← News

SOFTWARE

251 stories

1. SOFTWAREClaude Enterprise will now tell your boss what the team actually did with AI — and what it costAnthropic made smart reports, in-app surveys and pooled group budgets generally available for Claude Enterprise on Wednesday, Oct. 28, 2026.
2. SOFTWAREGoogle's Gemini 4 Carbon arrives two weeks after Argon — and cyber defenders get it firstGoogle DeepMind announced Gemini 4 Carbon on Wednesday, Oct. 28, 2026, two weeks after launching Gemini 4 Argon, with state-of-the-art claims on cybersecurity and long-horizon coding and a staged rollout that starts with cyber defenders.
3. SOFTWAREOracle puts its apps inside ChatGPT, lets Codex build on OCI, and becomes the first cloud outside Google to rent out TPUsOracle said Tuesday, Oct. 27, 2026, at Oracle AI World in Las Vegas that it is bringing OpenAI's ChatGPT and Codex into its applications and cloud, and that Google's TPUs will be available on Oracle Cloud Infrastructure.
4. SOFTWAREApple Creator Studio adds AI assistants to Keynote, Pages, Numbers and Logic Pro — and opens the apps to outside AI agentsApple said Tuesday, Oct. 27, 2026, that Apple Creator Studio is getting intelligent assistants in Keynote, Pages, Numbers and Logic Pro, plus MCP support for external AI agents, as a free update for subscribers in November.
5. SOFTWAREOpenAI ships GPT-6.1 Terra, a cheaper middle model it says matches Astra on agent work at a tenth of the priceOpenAI released GPT-6.1 Terra on Monday, Oct. 26, 2026, for ChatGPT Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and in the API as gpt-6.1-terra.
6. SOFTWAREBlock's Dessa is a voice AI trained only on human speech, and it laughs and interrupts like a personBlock's AI research lab unveiled Dessa on Oct. 23, 2026, a full-duplex speech-to-speech model it says was trained from scratch on nothing but recorded human conversation, with no text and no pretrained models, released as a research preview with demo recordings, a public waitlist and early access for a small group of AI safety researchers.
7. SOFTWAREMoonshot's Kimi K3.1 Pro lands within a point of Claude Opus 5.5 on a top coding test, and the weights are comingMoonshot AI released Kimi K3.1 Pro on Friday, Oct. 23, 2026, the first model in its K3.1 series, with company-reported benchmark scores that put it close to the top proprietary models on coding and agent tests.
8. SOFTWAREAnthropic's Claude Duet is an agent that comes to you first: it watches your apps, works while you're away, and asks before it spends or sendsAnthropic on Wednesday, Oct. 21, 2026, opened a research preview of Claude Duet, a standalone app for iOS, Android and claude.ai where Claude acts as a proactive agent that monitors connected apps, sends a morning brief, runs tasks in the background on its own cloud computer, and asks for approval before it sends, pays, signs or deletes anything.
9. SOFTWAREOpenAI's GPT-6 models are coming to Google Cloud, putting ChatGPT's maker on all three big U.S. cloudsOpenAI and Google Cloud said on Wednesday, Oct. 21, 2026, that OpenAI's frontier models and its Codex coding agent are coming to Google Cloud's Model Garden, with GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna rolling out now in private preview to select enterprise customers and general availability planned in the coming weeks.
10. SOFTWAREGoogle's Gemini can now read your Gmail, calendar and photos while you drive, as Personal Intelligence comes to Android AutoGoogle said on Tuesday, Oct. 20, 2026, that it is starting to roll out Personal Intelligence for Gemini on Android Auto, letting the assistant draw on a driver's Gmail, Google Calendar, Google Photos, Google Maps, YouTube and Search to answer questions in the car, with cars that have Google built-in getting it in the coming months.
11. SOFTWAREGoogle open-sources TorchTPU so PyTorch code can run on its TPU chips, a day after Amazon did the same for TrainiumGoogle said on Tuesday, Oct. 20, 2026, at PyTorch Conference 2026, that TorchTPU, its native PyTorch backend for Google's Tensor Processing Units, is now open source on GitHub, along with new vLLM and SGLang serving engines for TPUs and a preview of a TPU agent suite that helps developers port and tune models.
12. SOFTWAREDeepSeek open-sources DeepInfer, the software it uses to run its own AI models, on Nvidia and Huawei chipsDeepSeek on Tuesday, Oct. 20, 2026, released DeepInfer, the in-house inference framework that serves its DeepSeek-V4.1-Flash model in production, as free open-source code under the Apache 2.0 license, with support for Nvidia's newest data-center GPUs and Huawei's Ascend 950 chips.
