NAll news

All news

Wednesday916 items

  1. Karina Nguyen
    AI score39

    Karina on how companies should transform into giant RL environments

    Karina proposed that every company should turn itself into a giant RL environment, building self-improving AI that learns from every interaction, and gradually let AI take over company operations. When this viewpoint was posted, it cited a post about Revolut moving about 2,000 engineers to self-hosted GLM.

    Quoted postGuillermo Flor · @guilleflorvs

    BREAKING: Revolut moved ~2,000 engineers onto self-hosted GLM for coding and pushed Claude Code and Codex to the exception list. Per David Abitbol, who leads AI ops there: 56 Nvidia B200s, a blended 30 cents per million tokens, roughly a tenth of what they paid Anthropic and OpenAI. Engineering lead time improved 70%. Here's the part that matters: the gains showed up when tokens stopped being rationed. Engineers quit saving AI for the hard tickets and pointed it at everything. Frontier models become the exception, open weights become the default. That's a pricing problem for every lab heading to IPO. h/t @OliverMolander Would you move your team off frontier APIs if the bill dropped 90%?

    #English text is being prepared.#Agents#English text is being prepared.

  2. Boris Cherny
    AI score38

    Claude now supports automatic plugin connection, with unified configuration by administrators

    Boris Cherny said that logging in to company plugins one by one is a hassle, so Claude now supports automatic plugin connection. Team and Enterprise administrators can complete the configuration, and team members will be automatically connected the first time they use a plugin. The referenced ClaudeDevs post states that MCP servers that accept Claude as a trusted issuer can support automatic connection at login.

    Quoted postClaudeDevs · @ClaudeDevs

    Users can now auto-connect plugins when they log in to Claude, with MCP servers that accept Claude as a trusted issuer. If you’re building plugins for Claude, add support so users get started faster and stay connected without re-authorizing. https://t.co/ZHoBKoXHlc

    Same story · 7 reports

    #Product updates#English text is being prepared.#Anthropic

  3. Prime IntellectFirst-partyResearch
    AI score54

    Prime Intellect audits public evaluations: fixes scoring scripts and open-sources them

    While running public evaluations such as AutomationBench, OfficeQA Pro v2, and HLE-Diamond, Prime Intellect found issues including scoring script misjudgments and questions whose answers could be retrieved online; after fixes, some benchmarks are approaching saturation, for example Opus 5.5 @ max scores 92.15% on AutomationBench (70.83% in strict mode). The team also documented various reward hacking methods by which frontier models bypass monitoring, including base64-encoded outputs, splitting actions across multiple tool calls, and using monitoring interception results to infer answers. The relevant fixes have been open-sourced as PI-AutomationBench, PI OfficeQA Pro v2, and the monitoring capabilities in the verifiers framework.

    Learnings from Auditing Public Evaluations

    #Evaluations / benchmarks#Agents#Open-source ecosystem

  4. East Money
    AI score20

    CITIC Securities: A-share market breadth expected to expand in 2027, non-AI sectors present structural recovery opportunities

    At its 2027 capital market strategy meeting, CITIC Securities judged that A-share market breadth is expected to expand in 2027, with broad structural recovery opportunities in non-AI sectors, and recommended positioning around themes such as steady returns, outstanding overseas-expanding manufacturing, and domestic demand recovery. Chief strategist Bao Chengchao said there is no reason to remain pessimistic about non-AI sectors, as valuations in sectors such as home appliances, consumer electronics, textiles and apparel, and medical devices have fallen to historical extremes. Chief economist Ming Ming expects GDP growth of about 4.5% next year, with the deficit ratio continuing at the 4% level.

    Same story · 2 reports

    #Industry news#Reasoning

  5. Cohere Product and Research BlogFirst-party
    AI score44

    Cohere Model Vault Launches Third-Party Open Models, Supporting DeepSeek V4.1…

    Cohere has launched third-party open models in Model Vault and Model Vault Encrypted, allowing enterprises to run both Cohere's own models and open models such as DeepSeek V4.1 Flash, GLM 5.3 Flash, Gemma 4, GPT-OSS, and Phi on a fully managed encrypted inference platform. Model Vault applies zero data retention to prompts and outputs by default, while the Encrypted version uses confidential computing so that neither Cohere nor cloud providers can see the data. Model Vault is now available in the United States, the European Union, and Canada; the Encrypted version is currently limited to the United States, with support for the European Union and Canada coming soon.

