AIToday

Today's AI, for Today's work.

Start free60,000+ monthly readers

Updated: 21h ago

Latest AI News

Yesterday

Microsoft Agent Framework ports Python vector stores

Microsoft Agent Framework has ported Semantic Kernel's Python vector-store work forward, keeping typed records…

Semantic Kernel Blog21h ago

Claude Code review-check system catches only half

A developer catalogued past review complaints into 40 "pattern" types and had Claude Code check merge requests…

Zenn AI/ML21h ago

Orca review: Sonicmove engineer nohe finds parallel agent work seamless

nohe, an engineer at 株式会社ソニックムーブ, switched to Orca v1.4.224 about a month ago and found the move from a VS Cod…

Zenn AI/ML21h ago

Developer builds AFM-powered text adventure demo 'Night of the Blood Pond'

A developer released a demo text adventure called 'Night of the Blood Pond,' set in Oita Prefecture, that uses…

Zenn AI/ML21h ago

Claude Code 2.37337.1 update: gripes stay on X, bugs pile up on GitHub

After Anthropic's Claude Code desktop app update 2.37337.1, an AI-led search of X turned up user complaints ab…

Zenn AI/ML21h ago

Agent Script, out at Dreamforce 2025, builds Salesforce employee agents

A Zenn walkthrough shows how Agent Script, announced at Dreamforce 2025 and in public beta since November 2025…

Zenn AI/ML21h ago

Claude Code team-rollout guide, tested on 2.1.294/2.1.296

An unofficial guide for people rolling out Claude Code across a team reports hands-on checks of managed settin…

Zenn AI/ML21h ago

Claude Code MCP servers: just a script, no SDK needed

The author registered a memo server in Claude Code with "claude mcp add memo -- node /path/to/memo-mcp.mjs", t…

Zenn AI/ML21h ago

Bjorn Roche: AI has changed my engineering leadership role

Engineering leader Bjorn Roche published an essay titled 'AI has changed my engineering leadership role,' shar…

Hacker News1d ago

Google Play and Gemini to let users build AI apps with no developer account

Google announced on the 13th that Google Play and Gemini will get new features, including a way to create and…

Impress Watch1d ago

AUC 0.03 stopped a model review; four books fixed it

A modeler's improvement proposal was halted in internal review despite an offline AUC gain of 0.03, because he…

Qiita 機械学習1d ago

Claude's thinking-block 400 explained: preserved thinking, not a bug

A test on October 8 with the thinking-binding-controls-2026-08-01 header showed claude-opus-5-5 kept its think…

Zenn AI/ML1d ago

Agent 'done' claims checked against logs: VERIFIED 3/6(50%)

The build splits work into Task, Agent, ExecutionLogger and Verifier layers, and a 6-task demo run (2 fake=Tru…

Zenn AI/ML1d ago

Claude Code blamed sleep, skipped logs in 2 task outages

On Aug 29, 2026, a scheduled note task did not run; in its work log, Claude Code wrote that the cause was unve…

Zenn AI/ML1d ago

GitHub suspends developer after AI agents fire 50 PRs in a day

Running Claude Code subagents across 15 repositories produced about 50 pull requests in one day, with repeated…

Zenn AI/ML1d ago

ahic shows a simple RAG build on Gemini API and Python

ahic published a step-by-step guide to a simple RAG (retrieval-augmented generation) system that concatenates…

Zenn AI/ML1d ago

Gemini 4 Argon's 32,768-token cap trips up a three-AI test

On 12 programming problems run 120 times, Claude Opus 5.5 scored 36 of 36, Gemini 4 Argon 35 of 36, GPT-5.6 So…

Zenn AI/ML1d ago

Seedance 2.5 run logs show how to avoid paying twice

A Higgsfield API estimate returned a price formula, not a price, for Seedance 2.5, so the author computed it a…

