent Lens brings human-friendly insights to agent observability
Forge (CoreWeave)
CoreWeave launches Forge, the whole agentic loop in one place
NVIDIA Vera CPU (CoreWeave)
CoreWeave to offer NVIDIA Vera, the first CPU built for AI agents
Sandboxes (CoreWeave)
CoreWeave Sandboxes go GA as part of Forge
Monthly roundup: ThursdAI Oct 2026 ↗
Monthly roundup
Dots (OpenAI)
OpenAI launches dots, always-on ChatGPT agents on GPT-6 Astra
Codex cloud (OpenAI)
Codex now runs fully in the cloud
UltraFast mode (OpenAI)
OpenAI launches UltraFast mode on Cerebras with a $500 Pro tier
Decisions API (OpenAI)
OpenAI previews a Decisions API for fast fixed-choice answers
cf CLI (Cloudflare)
Cloudflare goes agent-first with cf, one CLI for its whole dashboard
jevgrep (jevgrep)
jevgrep: a Jev-powered CLI that claims 40% cheaper coding agents
OpenAI "Dots" persistent agents
Always-on agents with their own cloud computer/browser that keep working while your device is off; Agents API gets Computer Use.
Grokbot (xAI)
Grokbot connects to Tesla and 1Password
Monthly roundup: ThursdAI Sep 2026 ↗
Monthly roundup
Agents API (OpenAI)
OpenAI ships the Agents API in public beta: the Codex harness as a managed service
Assistant Benchmark (Merit Systems)
Assistant Benchmark ranks 116 submitted AI assistants across 16 hand-tested dimensions
Projects (Cursor)
Cursor launches Projects: one view for agents, chats and work
Trusted Person network (Instinct)
Instinct adds email and a Trusted Person network for agent-to-agent coordination
Navier-Stokes solution claim (agent swarm) (OpenAI)
OpenAI claims a Navier-Stokes Millennium Prize solution from a 10,000-agent swarm
Automated research intern milestone (OpenAI)
OpenAI says it reached its automated research intern milestone, targets an automated AI researcher by March 2028
Muse Code (Meta AI)
Muse Code leaves beta with $5 / $20 / $50 plans and a TypeScript SDK preview
Codex with GPT-6 Astra (OpenAI)
Codex gets GPT-6 Astra: notes across context windows and async questions
OpenClaw 2.0 (OpenClaw)
OpenClaw 2.0: native computer use via Cua Driver, cloud fleets, 16,977 PRs
ChatGPT Work (OpenAI)
ChatGPT Work adds website sign-in via credential handoff
Foundation (Chroma)
Chroma launches Foundation: unified memory for your agents
Hermes desktop bot mode (Nous Research)
Nous Research ships bot mode for Hermes desktop
OpenBot (CopilotKit)
CopilotKit releases OpenBot, an open take on the bot pattern
Computer History (Cua)
Cua open-sources Computer History, memory for computer-use agents
Mojo (Modular)
Mojo goes fully open source under Apache 2.0
DeepSeek Harness (DeepSeek)
DeepSeek ships its own open-source agent harness, 23K stars in days
Cloudflare OS (Cloudflare)
Cloudflare OS: Kenton Varda's Sandstorm reborn as open source agent infrastructure
Muse Code + Muse Spark 1.2 (Meta AI)
Meta ships Muse Code, a terminal coding agent that's 12-21x cheaper if you feed Meta your data
Prime Agent (Prime Intellect)
Prime Intellect's Prime Agent: a self-improving RLM harness claiming 95.5% on ARC-AGI 3's public set
DeepSeek V4-Flash (DeepSeek)
DeepSeek V4-Flash enters public beta, beating its bigger sibling on agent benchmarks at $0.14/$0.28
Production-trained router (Cursor)
Cursor launches a router trained on production traffic: near-Fable satisfaction at ~60% lower cost
Terminal Bench 2.0 — GPT-5.6 results (Wolfbench)
Wolfbench: GPT-5.6 Sol on max thinking beats GPT-5.5 on both score and cost
