Why it matters: Marks a major brokerage embedding frontier AI directly into advisor workflows.
Charles Schwab is bringing Anthropic's Claude for Financial Advisors into its Advisor Center, saying it is the first RIA custodian to offer the advisor-specific AI to independent firms.
Why it matters: Ads inside AI Mode expand how commerce media works as search shifts to AI answers.
Google is scaling ads in AI Mode to 10 additional countries including Australia, Canada, India, Kenya, Malaysia, New Zealand and Nigeria, and launching multiturn ads that appear in follow-up queries in the US.
Articul8 launched Bivia, a product bringing domain-specific decision intelligence to enterprise AI agents, aimed at grounding agent decisions in industry-specific data and logic.
Why it matters: Google's earnings show AI capex pressure colliding with slowing ad growth.
Alphabet shares tumbled 9% after a $51B AI outlay deepened cash burn, with revenue and profit missing estimates as search undershot. Parent revenue still jumped 25% as homegrown TPU chips take on Nvidia.
Sierra's Compliance Console adds configurable guardrails to its Horizon platform so AI agents follow rules on outreach across channels, states and countries.
Why it matters: Microsoft's cloud beat is a key signal on whether AI capex is translating into revenue.
Microsoft reported better-than-expected cloud revenue, suggesting its massive AI investment is paying off. Shares climbed 5%; revenue growth accelerated slightly but data centers dampened cash flows.
Deep Hat Sift is a model that decides which incoming alerts or logs an AI agent should investigate, avoiding costly uneventful runs at enterprise volumes.
Insurify launched an AI Agent Gateway that lets AI assistants shop for insurance while giving carriers consumer-choice protections and controls over how their products are presented.
Cloudera introduced preconfigured AI applications built with Nvidia technology and expanded partnerships with Neo4j and Mistral AI to help enterprises move AI into production.
Why it matters: Meta's giant raise and dual Nvidia/AMD commitments shape the AI compute landscape.
Nvidia and Meta partner on personal AI agents, while AMD wins a major Meta deal; Meta is raising $97.5 billion with Nvidia and AMD buying in.
Why it matters: Regulatory scrutiny of the deal could reshape control of the open-model ecosystem.
The US Justice Department is making inquiries about Nvidia's roughly $13 billion agreement to acquire Hugging Face, part of broader antitrust scrutiny of the chip giant.
Particle-1 is a post-trained DeepSeek V4.1 Flash available via API; Particle says it used fewer tokens across four software and data-engineering benchmarks.
Why it matters: Agent orchestration across projects and models pushes AI coding tools toward cheaper, multi-model workflows.
GitHub's Copilot Fusion promises frontier-level AI coding at half the cost and turns Copilot into an agent orchestrator that works across projects and models.
Why it matters: Gives enterprises spend visibility that could accelerate Claude adoption.
Anthropic launched Claude Enterprise features including Smart Reports, Surveys and pooled budgets to show companies what their AI spending buys, ahead of its planned IPO.
Google launched a Multi-Party Compute custody platform letting banks hold and transfer digital and tokenized assets more securely, pitched as a defense against AI and quantum threats.
Why it matters: Banks pushing back on AI agent access could slow agentic commerce adoption.
Big banks are blocking AI agents from accessing customer accounts and warning against sharing banking passwords, as agentic payments create liability and security concerns. Banks want safer alternatives like tokenized access.
Why it matters: A new frontier model topping harder benchmarks and aimed at cyber defense marks a major release.
Google unveiled Gemini 4 Carbon, retaking the top spot on harder benchmarks, with cyber defense benchmarks and restricted access initially for Fairwind cyber defenders at $2/$10 per 1M tokens.
MCode Stack is an open-source collection of 15 agent skills that MiniMax's own coding team uses daily, giving developers reusable capabilities for coding agents.
TaxDome released a public API and an MCP server letting AI agents and developers connect to its tax and accounting practice management platform.
Microsoft's Copilot Harness, a set of building blocks for running Copilot-style agents, is now offered via Azure AI Foundry so developers can access it directly.
Webull and Sentinel AI are piloting a program that surfaces public company information to retail investors at the point of trade, connecting issuers with brokerage users.
