What shipped & dropped across AI labs, today.
Today
Thu · Jul 16 · 6 postsAutomations in Grok
Describe a job once and Grok runs it on a schedule or when an email arrives, then reports back.
Evolving Spec-Driven Development: Conductor Now Supports Antigravity
Conductor has evolved from a Gemini CLI extension into a portable plugin, bringing conversational Spec-Driven Development (SDD) to ecosystems like Antigravity CLI and Claude. Rather than relying on strict command sequences, developers can now chat naturally with their AI assistant while it dynamically manages persistent markdown artifacts (like spec.md and plan.md) in the background. This update eliminates workflow friction while ensuring your repository remains a version-controlled, single source of truth for your project's architecture and state across different AI tools.
How Anthropic runs large-scale code migrations with Claude Code
A step-by-step guide to running large code migrations with AI agents — including Bun's million-line Zig-to-Rust port.
Building scalable AI agents with modular prompt transpilation
To resolve the scaling bottlenecks and runtime errors caused by monolithic system prompts, engineering teams should treat prompts as build artifacts by modularizing instructions into reusable templates. By running these modular "skill files" through a transpiler, developers can enforce static validation, catch missing dependencies at build time, and integrate prompt generation directly into their CI/CD pipelines. This deterministic approach prevents code drift and ultimately establishes a safe framework where agents can propose updates to their own logic via standard pull requests.
Working with Claude Fable 5 in Claude Cowork
Anthropic's Claude Fable 5 can carry long, complex work on its own. Here's how to get the most out of it in Claude Cowork.
Expanding Choice in Gemini Enterprise Agent Platform: Introducing Grounding with Parallel Web Search
Google Cloud has partnered with Parallel Web Systems to natively integrate Parallel's search infrastructure as a web grounding provider on the Gemini Enterprise Agent Platform. This integration enables developers to anchor their AI agents in verifiable, real-time web results, significantly improving factual accuracy for complex enterprise workflows. Additionally, the partnership offers expanded architectural flexibility, allowing users to programmatically extract, permanently cache, and process web data alongside other large language models.
Yesterday
Wed · Jul 15 · 2 postsGrok Build is Now Open Source
Explore the harness behind our coding agent and TUI.
Working at the frontier: Why Base44 trusts Claude Fable 5 with their most challenging engineering work
Why Base44 trusts Anthropic's Claude Fable 5 with its most complex product and engineering tasks.
Tuesday
Tue · Jul 14 · 5 posts
GitHub Copilot for JetBrains expands BYOK capabilities
This update brings major advances in customization and model provider flexibility to all tiers of GitHub Copilot for JetBrains IDEs. With richer plugin and provider experiences, improved conversational interactions, and…
GitHub Copilot in Visual Studio — June update
June 2026 is about visibility and trust with a clearer view of your GitHub Copilot usage, a new trust layer for MCP servers, and the first C++ scenarios for the…
Introducing Claude for Teachers
Anthropic commits $10 million to Canadian AI research
Security reviews now available in the GitHub Copilot app
You can now run a security review on your in-flight code changes directly from the GitHub Copilot app. The /security-review slash command is shipping in public preview, bringing the same…
Monday
Mon · Jul 13 · 2 postsUnlocking the Next Era of On-Device AI with Google Tensor and Pixel
At Google I/O Connect India, Google showcased the future of 100% private, on-device AI powered by the custom Tensor SoC and TPU for the new Pixel 10 family. The event debuted the lightweight Gemma 4 E2B model, which runs natively on the device to enable completely offline multimodal features like AI chat, real-time image recognition, and personal agent tasks. Developers can start building these secure, edge-based applications today by accessing the newly announced Tensor SDK beta and its accompanying open-source resources.
Working at the frontier: How Hebbia builds AI for financial diligence that can't miss a detail
How Anthropic's Claude Fable 5 beat Hebbia's finance-specific model evaluations, achieving their biggest accuracy gain yet.
Friday
Fri · Jul 10 · 3 postsAgentic autofix for code scanning alerts in public preview
Fix code scanning alerts faster with agentic autofix.
Side Chats and Conversation Search
This release makes it easier to stay in flow with side chats that run alongside your main chat, the ability to search agent transcripts, and simplified project and repo pickers. #Side chats Open a side chat to ask questions, explore ideas, and investigate tangents without interrupting your main agent conversation. Use /side, /btw, or the plus button at the top of the chat panel to create a new side chat that has context from the main chat. Each side chat is a durable, full agent conversation that you can follow up on, revisit later, and at-mention to pull context back into the main thread. By default, side chats focus on reading, searching, and answering. Use them to ask clarification questions, research alternatives without committing to a pivot, and sanity-check a decision while the main agent continues running. #Conversation search Find past agent chats faster with search results that go beyond names and PR numbers. In the Agents Window, you can search agent transcripts from the command palette (Cmd+K). Cursor builds a local search index that scales search to thousands of conversations with snappy performance. You can also search within an existing conversation using Cmd+F. Jump between matches, see a match counter, and keep searching as you scroll through long transcripts. #Redesigned project and repo pickers We've simplified the project and repo pickers and made them more powerful. You can now stay in the picker for workflows that used to send you elsewhere. For example, you can create a project and connect GitHub, GitLab, or Azure DevOps without leaving the picker. Search is now scoped to where you're working—This Computer, Cloud, or a specific remote machine—instead of one global search box. You can also remove projects from Recents with one click. #New cloud agent hooks Cloud agents already support team hooks around tool execution and file/shell work. We've added new hooks that let you observe and control the agent conversation itself: prompts, responses, thinking, subagents, compaction, and turn completion. See all the supported hooks in our docs. New hooks like beforeSubmitPrompt, afterAgentResponse, afterAgentThought, stop, subagentStart, and more allow you to better observe output and reasoning, control subagents, and build self-correcting loops with cloud agents. Picker Improvements↓↑ Improved repo picker grouping with options under No Repo, On This Computer, and Cloud. Added the "Run on" picker that shows where your agent can run (Cloud, This Computer, or Remote Machines) and drills into the relevant choices (environments, local options, and so on) from there. The branch picker opens on your default branch and recently used branches instead of a long flat list, with search for everything else. Removed the Home concept so working without a repo is an explicit No Repo choice. Combined all remote options—your machines, team pools, and existing remote workspaces—into one searchable Remote Machines menu. Multi-repo and multi-root selection is a Select Multiple toggle inside the Cloud and This Computer flyouts, replacing the separate Set Up Workspace builder. Moved search to the top of the menu. Simplified the footer. Added the ability to find the No Repo option by typing none or no repo. If a repo exists only in the cloud, the picker suggests cloning it locally. Added clearer section dividers. Added folder icons with a small cloud badge for cloud repos. Removed redundant Current tag in the branch list.
Working at the frontier: How Cognition trusts Claude Fable 5 to work through the night
Cognition tested Claude Fable 5 in Devin, its AI software engineer. It's the first model its team trusts to run unattended for eight hours and deliver production-ready code.
July 9
Thu · Jul 9 · 2 postsCase StudyUST is bringing Claude to physical AI
LiteRT.js, Google's high performance Web AI Inference
We're excited to introduce LiteRT.js, the newest member of the LiteRT family! LiteRT.js is our powerful solution for running machine learning models directly in the browser, extending Google's cross-platform edge AI runtime to the web. Built for JavaScript developers, LiteRT.js delivers state-of-the-art ML model inference performance on WebGPU and upcoming WebNN, with a fallback to WebAssembly for CPU. This post provides a quick tour of LiteRT.js and gives web developers everything they need to get started.