Codex + ChatGPT Work (OpenAI)
Codex and the unified ChatGPT Work app cross 9 million active users
Codex Micro (OpenAI)
OpenAI ships its first hardware: a $230 Codex Micro keypad built with Work Louder
DOOMQL (Peter Gostev)
A playable Doom clone in ~2,000 lines of SQL, one-shotted by GPT-5.6 Sol Ultra
ChatGPT for Work (unified app) (OpenAI)
Codex becomes the unified ChatGPT app, with Work mode and hosted Sites
ZCode (Z.ai)
Z.ai launches ZCode, a GLM-5.2 agentic coding environment
Monthly roundup: ThursdAI Jul 2026 ↗
Monthly roundup
Aria (Weights & Biases)
W&B Aria auto-research agent goes GA
Linzumi (Linzumi)
Linzumi launches shared chat for fleets of AI coding agents
Fugu (Sakana AI)
Sakana AI launches Fugu multi-agent orchestration API
Codex Computer Use in Europe (OpenAI)
OpenAI rolls out Codex Computer Use, Chrome extension, Memory and Chronicle to European users
Kimi K2.7 Code on CoreWeave Inference (Weights & Biases / CoreWeave)
Kimi K2.7 Code goes live on W&B/CoreWeave Inference
HiveMind (Weights & Biases)
Weights & Biases launches HiveMind for coding-agent observability
Devin Desktop (Cognition Labs)
Cognition rebrands Windsurf into Devin Desktop multi-agent hub
Hermes Desktop (Nous Research)
Nous Research launches Hermes Desktop agent app for Mac/Win/Linux
RTX Spark (NVIDIA)
NVIDIA announces RTX Spark Arm + Blackwell platform for local AI PCs
Agent Arena (Arena (LMArena))
Arena launches Agent Arena for real-world agent workflow evals
Universal Cart / AP2 / UCP (Google)
Google launches Universal Cart, AP2 and UCP for agentic commerce
AI Studio native Android apps (Google)
Google AI Studio builds free native Android apps; 250K in week one
Cua Driver for Windows (Cua)
Cua Driver brings background computer-use agents to Windows
DeepSWE (Datacurve)
Datacurve's DeepSWE: a contamination-free coding benchmark
Antigravity 2.0 (Google)
Antigravity 2.0 becomes Google's central agentic coding harness
Gemini Spark (Google)
Gemini Spark announced as a 24/7 proactive personal AI agent
Google Search agentic capabilities (Google)
Google Search adds Gemini 3.5 Flash-powered agentic capabilities
Codex Mobile (OpenAI)
OpenAI Codex Mobile arrives in the ChatGPT mobile apps
Managed Agents (Gemini API) (Google DeepMind)
Gemini API gets Managed Agents with hosted sandboxes and the Interactions API
Grok Build (xAI)
xAI launches Grok Build, an agentic CLI coding tool in beta
CoreWeave Sandboxes (CoreWeave)
CoreWeave Sandboxes launch in preview via the W&B SDK
Hermes CLI agent (Nous Research)
Hermes passes OpenClaw as #1 CLI agent on OpenRouter, adds computer use
Coding Agent Index (Artificial Analysis)
Artificial Analysis Coding Agent Index benchmarks model + harness combos
Link wallet for agents (Stripe)
Stripe launches Link wallet giving AI agents scoped payments
Devin for Terminal (Cognition Labs)
Cognition launches Devin for Terminal CLI coding agent
Cursor SDK (Cursor)
Cursor launches SDK exposing the runtime that powers the IDE
Projects.dev (Stripe)
Stripe opens Projects.dev: 32 infra providers provisionable by agents
DELEGATE-52 (Microsoft)
Microsoft's DELEGATE-52 exposes stealthy document corruption
Monthly roundup: ThursdAI Apr 2026 ↗
Monthly roundup
Gemini Enterprise Agent Platform (Google DeepMind)
Google launches Gemini Enterprise Agent Platform
Gemini Deep Research Max (Google DeepMind)
Google ships Gemini Deep Research + Deep Research Max on Gemini 3.1 Pro