Why it matters: Opening its IDE to rival agents makes Android Studio a neutral platform in the coding-agent race.
Google's Bring Your Own Agent feature lets Android Studio run over 40 coding agents, including Anthropic's Claude Agent and OpenAI's Codex, plus terminal tools Google says finish UI tasks 77% faster.
Perplexity's Mac app voice mode now runs on OpenAI's gpt-live-1, which can listen while it speaks, replacing its previous voice stack. The upgrade is available in the Mac app today.
Unstoppable Domains launched Unstoppable.ai, a model-agnostic AI workspace that lets users work with multiple AI models in one place.
Whistle is an offline speech-to-text project that runs entirely on an ESP32-S3 microcontroller, enabling local voice transcription without a cloud connection.
Boston- and Geneva-based Read/Write Labs rebranded Dandelion and says it is developing software to turn brain models into therapies; public materials show no product demo, customer list or clinical validation yet.
A Stirling-based startup has launched a privacy-first AI tool aimed at helping businesses use AI while keeping sensitive data protected and on-premises.
Slack's Code API, which opens the code-channel system launched in August 2026 to developers' agents, is now generally available on all Slack plans after over 150 customers and partners requested access.
Why it matters: Anthropic is pushing Claude deeper into scientific research and drug discovery partnerships.
Anthropic expanded its AI for Science program to foundations and research institutions and outlined six drug-development partnerships plus in-house Mithril work.
HITRUST updated its certification platform with new AI security assessment tools and third-party risk management features for organizations handling sensitive data.
Why it matters: The SDK turns Cursor from an IDE into a platform for teams shipping their own coding agents.
Cursor's Agent SDK lets software teams define agents in code, test them with evals and event replay, and deploy on Cursor's infrastructure or their own. Released October 20.
Why it matters: Native code scanning for AI apps gives developers a first-party way to catch AI-specific vulnerabilities.
GitHub made code scanning for AI applications available in public preview, aiming to find security issues in AI app code.
OpenMausBot, published on GitHub by Milind Soni, is an open-source personal agent chat app where each bot is a local Claude or Codex agent with its own personality, model, cloud computer and connected apps, used like messaging contacts.
Anthropic added ready-to-use Legal Artifacts, interactive tools in a gallery for legal users, expanding its push into the legal vertical for Claude customers.
Fibr AI launched AXO, a tool that adapts existing website content for AI discovery and lets agents perform supported tasks. Results so far come from Fibr's own site and are unverified.
Voice AI startup Rime launched Prism, a fine-tunable speech-to-speech model aimed at enterprises that want to avoid black-box voice AI.
Zocks, a privacy-first AI platform for financial services, launched Zocks Account Opening, which automates the onboarding workflow to speed client setup and add capacity.
Mistral opened its Responses API in public preview, borrowing OpenAI's format to help developers build AI agents.
esp32-pie-asr is an open-source project that runs live English speech recognition entirely on an ESP32-P4 microcontroller, with no cloud service or separate AI accelerator, displaying words as they are spoken.
Otter added a Work mode to its AI Chat assistant that automates more tasks before and after meetings, part of its push beyond transcription following a $320 million funding round last week.
A US company launched modular, Lego-like building kits for constructing offices, security checkpoints and robotics or drone facilities.
Amazon announced smart home updates including an Alexa-powered interactive 3D map of a home's interior and exterior, among its biggest announcements of the day.
Maxinsights raised $30M in seed funding for a platform that pays people to record everyday activities like cooking and cleaning, generating training data for robots and AI models. Founders are robotics and autonomous-driving veterans.
Altera announced a new Agilex FPGA portfolio plus an agentic AI developer tool aimed at helping developers build and deploy AI designs on its FPGAs.
Amazon's new Alexa Photo Frame lets users ask the voice assistant to display favorite photos. ZDNET reviews why the device fills a gap for the reviewer.
Why it matters: 512GB of unified memory makes the Mac Studio a serious local-model workstation.
Apple's Mac Studio with M5 Ultra is now available with 512GB of unified memory starting at $22,299, a configuration local LLM users have awaited.