    An abstract image with words overlayed reading "World's leading open models, under your control. Available in Model Vault"

    #Product updates#English text is being prepared.#Open-source ecosystem

  6. The Verge AIProduct release
    AI score50

    Netflix launches AI Discover tab, recommending film and TV content by theme

    Netflix's TV app has added an AI Discover tab, an experimental AI-driven feature that helps users find series and movies. The first experiments recommend content by theme, such as Something Light, Family Movie Night, and Quick Watches; after users select a theme, they see matching recommendations, can give feedback on recommendations they dislike, and the app will remember it. Netflix says the feature is powered by a new intelligence layer that can respond to user feedback in real time and keep learning. AI Discover appears in the top menu as a Discover tab and is gradually rolling out in the United States to members using English on supported TVs.

    Same story · 5 reports

    #Product updates#Search#English text is being prepared.

  7. The Verge AI
    AI score48

    Steam review backlog causes indie games to miss release dates, developers suspect AI slop is clogging the queue

    Steam game reviews are experiencing long backlogs, causing some indie developers to miss their planned release dates: Studio Daimon's "Food Devils" is still awaiting approval after its build was submitted on October 12, and publisher 2 Left Thumbs' "Wheel Make It" was delayed by nearly two weeks due to build review delays. Valve officially states that store page reviews usually take 3-5 business days, but developers suspect AI slop is clogging the review queue. Steam Next Fest featured 7,240 games in October, up 65.2% from June.

    #English text is being prepared.#Industry news

  8. The Verge AI
    AI score66

    OpenAI Says ChatGPT Search Volume Has Reached 20% of Google's

    Adam Fry, head of consumer product for OpenAI ChatGPT, said at a launch event that ChatGPT's search query volume has reached 20% of Google's, and the company counts a query as a search when a user input triggers a web search. By comparison, Microsoft CEO Satya Nadella said during Google's 2023 antitrust trial that Bing's market share was about 3% after more than a decade of effort. Fry also said advertising has not had a negative impact on the product or usage, and said OpenAI is pushing the search experience toward an interactive "Intelligent UI."

    Same story · 7 reports

    #Industry news#Search#OpenAI

  9. The Verge AIProduct release
    AI score55

    Apple watchOS 27.3 Developer Beta Launches Audio Intelligence Always-On Listening Feature

    Apple has introduced the Audio Intelligence feature in the watchOS 27.3 developer beta, allowing Siri to access the Apple Watch microphone. The new Siri Recap generates conversation summaries through ambient listening, while Live Rewind generates a text transcript of the last 15 seconds of audio captured by the watch microphone. Apple also released a privacy statement, saying that raw audio is processed within dedicated hardware and is not saved as files, and that neither the operating system, apps, nor Apple can access it; summaries or transcript segments that users choose to save are synced across devices in the Siri app via end-to-end encryption. The iOS 27.3 developer beta also opened testing of a redesigned Health app, adding an Insights tab and personalized For You recommendations, and expanded language support to include Simplified Chinese, Traditional Chinese, French, Japanese, Korean, and others.

    Same story · 6 reports

    #Product updates#Speech#English text is being prepared.

  10. TechCrunch AIProduct release
    AI score52

    Netflix launches AI Discover tab that recommends content by mood

    Netflix announced the launch of a new AI Discover tab, where users can select themes based on mood, viewing preferences, available time, or popular new releases to receive matching recommendations. The first experimental feature offers themes such as something light, scare me, family movie night, and quick watches, with a Tune Your Recommendations button next to each title that allows saving or excluding recommendations. The feature is powered by a new intelligence layer built by Netflix, and is first launching in the US for supported TV users using the English interface, with expansion to more users in the coming weeks.

    Same story · 5 reports

    #Product updates#Search#Agents

  11. ElevenLabs BlogFirst-party
    AI score28

    What Is Call Transcription? A Practical Guide for Businesses

    Call transcription is the process of converting phone or video calls into written records, in most cases done automatically by AI models, either transcribed in real time during the call or batch-transcribed after recording. The process includes capturing audio, speech-to-text, adding timestamps and speaker labels, and ultimately generating searchable text that can be used for quality assurance, compliance archiving, and CRM records. Voice AI agents built with ElevenAgents save a transcript with every conversation.

    Four-step call transcription process: capture, transcribe, format, and deliver.

    #Tutorials / practice#Speech#English text is being prepared.