Codex Computer Use + Chronicle (OpenAI)
Codex gets background computer use on macOS plus Chronicle screen memory
W&B LEET Workspace Mode (Weights & Biases)
W&B LEET TUI ships workspace mode with multi-run compare and GPU metrics
Euphony (OpenAI)
OpenAIDevs releases Euphony, an open-source Codex session log visualizer
Windsurf 2.0 (Windsurf)
Windsurf 2.0 ships Agent Command Center and full Devin integration
Warp any-CLI-agent support (Warp)
Warp now supports any CLI agent with vertical tabs and mobile control
Marimo Pair (Marimo)
Marimo Pair drops coding agents inside reactive Python notebooks
Cursor remote agents & code review agent (Cursor)
Cursor ships remote agents and a code review agent
Monthly roundup: ThursdAI Apr 2026 ↗
Monthly roundup
W&B Automations (Weights & Biases)
W&B Automations launch: event triggers from training runs
MemPalace (MemPalace (Ben Sigman & Milla Jovovich))
MemPalace open-source AI memory system goes viral with 26K stars
OpenClaw 2026.4.5 (OpenClaw)
OpenClaw 2026.4.5 ships /dreaming memory consolidation
Cursor 3 (Cursor)
Cursor 3 ships as agent-first rebuild, dropping the VS Code fork
Claw Chief (Ryan Carson)
Ryan Carson open-sources Claw Chief, an AI chief of staff
claw-code (Ultraworkers (Sigrid Jin & Bellman))
Claw-code clean-room rewrite becomes fastest repo to 100K GitHub stars
ARC-AGI-3 (ARC Prize Foundation)
ARC-AGI-3 launches: humans score 100%, frontier models under 1%
Manus My Computer (Manus (Meta))
Manus launches 'My Computer' desktop app for macOS and Windows
W&B iOS App (Weights & Biases)
Weights & Biases launches native iOS app for monitoring training runs
Google AI Studio (vibe coding overhaul) (Google)
Google AI Studio gets full-stack vibe coding with Antigravity and Firebase
Codex Subagents (OpenAI)
OpenAI ships subagents for Codex with custom TOML configs
Unsloth Studio (Unsloth AI)
Unsloth Studio: web UI for local fine-tuning with 2x speed, 70% less VRAM
Cursor in JetBrains (ACP) (Cursor)
Cursor joins ACP registry and goes live in JetBrains IDEs
AutoResearcher (Andrej Karpathy)
Karpathy open-sources AutoResearcher for autonomous ML experiments
Paperclip.ing (Paperclip)
Paperclip.ing: open-source agent orchestration for zero-human companies
Codex app for Windows (OpenAI)
OpenAI launches the Codex app on Windows
Google Workspace CLI (Google)
Google releases a Google Workspace CLI
Symphony (OpenAI)
OpenAI releases Symphony on GitHub
Wolf Bench (Weights & Biases)
Wolfram previews Wolf Bench, a multi-metric agent eval from W&B
Devin 2.2 (Cognition Labs)
Devin 2.2: computer use, browser, and self-verifying autonomous work
Nous Research Agent (Nous Research)
Nous Research ships a research agent
Perplexity Computer (Perplexity)
Perplexity introduces Perplexity Computer
Cloud Agents (Cursor)
Cursor launches cloud agents
LMLink (LM Studio)
LM Studio launches LMLink for remote access to local models
Time Horizon Benchmark (METR)
METR Time Horizon goes vertical: Opus 4.6 hits ~14.5-hour tasks
Dreamer (Dreamer)
Dreamer launches beta platform for building agentic apps with no-code AI
Code Factory (Ryan Carson)
Ryan Carson publishes the viral Code Factory agentic engineering blueprint
Deep Research (GPT-5.2) (OpenAI)
OpenAI upgrades Deep Research to GPT-5.2 with app integrations
WebMCP (Chrome 146) (Google (Chrome))
Chrome 146 introduces WebMCP, a native browser API for AI agents
AntFarm (Ryan Carson)