13. SOFTWAREAmazon's TorchNeuron lets PyTorch code run on its Trainium chips with one setting change, taking aim at Nvidia's software moatAmazon said on Monday, Oct. 19, 2026, that TorchNeuron, a native PyTorch backend for its AWS Trainium AI chips, lets developers run existing PyTorch code on Trainium with a single configuration change instead of a rewrite, and that it will be released as open source on Oct. 20.
14. SOFTWAREDeepSeek's V4.1-Flash update tops its own coding chart, and its search engine is now for sale to every developerDeepSeek on Saturday, Oct. 17, 2026, released DeepSeek-V4.1-Flash-1017, an update to its V4.1-Flash model that it says is stronger, faster and terser, and opened its in-house web Search and Fetch APIs to all developers.
15. SOFTWAREMeta releases Muse Spark Open, an 816B open-weight model that runs on two Mac StudiosMeta on Thursday, Oct. 15, 2026, released the weights for Muse Spark Open, a natively multimodal model with 816 billion total parameters (42 billion active) and a 1-million-token context window that it calls the most capable US open-weight model to date, and said it worked with Apple so the model can run on a desktop pair of M5 Ultra Mac Studios.
16. SOFTWAREGoogle releases Gemini 4 Argon to everyone who pays, two weeks after holding it back for cyber defendersGoogle on Tuesday, Oct. 13, 2026, made Gemini 4 Argon, its new frontier model, globally available to developers and enterprises through the Gemini API, to Google AI Ultra subscribers in its Antigravity coding platform, and to Google AI Pro and Ultra subscribers in the Gemini app and AI Mode in Search. Google first announced Argon on Sept. 30 but limited it to trusted cyber defenders while it hardened safeguards. Google says Argon ranks first on the Vals Index of finance, coding, legal and tax work.
17. SOFTWAREAnthropic's code is now written so fast that git broke, so seven engineers and Claude built a replacement in five monthsAnthropic said on Tuesday, Oct. 13, 2026, that it has moved its main code repository off git onto Scion, an in-house version control system built for AI agents rather than people. A team of seven built Scion, a code-review tool called Antique and half a dozen related systems in five months, largely with Claude, while the volume of code Anthropic merges each day grew tenfold.
18. SOFTWAREMeta launches Muse Wonder, its most capable AI models yet, and says it has caught up to the frontierMeta on Monday, Oct. 12, 2026, introduced Muse Wonder and Muse Wonder Full, calling them "our most capable models to date" and claiming state-of-the-art performance on personal and professional agent tasks. Muse Wonder is available now through the Meta Model API and Muse Code, Muse Wonder Full is available to Muse Code subscribers, and Wonder will soon power the Muse personal agent app, rolling out to subscribers first. Meta's shares rose about 4% in afternoon trading, Reuters reported.
19. SOFTWAREMicrosoft says its new Maia 300 chip will be the main engine behind its own AI starting in 2027Microsoft unveiled Maia 300 on Monday, Oct. 12, 2026, the third generation of its custom AI accelerator, and said it will scale the Azure Maia family to multi-gigawatt deployments starting in 2027, making Maia "the primary inference system for Microsoft products and services." Reuters reports the company plans to deploy potentially more than a million of the chips next year.
20. SOFTWAREMicrosoft's new AI model doesn't write anything. It just decides, and it's built on Alibaba's QwenMicrosoft introduced Microsoft-Decision-1 late on Friday, Oct. 9, 2026, a "decision-scoring" model for routing, classification, prioritization, verification and agent control. Instead of generating text, it scores a fixed set of answers and returns a calibrated probability for each. It is available now in Microsoft Foundry and through OpenRouter, and was built by post-training Alibaba's open-weight Qwen3.5-9B.
21. SOFTWAREGoogle's Gemini agent gets its own work email and can run jobs for days across Workspace, Microsoft 365 and SlackGoogle Cloud announced on Thursday, Oct. 8, 2026, at its Gemini at Work event that it is launching the Gemini agent — a single enterprise agent that can take objectives, plan the work, connect to business systems, and bring back finished output inside the tools people already use. Coworker-mode agents get their own @agents.company.com email, calendar, Drive and directory presence, and Google says the agent can keep running for hours or days after you close your laptop. The Verge reports it is currently available to enterprise customers in private preview.
22. SOFTWAREAnthropic's new Claude Haiku 5.5 costs up to 90% less than the last oneAnthropic released Claude Haiku 5.5 on Wednesday, Oct. 7, 2026, the third model in its Claude 5.5 family in about a month. It is the company's cheapest and fastest small model yet, aimed at jobs like summaries, sorting and classifying text, live customer support and acting as a helper agent for bigger Claude models. For most requests it is priced 90% below Haiku 4.5, and Anthropic says it costs about 75% less to run on average.
