When you edit your Taste Profile, you'll impact your personalized playlists like Discover Weekly, recommendations, and Wrapped.
According to Index Ventures Partner Shardul Shah, cybersecurity startup Wiz sits “at the center of three tailwinds: AI, cloud, and security spend.” Those tailwinds powered what just became the largest venture-backed acquisition in history — Google’s $32 billion deal, finalized after a declined 2024 offer, antitrust review on both sides of the Atlantic, and an […]
Databricks shares insights on ai startup funding. From @databricks (Databricks): "What are the defining traits of companies that put their AI initiatives into production? We looked at actual usage data across 20,000+ customers worldwide to identify the building blocks to a successful enterprise AI strategy. One key finding: AI evaluation tools have become a"
Alpha Signal shares insights on openai/gpt. From @AlphaSignalAI (Alpha Signal): "OpenAI just turned your coding agent into a background employee. OpenAI shipped two updates to the Codex desktop app: custom themes and Automations going generally available. Automations now let you: > Pick model and reasoning level per run > Run in an isolated worktree or" | From @AlphaSignalAI (Alpha Signal): "Stop rendering one video at a time. OpenAI just opened its full Video API to all developers, powered by Sora 2. It's a programmatic interface for generating production-ready video at scale. > Custom characters across scenes > 16:9 and 9:16 aspect ratios > Clips up to 20 seconds"
Multiple AI thought leaders discussing ai hardware. From @TechCrunch (TechCrunch): "How to watch Jensen Huang’s Nvidia GTC 2026 keynote" | From @nvidia (NVIDIA): "You can still join us in person in San Jose with an Exhibits Only pass. Register for #NVIDIAGTC:"
38 points • 2 comments on Hacker News
Ollama shares insights on meta/llama. From @ollama (Ollama): "RT @simonguozirui: Running OpenJarvis on my Mini with @ollama to profile Intelligence per Watt! https://t.co/puAvvnYGjw"
Google DeepMind shares insights on google/gemini. From @GoogleDeepMind (Google DeepMind): "RT @pushmeet: Happy to share new progress in AI for Maths @GoogleDeepMind . In extremal combinatorics, AlphaEvolve has helped establish n…"
For decades, manufacturers have pursued automation to drive efficiency, reduce costs, and stabilize operations. That approach delivered meaningful gains, but it is no longer enough. Today’s manufacturing leaders face a different challenge: how to grow amid labor constraints, rising complexity, and increasing pressure to innovate faster without sacrificing safety, quality, or trust. The next phase…
24 points • 2 comments on Hacker News
Peacock is betting on new AI-powered video experiences, vertical clips, and mobile games to help its growth.
After selling his AI startup to AMD for $665 million, Peter Sarlin is back with Qutwo, a new venture building the infrastructure it believes enterprises will need when quantum computing finally arrives.
Simon Willison published an analysis of a pull request where Shopify founder Tobi Lütke used AI-assisted research (autoresearch) to improve the benchmark performance of Liquid, the open-source template language Shopify created 20 years ago, by 53%. Willison noted the case as a striking example of AI improving existing, mature production-grade code rather than just generating new code. The PR is publicly visible and the performance gains are reproducible. For developers maintaining large legacy codebases, the story offers a template for using AI to find and implement optimizations that would be impractical to find manually.
The US military might use generative AI systems to rank lists of targets and make recommendations—which would be vetted by humans—about which to strike first, according to a Defense Department official with knowledge of the matter. The disclosure about how the military may use AI chatbots comes as the Pentagon faces scrutiny over a strike…
Company says it doesn't know how long it will take to restore its Microsoft environment.
Facebook Marketplace is adding a bunch of new AI-powered tools that are supposed to make selling items on the platform a little more efficient. One feature will use Meta AI to automatically respond to those annoying "Is this still available?" messages. You can toggle on the auto-reply option when creating a listing, and Meta AI will draft an editable response to any questions related to availability. In an example shown by Meta, its AI assistant creates an auto-reply saying, "Yes, it's still ...