AIMOGA Robotics, an OMODA and JAECOO ecosystem partner, unveiled its first home companion robot, moving automotive perception and control tech into homes.
Why it matters: Adds another Android XR entrant to the AI glasses race against Meta and Apple.
XREAL's Aura smart glasses run Android XR with Google Gemini built in, bringing AI assistance to a wearable display. Pricing is high, the report says.
PsiQuantum developed a device that routes signals between GPUs using optical links, which it says could exponentially speed up AI data center processing, per Semafor.
Samsung is partnering with Lark Health to offer eligible US seniors AI diabetes prevention coaching through Samsung Health, with Medicare eligibility restrictions.
Pakistan's Peshawar Safe City project has linked drone and anti-drone technology into its central command for city-wide surveillance and security, the first such integration in the country.
OrcaRouter's Orca Connect turns any local GPU into a free, globally accessible AI API, the company says.
About Amazon · 10h · 3 sources
#
Why it matters: Amazon's Alexa+ push into camera-equipped home devices expands ambient AI but raises privacy concerns.
Amazon refreshed its Echo lineup: Echo Dot gains a built-in smart home hub and Wi-Fi extender, Echo Studio adds a camera for Alexa+ visual help, and Echo Show 16 offers a larger display. Privacy questions remain about the Studio camera.
Samsung unveiled the Galaxy A18 and A08 budget phones with AI features and six years of security updates; the A08 debuts in India with a 6,000mAh battery.
Why it matters: Vera CPU deployment scales Oracle's AI cloud and signals Nvidia's broader hardware push.
Nvidia and Oracle launched standalone Vera CPU instances in a multi-gigawatt OCI deployment, with Cerebras gaining 10% on an Oracle Cloud deal. Nvidia also admitted Vera Rubin is less powerful than initially claimed.
Nvidia GeForce RTX 5080 cards are now selling for close to $2,000 at US retailers, about $800 higher than two months ago, with some Newegg inventory running dry.
XPeng's IRON humanoid robot moved its solid-state battery into small-batch production, with scale manufacturing targeted by year-end.
Huawei unveiled the FreeClip 2 open-ear earbuds in three colors at its Double Eleven event in China, alongside the Pura 90 Plus and Enjoy 100 phones, with louder drivers and NPU AI.
EverXCloud has opened pre-launch access to a US spot exchange for renting cloud GPUs by the hour.
Realbotix relaunched its Ask Aria AI avatar, showcasing the Echo robot platform. Aria is an interactive AI companion avatar users can talk with.
Why it matters: Chip smuggling seizures show export controls are being tested as AI hardware demand surges.
Malaysian customs seized advanced AI chips worth $12.1 million concealed inside 30 servers at an airport, suspected of violating export controls. The haul underscores efforts to curb chip smuggling.
oToBrite showcases a 3D stereo vision camera at ROBOTWORLD 2026 and teams with YUAN on a GMSL2-to-HSB demo for robotics vision.
Why it matters: Puts a robot foundation model with task planning into a mainstream cloud for developers.
Google brought Gemini Robotics ER 2 to Google Cloud in public preview, adding task planning capabilities for robots.
Why it matters: RTX Spark updates push local AI agent capabilities onto consumer and enterprise laptops.
Nvidia says RTX Spark Windows laptops gain new AI models, agent tools and games, as on-premises AI computing demand rises alongside cloud costs. Asus and Intel also unveiled the commercial ExpertBook A7 laptop in Taiwan.
Rokid launched AI glasses with a full-color display, weighing 55g and priced at 4,999 yuan. CEO Zhu Mingming called it the industry's 'iPhone 1 moment.'
Meta's developer MCP server gains six tools that let AI agents create and manipulate apps, extending agent capabilities on the Meta developer platform.
Version 3.1.0 of the dictation tool, which uses LLM reformatting, is available across macOS, Windows and Linux.
Articul8 launched Bivia, a product bringing domain-specific decision intelligence to enterprise AI agents, aimed at grounding agent decisions in industry-specific data and logic.
Amp, the frontier coding agent and development environment, ships a new CLI build version 0.0.1793217669-gb7b7f6, installable via npm with npm i -g @ampcode/cli.