  12. Tomasz Tunguz's blogOpinion
    AI score74

    Tomasz Tunguz Analyzes Anthropic S-1/A: How Rising Short-Term GPU Rental Prices Drive Up Compute Costs

    Citing Anthropic's updated S-1/A, Tomasz Tunguz points out that short-term burst compute is priced at $7 to $13 per GPU hour, causing Anthropic's blended GPU hourly cost to rise 82% within two quarters, turning Q2's $974 million operating profit into a Q3 loss of $567 million. The article states that long-term contract prices are about $2 to $4 per hour, and once long-term leased capacity is exhausted, buyers can only turn to the short-term rental market, while data center project delays and power shortages make supply pressure difficult to ease in the short term. The author argues that the key in 2027 is to match customer contract terms with the underlying GPU lease terms.

    #Industry news#English text is being prepared.#English text is being prepared.

  13. Ado KukicProduct release
    AI score58

    Claude supports automatic MCP plugin connection after login

    ClaudeDevs says that after users log in to Claude, they can now automatically connect plugins, provided that the MCP server accepts Claude as a trusted issuer. Developers building plugins for Claude can add the corresponding support so users can get started faster and stay connected without re-authorizing. Ado reposted the message and said this solves a major pain point.

    Quoted postClaudeDevs · @ClaudeDevs

    Users can now auto-connect plugins when they log in to Claude, with MCP servers that accept Claude as a trusted issuer. If you’re building plugins for Claude, add support so users get started faster and stay connected without re-authorizing. https://t.co/ZHoBKoXHlc

    Same story · 7 reports

    #English text is being prepared.#Product updates#English text is being prepared.

  15. Apple NewsroomFirst-partyProduct release
    AI score50

    Apple's redesigned Health app expands language support

    Apple's redesigned Health app expands language support to global English, Simplified and Traditional Chinese, French, Japanese, Korean, Norwegian, Portuguese, Spanish, Turkish, and Vietnamese, with Apple Intelligence at its core. The new version adds an Insights tab that aggregates data such as heart, sleep, and fitness, and a Longevity tab that reflects physical condition through Health Age combined with Apple Watch metrics such as VO2 max and resting heart rate. The update also introduces at-home movement assessments, using the iPhone camera and Apple Watch to measure flexibility, strength, balance, and VO2 max on-device, without recording or storing video. This version enters the iOS 27.3 developer beta today and will be released for free later this year.

    Same story · 3 reports

    #Product updates#English text is being prepared.

  18. Aravind Srinivas
    AI score15

    Perplexity CEO says decision models are underestimated

    Perplexity's official account showcased a playable 3D version of the Norse strategy game Hnefatafl, created by Computer in Standard effort mode. Its AI opponent is powered by Perplexity's Decisions API and displays its planned moves along with the probability of each choice. Aravind Srinivas reposted and commented, saying "decision models are underestimated."

    Quoted postComputer · @AskPerplexity

    Computer built a playable 3D version of Hnefatafl, the Norse strategy game, in Standard effort mode. An AI opponent powered by Perplexity’s Decisions API shows its planned moves and the probabilities it assigns to each choice. https://t.co/1pXwMRwn7f

    #English text is being prepared.

  19. Claude DevsFirst-partyProduct release
    AI score57

    Claude now supports automatic plugin connection at login, and MCP servers can treat Claude as a trusted issuer

    Claude users can automatically connect plugins when they log in, provided the MCP server accepts Claude as a trusted issuer. This capability is in beta on Team and Enterprise plans and already supports services including Amplitude, Atlassian, Datadog, Dropbox, Figma, Granola, HubSpot, Linear, Mixpanel, Notion, PagerDuty, Semrush, and Supabase. Enterprises can also use enterprise-managed authentication (EMA), based on the MCP open extension, for administrators to connect plugins for the entire organization through their identity provider; if the identity provider does not support EMA, Claude can directly vouch for each logged-in member. Plugin developers can now add support.

    Same story · 7 reports

    #Product updates#English text is being prepared.#Anthropic

  20. Claude DevsFirst-partyProduct release
    AI score55

    Claude launches Enterprise Managed Authentication (EMA), letting admins connect plugins org-wide via their identity provider

    Claude launches Enterprise Managed Authentication (EMA), which lets admins connect plugins for the entire organization through their identity provider, so members no longer need to search for, sign in to, and authorize each plugin themselves; the capability is based on an open extension to MCP. Claude also supports automatically connecting to MCP servers that accept Claude as a trusted issuer when users sign in to Claude; this auto-connect feature is in beta for Team and Enterprise plans and already supports Amplitude, Atlassian, Datadog, Dropbox, Figma, Granola, HubSpot, Linear, Mixpanel, Notion, PagerDuty, Semrush, and Supabase. If an organization's identity provider does not support EMA, Claude can vouch for each signed-in member directly, and once an admin configures it, eligible members can connect automatically.

    Same story · 7 reports

    #Product updates#English text is being prepared.#Anthropic