Computers ordering cappuccinos. A couple of weeks ago, Google and Samsung announced a big Gemini development coming to their newest devices: task automation. Starting with food delivery and rideshare apps, Gemini would be able to use certain apps on your behalf in a virtual window to take care of things like ordering dinner or getting a car to the airport - all based on simple prompts. You know, all the stuff that we've been promised for years AI assistants will be able to do. That feature ...
Pinecone expands its serverless vector database free tier to 100M vectors with no time limit. New features include hybrid search (dense + sparse), metadata filtering, and integration with LangChain, LlamaIndex, and OpenAI.
Google launches Gemini 1.5 Flash optimized for high-volume applications. The model offers 1M token context at 2x the speed and half the cost of 1.5 Pro. Ideal for chatbots, content moderation, and data extraction tasks.
OpenAI opens Sora text-to-video model to ChatGPT Plus subscribers. Users can generate 1080p videos up to 20 seconds with custom aspect ratios. Features include video extension, style presets, and frame-level editing.
LangChain releases version 0.2 with major architecture improvements including native streaming support, async-first agents, and built-in LangSmith observability. The update reduces latency by 40% for complex agent workflows.
Midjourney V7 brings photorealistic video generation with camera motion controls, supporting clips up to 60 seconds. New features include character consistency across frames, style transfer from images, and improved text rendering.
Figma introduces AI-powered design features allowing users to generate auto-layout designs from screenshots or voice descriptions. The update includes AI component suggestions, accessibility checks, and design-to-code export.
Replicate secures $200M funding to expand its AI model hosting platform. The startup now serves 50,000+ developers running 100M+ predictions monthly. New features include fine-tuning UI, model versioning, and enterprise SLAs.
Cohere launches Command R+ Enterprise with enhanced retrieval capabilities, supporting 128k token contexts. Features include citation grounding, multi-step tool use, and enterprise-grade security with SOC 2 compliance.
Elon Musk's xAI releases Grok-3, claiming top spot on MMLU with 89.2% accuracy. The model features real-time X integration, image understanding, and a 1M token context window. Available to X Premium+ subscribers immediately.
Mistral AI announces Codestral 2, a 32B parameter code-specific model achieving 92.1% on HumanEval benchmark. The model supports 80+ programming languages and offers a 256k context window for large codebase analysis. Released under Apache 2.0 license for commercial use.
Google rolled out spend caps for the Gemini API, letting developers define a hard monthly budget limit. Logan Kilpatrick (Google AI Studio DevRel) announced the feature and recommended all developers set a cap immediately. Simon Willison called it great news for anyone running Gemini prompts in CI or building agents that experiment with the API. There is up to a 10-minute delay before a newly set cap takes effect, and developers remain responsible for usage incurred in that window. The feature is the first in a planned series of cost-control tools for Gemini API users.
Anthropic's latest update to Claude will allow the AI chatbot to generate custom charts, diagrams, and other visualizations during your conversation. If Claude determines a visual is useful based on the context of your chat, it will insert the image in-line, rather than in its side panel. As an example, Anthropic says a conversation about the periodic table could lead Claude to generate a visualization of it, featuring interactive elements that let you click inside the table for more informat...
Anthropic rolled out two significant Claude updates this week. First, Claude can now build interactive charts and diagrams directly inside the chat window — available in beta on all plans including free. Second, Claude for Excel and Claude for PowerPoint now share full conversation context when multiple files are open, letting users pull data from spreadsheets into presentations without manually switching tabs. The dual release reinforces Anthropic's push into enterprise productivity workflows. Separately, the Ramp AI Index flagged Anthropic as the top AI stack choice for businesses, adding third-party validation to the product momentum.
Google launched Immersive Navigation, a new 3D navigation mode for Google Maps that Sundar Pichai called the product's biggest upgrade in over a decade. The view renders a vivid, real-time 3D picture of your surroundings with road-level details including lane markings and crosswalks. The update is part of a broader reimagining of Google Maps that the team described as built for the Gemini era, using AI to bring richer context and spatial understanding to everyday navigation. Logan Kilpatrick, who sat down with the Maps team, called it an impressive demonstration of Gemini in action at product scale.
A new Google AI initiative aims to improve heart health outcomes for people living in remote Australian communities.