LiteRT in Google Play services now offers GPU inference via ML Drift, about 1.4x faster, with a new CompiledModel API. Switching from bundled binaries cuts app size by roughly 8 MB and enables automatic runtime updates.
Antigravity 2.28.0 is an agent orchestration platform for macOS, Windows and Linux, updated with version 2.28.0.
NamiGPT is a beginner-friendly GitHub tutorial that walks through writing GPT-2 from scratch in PyTorch line by line to show how large language models work.
Zed, a multiplayer code editor with built-in AI agent support, ships version 1.26.2 for macOS, Windows and Linux.
An AWS blog walks through architecture and an open-source reference blueprint for a governed, multi-account agentic AI platform on Amazon Bedrock AgentCore, covering control plane, gateways and CI/CD evaluation.
Particle-1 is a post-trained DeepSeek V4.1 Flash available via API; Particle says it used fewer tokens across four software and data-engineering benchmarks.
Why it matters: Agent orchestration across projects and models pushes AI coding tools toward cheaper, multi-model workflows.
GitHub's Copilot Fusion promises frontier-level AI coding at half the cost and turns Copilot into an agent orchestrator that works across projects and models.
An AWS blog walks through fine-tuning Qwen2.5-1.5B with PyTorch FSDP across eight GPUs on two EFA-enabled g6.12xlarge instances on Amazon EKS using the Ray Train Deep Learning Container.
Part 3 of an AWS series on QA Studio with Amazon Nova Act adds apps that organize tests and report health metrics, plus team-based user management controlling test ownership.
A Google Developers post walks through the Agents CLI evaluation loop: continuous testing and grading to catch prompt regressions, compare models via side-by-side trace comparisons, and automate evaluation passes.
Reindeer joined Microsoft's Pegasus program and listed its managed, single-tenant platform that connects agents to systems like SAP and Oracle on Azure Marketplace, with eligible purchases counting toward Azure commitments.
AgentFlux is a platform for building visual workflows and autonomous AI agents, runnable with Docker on your own computer or server. It drew 101 GitHub stars in its first week.
PICO integrated its XR SDK with ten AI coding assistants, so developers can install it with a single npm line.
MCode Stack is an open-source collection of 15 agent skills that MiniMax's own coding team uses daily, giving developers reusable capabilities for coding agents.
TaxDome released a public API and an MCP server letting AI agents and developers connect to its tax and accounting practice management platform.
Microsoft's Copilot Harness, a set of building blocks for running Copilot-style agents, is now offered via Azure AI Foundry so developers can access it directly.
Fibo 2.0 Edit Turbo is an instruction-based image editing model that edits images in four guidance-free steps, using up to four reference images and an optional mask; a Hugging Face Space demo is available.
Why it matters: Opening its IDE to rival agents makes Android Studio a neutral platform in the coding-agent race.
Google's Bring Your Own Agent feature lets Android Studio run over 40 coding agents, including Anthropic's Claude Agent and OpenAI's Codex, plus terminal tools Google says finish UI tasks 77% faster.
Unstoppable Domains launched Unstoppable.ai, a model-agnostic AI workspace that lets users work with multiple AI models in one place.
This Hugging Face Space runs Wan 2.2 image-to-video generation in a browser demo, turning a still image into a short clip. It has about 100 likes and is a community-built Gradio app.
Controlled GitHub tests found Claude Code subscriptions burn five-hour quotas at different rates depending on client labels, including a roughly 2.9x gap on one Team plan.
Why it matters: The SDK turns Cursor from an IDE into a platform for teams shipping their own coding agents.
Cursor's Agent SDK lets software teams define agents in code, test them with evals and event replay, and deploy on Cursor's infrastructure or their own. Released October 20.
Raycast, a keyboard-driven launcher for Mac, Windows and Linux that lets you control tools with a few keystrokes, shipped versions 2.10.0.0 and 2.10.1.0.
Why it matters: Native code scanning for AI apps gives developers a first-party way to catch AI-specific vulnerabilities.
GitHub made code scanning for AI applications available in public preview, aiming to find security issues in AI app code.