23. SOFTWAREChatGPT stops just talking: GPT-6 arrives for everyone with answers built from charts, diagrams and buttonsOpenAI said on Wednesday, Oct. 7, 2026, that GPT-6 is rolling out globally in ChatGPT together with a new feature it calls Intelligent UI. Instead of only writing paragraphs, ChatGPT can now mix text with diagrams, charts, forms, tappable buttons and small interactive tools, deciding on its own what fits your question. Plus, Pro, Business and Enterprise users get it starting today; free and Go users get it on Thursday, Oct. 8.
24. SOFTWAREClaude now works inside Google Docs, Sheets and SlidesOn Tuesday, Oct. 6, 2026, Anthropic released Claude for Google Workspace in public beta on all paid Claude plans, adding Claude to Google Docs, Sheets and Slides. It also released beta connectors that let Claude create and edit Google files from the Claude app itself.
25. SOFTWAREMeta, Walmart, Stripe and Shopify back one set of rules for how AI agents deal with businessesOn Tuesday, Oct. 6, 2026, Sierra and Meta announced Personal Agent Protocol, an open standard they are building with Genesys, Instinct, Rocket, Shopify, Stripe and Walmart for how personal AI agents connect to businesses. A first draft of the specification is due later this month.
26. SOFTWAREMistral launches Large 4, a 1-trillion-parameter model it will release with open weightsOn Tuesday, Oct. 6, 2026, Mistral opened a public preview of Mistral Large 4, a 1-trillion-parameter model that reads text and images. Mistral says it is the strongest open-weight model built in the US or Europe and leads on cybersecurity work, and it plans to release the weights by the end of the month.
27. SOFTWARETikTok adds an AI shopping assistant and one-click checkout inside the appOn Monday, Oct. 5, 2026, TikTok announced Shopping Assistant, an AI chat agent that answers questions about products you see in ads, and Buy Direct, a one-click checkout that lets you buy straight from a brand from the For You feed. Both are starting with US brands through limited testing.
28. SOFTWAREReflection unveils Beam, a 501B open-weight model it says matches China's GLM 5.2 with 3–4× less computeReflection AI, the New York startup founded in 2024 by former Google DeepMind researchers, introduced Beam on Monday, Oct. 5, 2026, its first open-weight model. Beam is a mixture-of-experts model with 501 billion total parameters and 23 billion active per token, aimed at coding and agent work. Reflection says it scores close to Z.ai's GLM 5.2 on reasoning while using three to four times less inference compute. For now it is early access only; weights, a technical report, and a model card are promised later in October.
29. SOFTWAREOpenAI will start showing picture ads in ChatGPT while you generate imagesOn Monday, Oct. 5, 2026, OpenAI said it is adding a visual ad format to ChatGPT that uses images to show products and services. It will first be tested in the US later in October, appearing while people generate images, with ads clearly labeled and kept separate from the image being made.
30. SOFTWAREAleph Alpha open-sources Kolibri, a 78B German sovereign MoE with 1M contextOn Oct. 3, 2026 — Germany’s reunification day — Aleph Alpha released Kolibri, an English–German Mixture-of-Experts Transformer with 78.1 billion total parameters (about 3.46 billion active per token), context up to 1 million tokens, and full weights on Hugging Face under Apache 2.0 for sovereign on-premise deployment.
31. SOFTWAREAi2 releases Olmo-core 3 to train trillion-parameter MoE models more efficientlyOn Oct. 1, 2026, Ai2 said Olmo-core 3 switches MoE training to a DDP-plus-expert-parallelism design meant to keep large expert pools efficient — including benchmarks past one trillion total parameters.
32. SOFTWARETavus says Griffin is the first Human Interaction Model to pass a video Turing testTavus on Oct. 1 unveiled Griffin, a full-duplex video-to-video Human Interaction Model that saw, heard, spoke, and moved in real time; in the company's live study 26 of 54 participants (48%) believed their partner was human, and Griffin-Lite stays a limited research preview while safety disclosure work continues.
33. SOFTWAREMicrosoft ships MAI-Transcribe-2-Streaming plus MAI-Voice-2.1 for sub-second voice agentsMicrosoft AI said Thursday it launched MAI-Transcribe-2-Streaming — real-time speech-to-text in 60 languages with partials in just over 100 ms — alongside MAI-Voice-2.1 and MAI-Voice-2.1-Flash, so developers can build a full voice-agent loop on Microsoft Foundry and Azure AI Speech.