Zenn AI/ML1d ago

Folkbench: judge AI relay APIs on 4 coding tests

Folkbench published a guide arguing AI relay APIs (gateways that sit between coding tools and model providers)…

Zenn AI/ML1d ago

Rule Branch links GitHub rules to AI files via Draft PR

Rule Branch connects to GitHub and extracts rule candidates from a repository's README, existing AI instructio…

Zenn AI/ML1d ago

Popcar: AI-powered decompilation is a game-preservation boom — and a risk to all software

Indie developer Popcar describes a recent wave of AI-powered decompilation, enabling mashups between existing…

Simon Willison's Weblog1d ago

Gant Laborde's TensorFlow.js intro book works without Python, reviewer says

A reviewer assessed Gant Laborde's 'Learning TensorFlow.js,' translated by Yasushi Ando and published by O'Rei…

Qiita 機械学習1d ago

Epic Games Launcher: rebuilt "way faster," vibecoded with AI

Epic Games launched a preview of a rebuilt Epic Games Launcher for Windows, Mac, and Linux, saying it is "way…

The Verge AI1d ago

Turn today's AI news into today's work.

The day's essentials from 100+ sources, delivered to Email, LINE, or Slack. Free to start.

60,000+ monthly readers
Get Started Free

Earlier

Claude Code harness: start with CLAUDE.md and Skills only

Miyatchi of AI Orchestra, who teaches Claude Code to a course of over 200 students, laid out four harness mech…

Zenn AI/ML1d ago

Claude Code's Projects beta reviewed: Claude as coordinator, less visibility

A developer tested the beta Projects feature in Claude Code on his Muntin project, a small web framework writt…

Zenn AI/ML1d ago

TerminalHub lines up 30 Claude Code sessions on Windows

The author, an acquaintance of TerminalHub's developer, describes a Windows app that lists Claude Code, Codex…

Zenn AI/ML1d ago

Claude Code auto-resumes tasks after usage limit resets

Claude Code (v2.1.234 and later) now waits through a usage limit and restarts the task when the limit resets…

Zenn AI/ML1d ago

Claude Code + Rojo setup syncs AI code straight into Roblox Studio

The guide sets up VS Code, Rojo and Claude Code so that a Japanese instruction like "make a server script that…

Zenn AI/ML1d ago

SCHEMA wins in Neo4j SimpleKGPipeline test

Maru ran Neo4j GraphRAG for Python's SimpleKGPipeline on the same cookbook.md, which holds curry rice and niku…

Zenn AI/ML1d ago

AICraft opens one OpenAI-compatible key to 100+ models

AICraft launched an OpenAI-compatible API gateway at https://aicraftapi.net/v1 that reaches 100+ models under…

Zenn AI/ML1d ago

Bulchimbeon keeps Macs awake for Claude Code, but fails 3 real tests

A developer released Bulchimbeon, an MIT-licensed macOS menu-bar app that keeps the Mac awake only while agent…

Zenn AI/ML1d ago

Anthropic builds Git replacement Scion, already migrated

Anthropic announced Scion, a version-control system replacing Git, plus a code-review system, Antique

GIGAZINE AI2d ago

Python tutorial builds sentiment-based investing strategy with yfinance and scikit-learn

A Qiita tutorial shows how to pull stock data with yfinance, clean news text with nltk, classify sentiment wit…

Qiita 機械学習2d ago

Anthropic replaces Git with Scion as merges hit 30,000 per day

Anthropic built Scion, its own source control system to replace Git, completed the migration in five months wi…

Simon Willison's Weblog2d ago

Gemini 4 Argon hits llm-gemini at $2/million input, doubling Dec 1

Simon Willison added Gemini 4 Argon support to llm-gemini, his Gemini API tool

Simon Willison's Weblog2d ago

Claude Code false pass: `pytest tests; echo "EXIT=$?"` returns 0

The developer reproduced false passes locally with Python 3.14 and pytest 9.1 on Windows Git Bash: `pytest tes…