OpenMausBot, published on GitHub by Milind Soni, is an open-source personal agent chat app where each bot is a local Claude or Codex agent with its own personality, model, cloud computer and connected apps, used like messaging contacts.
Fibr AI launched AXO, a tool that adapts existing website content for AI discovery and lets agents perform supported tasks. Results so far come from Fibr's own site and are unverified.
Fibo 2.0 Edit Turbo is an instruction-based image editing model that edits images in four guidance-free steps, using up to four reference images and an optional mask; a Hugging Face Space demo is available.
This Hugging Face Space runs Wan 2.2 image-to-video generation in a browser demo, turning a still image into a short clip. It has about 100 likes and is a community-built Gradio app.
A merged image/video model built by blending two Minimax-H3 Ref2VA finetunes with a custom PRISM algorithm, no mergekit. Weights shared in fp16 on Hugging Face.
Tencent Hunyuan's Hy-SimulMT-30B-A3B-FP8 is a low-latency, high-quality model for real-time simultaneous translation, released under Apache 2.0 on Hugging Face alongside the Hy-Nomas model.
MiniCPM-OCR is an open document-reading model that also checks whether its own OCR output is trustworthy. OpenBMB released the 1.3B MiniCPM-OCR 1.0, scoring 93.68 on Real5-OmniDocBench for curved and real-world pages, plus an AI judge model.
Liquid AI released LFM3, a family of six open-weight models in the 2B class for building specialized multimodal models.
VDEBench is a Hugging Face dataset of 3,506 examples testing instruction-guided image editing on dense Chinese and English document pages, where models must change requested text or tables while preserving layout and typography.
Unlimited OCR 2 is an open document parser that uses sparse reference sliding-window attention to keep decoding cost nearly flat on documents of any length; weights and code are on GitHub and Hugging Face under the MIT license.
A paper introduces Hy-Nomas, an 8B diffusion video world model that follows successive free-form text commands and continuous camera control across multiple interaction turns, generating variable-length segments conditioned on cached history.
A paper introduces LociArt, a unified paradigm for local video stylization that specifies where to stylize and what style to apply while preserving non-target content, built on a GRNStyle component.
A paper proposes ThinkWorld, which uses world imagination to reinforce chain-of-dynamics reasoning in vision-language-action models for robot manipulation.
A new paper proposes Code as Video Caption, using executable programs instead of natural language to capture fine-grained spatial and temporal detail. It introduces CodeCap-Bench to evaluate the approach.
A research paper presents PDRVLA, a vision-language-action robot model whose lightweight Physical Dynamics Module extracts physical tokens from vision instead of generating future video, cutting latency in the control loop.
A paper argues quantization error is assumed to cause unavoidable semantic loss in discrete visual tokens, limiting their semantic fidelity versus continuous representations, and examines whether discrete tokens can serve as semantically lossless representations of visual signals.
A new benchmark evaluates multimodal agents that write code to render every frame and sound of a video, testing exact formulas, charts, rule-consistent games and synchronized narration.
Paper 'Zooming without Zooming v2' targets spatial perception and long-context multimodal understanding for embodied AI and agents.
NamiGPT is a beginner-friendly GitHub tutorial that walks through writing GPT-2 from scratch in PyTorch line by line to show how large language models work.
AgentFlux is a platform for building visual workflows and autonomous AI agents, runnable with Docker on your own computer or server. It drew 101 GitHub stars in its first week.
Kanzo is a self-hosted kanban board for both people and AI agents, shipped as a single 4 MB Rust binary that starts in 2 ms with zero setup.
TokenBar is an open-source, cross-platform app that monitors AI usage quotas for Codex, Claude, Cursor and 47 other services. Built with Flutter for Windows, macOS, Linux, Android and iOS.
goose is an open-source, extensible AI agent that installs, executes, edits and tests code with any LLM. Version 1.57.0 opens the release branch for the next update.
BLA (Blind LLM Agents) is an open-source adversarial defense that protects binaries, Electron apps, JS bundles, Python and WebAssembly against AI reverse-engineering agents by poisoning, exhausting and hallucinating them.
anymove is an open-source Python project that captures motion for any creature from one AI-generated video: it reads joints with DINOv2 across two cameras in one frame and outputs a fitted GLB rig.