34. SOFTWAREAmazon's Strands Labs open-sources Strands Decider 2B, a local decision model for agentsAmazon's Strands Labs on Thursday released Strands Decider 2B, an open-source 2-billion-parameter decision model that chooses among supplied options and returns calibrated confidence instead of free-form text, small enough to run locally for agent workflow steps.
35. SOFTWAREGoogle launches Guided Vision in Gemini Live for Android accessibilityGoogle said Oct. 1 it is launching Guided Vision in Gemini Live on compatible Android devices, a voice-forward visual assistant that describes surroundings, answers follow-ups, and coaches camera framing for people who are blind or have low vision.
36. SOFTWAREShopify launches Canvas so merchants can redesign whole stores with SidekickShopify said Thursday it is rolling out Canvas, a new design surface where merchants pan across every store page and work with Sidekick — Shopify’s AI agent — to turn chat into live theme-code changes in real time.
37. SOFTWAREGoogle puts federated learning inside TEEs so outsiders can verify the privacy mathGoogle Research said Friday it shipped a next-generation federated learning system that runs training inside Trusted Execution Environments so device data stays encrypted except inside attested code, with access policies published to a public transparency log — and Gboard is already using it for English and Japanese next-word prediction.
38. SOFTWAREDecagon joins OpenAI's B2B marketplace so enterprises can spend OpenAI credits on CX agentsDecagon said Friday it is a launch partner for OpenAI's new B2B marketplace, so eligible enterprise customers can put part of an existing OpenAI commitment toward Decagon's customer-experience agents while still contracting and paying Decagon directly.
39. SOFTWAREGridCARE's AI finds ~20 MW of spare capacity on an SCE Inland Empire circuitGridCARE said Friday it is working with Southern California Edison to use physics-based AI that identified about 20 megawatts of additional flexible capacity on a constrained Inland Empire circuit, without waiting for a full infrastructure buildout.
40. SOFTWAREMeta says Muse Spark helped mathematicians settle five open research questionsMeta AI Research on Friday published six papers from mathematicians working with Muse Spark 1.1 and 1.2 in Thinking Mode on meta.ai, saying five present answers to previously open research questions after a second mathematician group reviewed the work.
41. SOFTWARECloudflare launches Web Search API via AI Gateway with Ceramic, Exa, and LinkupCloudflare said Friday it is launching a Web Search API through AI Gateway with partners Ceramic.ai, Exa, and Linkup, so agents can pull live web results into inference instead of guessing page URLs and hitting 404s.
42. SOFTWAREApple tightens macOS Full Disk Access as AI agents raise the riskApple said Friday it will add stronger controls around Full Disk Access on macOS, warning that some apps use the permission in ways that can expose a user's files, mail, messages, and browsing history without clear understanding — a risk the company says will grow as AI agents become more autonomous.
43. SOFTWARENielsenIQ puts Optiq conversational AI into Spaceman planogram workflowsNielsenIQ said Friday it added agentic AI and automation to Spaceman, the space-planning platform used by more than 600 retailers in 65+ countries, including Optiq conversational AI for creating and optimizing planograms in plain language.
44. SOFTWARECreatify Labs launches Boreal-H3, a MiniMax H3 video model post-trained for adsCreatify Labs said Friday it launched Boreal-H3, a video model built by post-training MiniMax H3 for advertising work — keeping product and creator identity intact across a clip — available now in its Model Playground and API.
45. SOFTWAREXtalPi puts predictive AI into Fangda Carbon graphite electrode productionXtalPi Holdings said Friday that an AI model built with Fangda Carbon for choosing and blending graphite-electrode raw materials has cleared acceptance testing and is running in Fangda's production workflow to cut cost and speed substitute-material decisions.
46. SOFTWAREQodo 3.0 puts quality control inside the agentic software factoryQodo said it released Qodo 3.0, expanding its AI code quality and governance platform so enterprises building agentic software factories can connect agent workflows to codebase knowledge, architectural standards, and governance — including PR Triage that groups related pull requests into work packages.
47. SOFTWAREAnytime AI 3.0 ships agentic coworkers that finish plaintiff caseworkAnytime AI said it released Anytime AI 3.0, moving the platform from task tools to agents that complete multi-step plaintiff casework and return finished work product — medical chronologies, negligence analyses, discovery responses, and demand letters — with built-in agents for the case types firms actually take plus custom agents shaped around each firm’s workflows.