Today we’re talking about the messy, fast-moving situation at Anthropic, the maker of Claude that now finds itself in a very ugly legal battle with the Pentagon. The back-and-forth is complicated, but as of a few days ago, the Pentagon had deemed Anthropic a supply chain risk, and Anthropic has filed a lawsuit challenging that designation, saying the government has violated its First and Fifth Amendment rights by “seeking to destroy the economic value created by one of the world’s fastest-gr...
The impact of artificial intelligence extends far beyond the digital world and into our everyday lives, across the cars we drive, the appliances in our homes, and medical devices that keep people alive. More and more, product engineers are turning to AI to enhance, validate, and streamline the design of the items that furnish our…
Multiple AI agent companies announce major funding rounds, reflecting investor enthusiasm for autonomous AI systems. Notable raises include MultiOn ($35M), Adept ($150M), and Imbue ($200M) for agent development platforms.
The former Tesla AI director releases a comprehensive free course covering neural networks from scratch. Topics include backpropagation, transformers, and LLM training, with hands-on coding exercises in Python.
Zapier introduces AI-powered automation that understands natural language instructions. Users can describe workflows in plain English, and the AI builds and configures the appropriate Zaps across 6,000+ integrated apps.
SD4 brings photorealistic image generation and 4-second video clips from text prompts. The model shows significant improvement in text rendering and human anatomy, addressing long-standing issues with previous versions.
Windows 12 preview showcases Copilot deeply integrated into the OS, with ability to control settings, manage files, and automate workflows. New "Recall" feature provides photographic memory of user activity for instant retrieval.
Perplexity Enterprise allows companies to connect internal documents, databases, and wikis for AI-powered search. New features include role-based access control, audit logs, and integrations with Slack, Notion, and Google Workspace.
The AI-powered code editor Cursor secures major funding to expand its team and capabilities. The company reports 2M+ active developers and plans to introduce collaborative coding features and enterprise security controls.
SmolVLM enables multimodal AI on smartphones and IoT devices with models under 2B parameters. The release includes optimized versions for iOS and Android, bringing vision capabilities to mobile apps without cloud dependencies.
Meta's Llama 4 family includes models from 8B to 400B parameters, with the largest variant matching GPT-4 on most benchmarks. Released under permissive license for commercial use, marking a significant milestone for open source AI.
Gemini 2.5 Pro introduces native multimodal understanding across text, images, audio, and video. New agentic capabilities allow the model to perform complex tasks autonomously, including research, data analysis, and content creation.
Claude 3.7 Sonnet sets new records on SWE-bench, solving complex software engineering problems with 62% accuracy. The model introduces enhanced tool use capabilities and improved instruction following for enterprise workflows.
OpenAI announces GPT-4.5, featuring significant improvements in mathematical reasoning and code generation. The new model demonstrates 15% better performance on MATH benchmark and supports longer context windows up to 256k tokens.
Most of the devices are made by Asus and are located in the US.
Augment Code argued that modern development is shifting away from classic IDE assumptions and toward workspaces where developers define intent and delegate execution to agents. The company said the basic unit of interest is no longer a single file but an agent, suggesting the next generation of developer tooling will be organized around orchestration rather than manual code navigation. It is a notable framing because it pushes the coding-agent conversation beyond model quality into the shape of the actual interface developers may end up using every day.
Replit announced it raised $400 million at a $9 billion valuation, with investors including Georgian and G Squared. On the same day, it launched Replit Agent 4, featuring real-time multi-user collaboration — multiple people building in the same workspace simultaneously — and a new canvas mode that renders live app previews inline while you code. Early testers described the leap as the biggest product improvement they had felt in any tool. The timing of the funding and launch together signals Replit is positioning itself as the primary platform for AI-native software creation.
Lightning AI promoted Nvidia's Nemotron 3 Super as a model developers can customize, fine-tune, and deploy for reasoning agents in minutes, while related posts from Nvidia and Artificial Analysis emphasized the model's open weights, efficiency, and launch-day availability across inference providers. Taken together, the posts frame Nemotron 3 Super as more than another model release: it is being positioned as an open reasoning model with a real deployment ecosystem already wrapped around it. That combination of openness, benchmark credibility, and immediate infrastructure support is what gives the launch its weight.