The Deep View profiles Halo AI glasses, which emphasize radical openness and privacy; adjacent items cover Claude for Mac 4.29.0, ChatGPT for Mac updates and Gemini CLI 0.66.0.
AstrBot-for-Android is a Kotlin app that runs AstrBot and NapCat locally on Android via a bundled PRoot Ubuntu environment, with native manager UI, WebUI, logs, backup/restore and updates, no root needed.
BAMA (Build Any Mac App) is an open-source repo of agent skills that research, clone, design, build, optimize and test native macOS apps in Swift and SwiftUI. It drew 103 stars in its first week.
Trompe is a TypeScript tool that textures a whole game level from a gray blockout using image models: it plans cameras, paints, projects and bakes real shadows, exporting glTF.
Kinetograph is a TypeScript programmatic video engine designed for AI agents to generate video via code; the new GitHub repo hit 100 stars in its first week.
An open-source Rust web crawler and content-extraction engine for AI agents and RAG that turns pages into clean Markdown, with adaptive crawling, PDF/RSS parsing, browser rendering and Firecrawl-compatible APIs.
Version 3.1.0 of the dictation tool, which uses LLM reformatting, is available across macOS, Windows and Linux.
Unstoppable Domains launched Unstoppable.ai, a model-agnostic AI workspace that lets users work with multiple AI models in one place.
Raycast, a keyboard-driven launcher for Mac, Windows and Linux that lets you control tools with a few keystrokes, shipped versions 2.10.0.0 and 2.10.1.0.
Monologue is AI voice dictation that adapts to your writing style, available on macOS, Windows and Linux. Version 1.5.3 is an update to the desktop app.
Nvidia released the RTX Remix 1.6 update, which adds path tracing support in Minecraft.
Google's AI Overviews in Search cannot be disabled, but PCMag offers workarounds for users who want to avoid the AI-generated summaries.
Otter added a Work mode to its AI Chat assistant that automates more tasks before and after meetings, part of its push beyond transcription following a $320 million funding round last week.
Klaviyo released an AI agent that generates email marketing automation flows from a prompt, aimed at its marketing customers. It turns plain-language instructions into ready-to-run flows.
Digit reports Gemini could soon handle eSIM setup for users, simplifying the process of activating mobile plans on a phone.
Eney is an AI assistant that understands what is happening across a user's apps, files and system, prepares next steps and asks before acting. It offers copy-triggered suggestions and bot-free meeting transcription.
Clearafi compares answers from multiple AI models and scores their agreement before you rely on the result; the tool costs $50.
Zocdoc now lets registered AI agents search providers, check insurance and book appointments, beginning with a partnership with Instinct.
Chrome's enhanced autofill now uses Gemini to pull info from Gmail when filling out forms, and Google is adding more Connected Apps to Gemini and Search via a new developer platform.
The Washington Post put chatbots to the test to see whether they can reliably create voter guides, assessing trustworthiness and accuracy.
Rori, an AI math tutor, has reached roughly 250,000 students across Liberia, Rwanda, Ghana and Sierra Leone, with parents embracing the tool.
Adalat AI says its paperwork tools are used in 6,000 courtrooms across 11 Indian states, speeding up the justice system.
Meta launched Waddle World, a tamagotchi-style 16-bit pet care game whose virtual creatures chat with users and create doodles, described in store listings as a daily journal disguised as a pet.
WIRED reports that AI chatbots could help with emergency communication, since humans struggle with it, but only if people take the AI alerts seriously.
Users reported Microsoft Copilot issues including errors with AI generation, the app and website, and repeated 'Try Again' prompts, according to outage trackers compiled by The Sunday Guardian.
ChatGPT can now connect to Apple's iCloud Mail and Calendar, letting it read email and calendar data to answer questions and draft replies. The integration is rolling out to ChatGPT users.
A new survey finds under-30s are dating more but remain wary of AI-infused dating apps, a challenge for companies adapting products for younger users.
Jan is an offline AI chat tool that runs models locally on your own computer instead of the cloud. Version 0.8.8 is out for macOS, Windows and Linux.