Zenn AI/ML2d ago

GitHub ships Stacked pull requests to tame AI-sized code changes

GitHub released Stacked pull requests in public preview on July 30, 2026 and made them generally available on…

Zenn AI/ML2d ago

AI efficiency boom leaves team leader more work, not less

A team leader says AI-driven development has not reduced his workload

Zenn AI/ML2d ago

Claude Code rollout guidebook for small-business IT staff

A guidebook for small-business IT managers and part-time engineers lays out Claude Code adoption in order, fro…

Zenn AI/ML2d ago

Wells Fargo puts AI to work in banking, assets top $2.7 trillion

Wells Fargo CEO Charlie Scharf said the bank uses AI and autonomous agents to write code faster, close previou…

Top Companies AI2d ago

Google ships Gemini 4 Argon worldwide at $2 input

Google made Gemini 4 Argon available worldwide today through its API, the Gemini app, and Google Antigravity

THE DECODER2d agoAlso reported by AI Watch (Impress) and 5 others

Pluto Security raises $42M as AI coding spreads

Pluto Security left stealth with $42 million — an $8.5 million seed led by TLV Partners and a $34 million Seri…

TechCrunch AI2d ago

Google Play opens private app launches via Google AI Studio in U.S.

Google said Android owners can build an app with Google AI Studio and launch it privately through Google Play…

TechCrunch AI2d ago

Google puts AI Studio apps straight into the Play Store

On Tuesday, Google announced a US feature letting people launch personal Android apps vibe-coded in its AI Stu…

The Verge AI2d ago

Epic Games mandates AI tools on Disney game

Epic Games is directing developers on an unannounced Disney shooter to use AI tools like Anthropic's Claude, w…

The Verge AI2d ago

X Subscription bundles Grok and Cursor from $8 to $200 per month

The new X Subscription combines X's paid features with AI tools like Grok and Cursor, with plans ranging from…

TechCrunch AI2d ago

Developer builds minutes app on Gemini API and Google Apps Script

The app records audio in the browser, stores it in Cloud Storage, and has Gemini API's gemini-2.5-flash transc…

Zenn AI/ML2d ago

Platforms for AI agents: CLI, YAML turn intent into execution

The author describes rebuilding an existing frontend delivery platform so AI agents can use it, giving them th…

Zenn AI/ML2d ago

Mitate turns plain-language requests into designed systems

In a live Gemini test on 2026 年 10 月 10 日, R2-san asked only to hear club members' event dates and send remind…

Zenn AI/ML2d ago

Shipaton app "How We Work" misses ship after 2-week review stall

The builder made the iPhone app "How We Work" for RevenueCat's Shipaton with no code, Markdown or Figma edits…

Zenn AI/ML2d ago

Peter Mattis: AI put me back on the side of writing code

In episode #36 of The Pragmatic Engineer, Cockroach Labs CTO Peter Mattis said AI has returned him to writing…

Zenn AI/ML2d ago

CLAUDE_CONFIG_DIR=~/.claude-work still leaks ~/.claude/CLAUDE.md, kills AGENTS.md

With three rule files in place, launching CLAUDE_CONFIG_DIR=~/.claude-work in ~/repo pulled in the personal ~/…

Zenn AI/ML2d ago

One-CEO firm pins all nine Claude Code jobs to --effort low via claude-run.sh

Joinclass LLC (合同会社ジョインクラス) routes all nine unattended Claude Code jobs to --effort low through its claude-run…

Zenn AI/ML3d ago

River Review: AI code review via Review Judgment as Code

The project's developer laid out River Review's core design, placing review criteria in repository-owned Skill…

Zenn AI/ML3d ago

カンパニーブレイン converges on Markdown bodies with YAML labels

The author argues company knowledge for AI agents is settling on Markdown bodies with YAML labels, noting Anth…