Ryan Carson releases AntFarm for agent coordination
Moltbook (Moltbook)
Moltbook: a Reddit built for and by AI agents
Codex App (OpenAI)
OpenAI launches standalone Codex app for managing parallel coding agents
OpenAI Frontier (OpenAI)
OpenAI Frontier: enterprise platform for AI agents as coworkers
Chrome Auto-Browse (Google)
Google launches agentic Auto-Browse in Chrome with Gemini 3
Gemini 3 Flash Agentic Vision (Google)
Google adds Agentic Vision to Gemini 3 Flash
Kimi Code (Moonshot AI)
Moonshot AI releases Kimi Code coding agent
Clawdbot (Peter Steinberger)
Clawdbot: open-source self-improving personal AI assistant for macOS
AI Prescription Renewals (Doctronic)
Doctronic launches first US pilot for AI prescription renewals
Ralph Wiggum (Ryan Carson)
Ralph Wiggum autonomous coding technique hits 1.2M views
Cursor 2 + Composer (Cursor)
Cursor 2 and the Composer model level up IDE agents
Monthly roundup: ThursdAI Dec 2025 ↗
Monthly roundup
Deep Research (OpenAI)
OpenAI Deep Research scores 26.6% on Humanity's Last Exam
Operator (OpenAI)
OpenAI Operator: first agentic ChatGPT with browser control
Code Maps (Windsurf)
Windsurf Code Maps generates flowcharts of entire codebases
Monthly roundup: ThursdAI Dec 2025 ↗
Monthly roundup
LLM Evaluation Jobs (Weights & Biases)
W&B launches LLM Evaluation Jobs for OpenAI-compatible APIs
GPT-5.1-Codex-Max in Cursor (Cursor)
Cursor offers GPT-5.1-Codex-Max for free through Dec 11
Serverless LoRA Inference (Weights & Biases)
W&B launches Serverless LoRA Inference on CoreWeave
Grok 4.1 Fast + Agent Tools API (xAI)
Grok 4.1 Fast: 2M context and Agent Tools API at 10x lower cost
Monthly roundup: ThursdAI Nov 2025 ↗
Monthly roundup
Antigravity (Google DeepMind)
Antigravity: Google's free agent-first IDE powered by Gemini 3 Pro
Marimo VS Code / Cursor extension (Marimo)
Marimo ships reactive Python notebooks extension for VS Code and Cursor
W&B LEET (Weights & Biases)
W&B ships LEET, an open-source terminal UI for monitoring ML runs
Terminal-Bench 2.0 (Laude Institute / Stanford)
Terminal-Bench 2.0 and Harbor launch as new bar for coding agents
Code Arena (LMArena (LMSYS))
LMArena launches Code Arena for live agentic coding evaluations
Cursor in-IDE browser (Cursor)
Cursor adds a built-in browser inside the IDE
Monthly roundup: ThursdAI Nov 2025 ↗
Monthly roundup
Codemaps (Windsurf (Cognition))
Windsurf ships Codemaps, AI-annotated navigable maps of your codebase
AWS-OpenAI infrastructure partnership (Amazon Web Services)
AWS announces multi-year strategic infrastructure partnership with OpenAI
Cursor 2.0 & Composer (Cursor)
Cursor 2.0 ships with Composer, its own 4x-faster coding model
Pomelli (Google (Labs))
Google Labs launches Pomelli, an AI marketing agent
Email Assistant (Perplexity)
Perplexity launches privacy-first Email Assistant for inbox management
Pokee (Pokee AI)
Pokee: an agentic workflow builder
Director 2.0 (Browserbase)
Browserbase launches Director 2.0 with 1Password delegated auth
ChatGPT Atlas (OpenAI)
OpenAI launches ChatGPT Atlas, its agentic AI browser
AI Studio Vibe Coding (Google DeepMind)
Google AI Studio launches 'Vibe Coding' build experience
TorchForge (Meta AI (PyTorch))
TorchForge: PyTorch-native library for scalable RL post-training
Amp Free (Amp)
Amp launches a free tier powered by ads and surplus model capacity
ChatGPT Memory (OpenAI)
OpenAI ships smarter ChatGPT memory management, no more 'memory full'