Fathom is an app that records and transcribes video conferences. Version 3.12.0 is out for macOS, Windows and Linux.
Wispr Flow updates its voice-to-text dictation app with AI-powered auto-editing, available on macOS, Windows and Linux.
Google is adding features that make it simpler for Gemini users to find and delete older conversations, a privacy control update for the chatbot.
Meta's developer MCP server gains six tools that let AI agents create and manipulate apps, extending agent capabilities on the Meta developer platform.
Articul8 launched Bivia, a product bringing domain-specific decision intelligence to enterprise AI agents, aimed at grounding agent decisions in industry-specific data and logic.
Amp, the frontier coding agent and development environment, ships a new CLI build version 0.0.1793217669-gb7b7f6, installable via npm with npm i -g @ampcode/cli.
LiteRT in Google Play services now offers GPU inference via ML Drift, about 1.4x faster, with a new CompiledModel API. Switching from bundled binaries cuts app size by roughly 8 MB and enables automatic runtime updates.
Antigravity 2.28.0 is an agent orchestration platform for macOS, Windows and Linux, updated with version 2.28.0.
Zed, a multiplayer code editor with built-in AI agent support, ships version 1.26.2 for macOS, Windows and Linux.
AWS Machine Learning · 3h
#
An AWS blog walks through architecture and an open-source reference blueprint for a governed, multi-account agentic AI platform on Amazon Bedrock AgentCore, covering control plane, gateways and CI/CD evaluation.
Particle-1 is a post-trained DeepSeek V4.1 Flash available via API; Particle says it used fewer tokens across four software and data-engineering benchmarks.
GitHub's Copilot Fusion promises frontier-level AI coding at half the cost and turns Copilot into an agent orchestrator that works across projects and models.
AWS Machine Learning · 3h
#
An AWS blog walks through fine-tuning Qwen2.5-1.5B with PyTorch FSDP across eight GPUs on two EFA-enabled g6.12xlarge instances on Amazon EKS using the Ray Train Deep Learning Container.
AWS Machine Learning · 3h
#
Part 3 of an AWS series on QA Studio with Amazon Nova Act adds apps that organize tests and report health metrics, plus team-based user management controlling test ownership.
A Google Developers post walks through the Agents CLI evaluation loop: continuous testing and grading to catch prompt regressions, compare models via side-by-side trace comparisons, and automate evaluation passes.
Reindeer joined Microsoft's Pegasus program and listed its managed, single-tenant platform that connects agents to systems like SAP and Oracle on Azure Marketplace, with eligible purchases counting toward Azure commitments.
PICO integrated its XR SDK with ten AI coding assistants, so developers can install it with a single npm line.
MCode Stack is an open-source collection of 15 agent skills that MiniMax's own coding team uses daily, giving developers reusable capabilities for coding agents.
TaxDome released a public API and an MCP server letting AI agents and developers connect to its tax and accounting practice management platform.
Microsoft's Copilot Harness, a set of building blocks for running Copilot-style agents, is now offered via Azure AI Foundry so developers can access it directly.
Google's Bring Your Own Agent feature lets Android Studio run over 40 coding agents, including Anthropic's Claude Agent and OpenAI's Codex, plus terminal tools Google says finish UI tasks 77% faster.
Controlled GitHub tests found Claude Code subscriptions burn five-hour quotas at different rates depending on client labels, including a roughly 2.9x gap on one Team plan.
Cursor's Agent SDK lets software teams define agents in code, test them with evals and event replay, and deploy on Cursor's infrastructure or their own. Released October 20.
GitHub made code scanning for AI applications available in public preview, aiming to find security issues in AI app code.
OpenMausBot, published on GitHub by Milind Soni, is an open-source personal agent chat app where each bot is a local Claude or Codex agent with its own personality, model, cloud computer and connected apps, used like messaging contacts.
Fibr AI launched AXO, a tool that adapts existing website content for AI discovery and lets agents perform supported tasks. Results so far come from Fibr's own site and are unverified.
Mistral opened its Responses API in public preview, borrowing OpenAI's format to help developers build AI agents.
Willison argues AI coding agents have become reliable enough to change game development, comparing it to the late-2025 coding-agent tipping point.