Zenn AI/ML3d ago

ColGrep-Agent-2B cuts Claude Code tokens 34% with a small local search agent

ColGrep-Agent-2B delegates repository search to a local 2-billion-parameter model, cutting API tokens 34% and…

Zenn AI/ML3d ago

HACH developer: persist AI agent task steps, not cursors

The developer of HACH, a resident AI agent for Slack, Teams and Chatwork, published a design guide arguing lon…

Zenn AI/ML3d ago

yagura v0.1.0 lets Claude Code and Codex CLI work side by side

yagura v0.1.0 was released as a desktop app for Windows, macOS and Linux that runs Claude Code, Codex CLI, Ant…

Zenn AI/ML3d ago

Claude Code 2.1.286〜2.1.288 hidden-failure traps: three fix points

A Claude Code Mod developer traced silent, error-free failures on the desktop Code tab to three causes and con…

Zenn AI/ML3d ago

AGENTS.md and CLAUDE.md: one folder, two AI tools, one rulebook

Tiero keeps all instructions in AGENTS.md and makes CLAUDE.md a single line that imports it, so Codex and Clau…

Zenn AI/ML3d ago

AIGoCode offers one API key for OpenAI, Anthropic, Gemini models

AIGoCode is an AI API gateway that bundles access to OpenAI, Anthropic and Gemini-compatible models under one…

Zenn AI/ML3d ago

Epic Games delays Marvel and Star Wars shooter, tells devs "No more hand-written code"

Epic Games delayed its Disney collaboration — a shooter featuring Marvel and Star Wars characters — that had b…

Top Companies AI3d agoAlso reported by Ars Technica AI

Anthropic's dynamic workflows let Claude run its own agents

Anthropic announced dynamic workflows, a new multi-agent feature in Claude Console where the lead agent writes…

Latent Space3d ago

GitHub Copilot app for Beginners: 4 ways to teach Copilot your codebase

GitHub published the latest installment of its GitHub Copilot app for Beginners series, outlining four ways to…

GitHub Copilot Blog3d ago

Build your own AI agent: three routes, but ChatGPT GPTs end December 11, 2026

AI Now laid out three do-it-yourself routes — chat-type skills in Gemini or Claude, automation-type flows in n…

AINOW3d ago

AWS shows llama.cpp deployment on Amazon SageMaker AI

AWS published a technical how-to for deploying decision models with llama.cpp on Amazon SageMaker AI, serving…

Amazon AI Blog3d ago

Claude Code deny rules leak; PreToolUse hook plugs gaps

The developer's settings.json deny list blocked commands starting with the literal text like git push, but off…

Zenn AI/ML3d ago

AI instruction sets should mirror a company org chart, essay argues

An essay proposes mapping custom assets inside .github so copilot-instructions.md and instructions set policy…

Zenn AI/ML3d ago

subagent-driven-development adds progress.md ledger to stop rework

The subagent-driven-development skill writes a progress.md ledger under .superpowers/sdd/ for each plan, loggi…

Zenn AI/ML3d ago

Mitate Hits 2,990 Commits in 6 Weeks on Spare Time

R2-san's Mitate, a project for the 5th Agentic AI Hackathon with Google Cloud, reached about 2,990 commits and…

Zenn AI/ML3d ago

Claude Max, Team plans add $100–$500 monthly API credits unusable in Claude Code

Since 2026-10-07, Max 5x, Max 20x and Team plans include monthly API credits — $100, $200 and up to $500 per s…

Zenn AI/ML3d ago

Zenn: reply to '20 Claude Code Max vs your salary' by conceding the math

A Zenn piece lays out a scripted reply for engineers facing the interview question 'for your 8 million yen sal…

Zenn AI/ML3d ago

Today's AI, for Today's work.

Get curated AI news from 100+ sources delivered daily to your inbox. Free to use.

60,000+ monthly readers

Get Started Free

Free · 30 seconds with Google · unsubscribe anytime