48. SOFTWAREIvo open-sources Sage, a legal model post-trained for long-horizon contract workIvo said at its Inscribe customer event that it released Ivo Sage — described as the first free open-source model from a legal AI company post-trained for long-horizon contract work — so attorneys, researchers, and developers can download, run, and fine-tune it, while also previewing the Ivo-micro1 Contract Bench.
49. SOFTWARECloudflare open-sources Clef decision models on Workers AICloudflare said it released Clef and Clef-flash, the first models trained by its Workers AI team, hosted on Workers AI and open-sourced under Apache 2.0 on Hugging Face — decision models that read a state and typed questions and return probabilities for every allowed answer, plus a new reinforcement-learning fine-tuning service.
50. SOFTWAREC1.ai launches LLM Gateway to route enterprise model calls by policyC1.ai said C1 LLM Gateway is available today as the final launch of C1 Launch Week, putting governed routing and metered inference in front of enterprise model traffic so security teams decide which data may leave the boundary and finance teams can see who spent what before the invoice lands.
51. SOFTWAREClassie ships Supervise to watch and stop enterprise AI agents in real timeClassie said Classie Supervise is generally available, giving CIOs and CISOs a way to monitor sanctioned and unsanctioned AI agents as they run, enforce policy inline, and keep a chain-of-custody record of what each agent did and what it cost.
52. SOFTWAREDeepKeep ships AI Lens for Developers to watch coding agents like Cursor and Claude CodeDeepKeep announced AI Lens for Developers, extending its AI usage-control and runtime-protection modules so security teams can monitor and govern coding agents that read files, run shell commands, and invoke MCP tools on developer machines — with Cursor and Claude Code supported at launch.
53. SOFTWAREOpenAI parts ways with three safety-team staff over sensitive-information policiesOpenAI said Thursday it has parted ways with three individuals for violating its policies on accessing and handling sensitive company information, after the Wall Street Journal reported the exits involved safety-team researchers allegedly sharing confidential details with a third-party AI-safety organization.
54. SOFTWAREChatGPT adds virtual try-on for clothes and accessories in shopping resultsOpenAI’s ChatGPT shopping experience now lets users virtually try on eligible clothing and accessories from a Try On button in product listings, generate previews with ChatGPT Images from a selfie, and save products to Favorites folders in ChatGPT Library — available on mobile and web.
55. SOFTWAREIBM opens self-hosted IBM Bob for on-prem and air-gapped AI codingIBM said Thursday it introduced self-hosted deployment for IBM Bob, so organizations can bring AI software development and modernization to on-premises, private-cloud, sovereign-cloud, and air-gapped environments without moving sensitive code outside their control.
56. SOFTWARESidero ships Talos Linux for AI and Talos Cupar for private AI stacksSidero Labs said Thursday it introduced Talos Linux for AI, an open-source immutable OS for AI workloads, and Talos Cupar, a commercial platform to deploy and manage models, GPUs, and private data on customer-owned infrastructure.
57. SOFTWAREMixpanel launches Agent Intelligence to tie agent chats to real customer outcomesMixpanel said Thursday it introduced Agent Intelligence in early access — a product that stitches AI agent conversations to customer behavior and business outcomes so teams can measure whether an agent helped, stalled, or converted a user.
58. SOFTWAREDoorDash launches a text-to-order AI agent inside Apple MessagesDoorDash said Wednesday it is opening a U.S. waitlist for a text-based AI ordering agent that works in Apple Messages, letting people ask for a meal or restock in plain language while the agent builds a cart and handles checkout in the thread.
59. SOFTWAREOpenAI says it disrupted a coordinated model-distillation campaign tied to MoonshotOpenAI said Wednesday it identified and disrupted a coordinated campaign to extract protected reasoning from its models — activity consistent with adversarial distillation — and attributed a core cluster of the activity to individuals associated with Moonshot AI.
60. SOFTWAREElevenLabs hits $22B valuation in $300M employee tenderElevenLabs said Wednesday it closed a $300 million employee tender offer that values the voice and conversational-agent company at $22 billion — double its February Series D — led by Wellington and T. Rowe Price.
61. SOFTWAREGMI Cloud raises $668M to expand GPU cloud across the U.S. and AsiaMountain View AI-cloud company GMI Cloud said Wednesday it raised $668 million in new financing — $223 million Series B equity led by ARCHIV with NVIDIA participating, plus a $445 million credit facility led by CTBC — to expand GPU capacity in the United States, Taiwan, and Southeast Asia.
62. SOFTWARECScale exits stealth with $145M for optical interconnect at AI scale-upPalo Alto startup CScale said Wednesday it exited stealth with $145 million in Series C funding to build resilient optical interconnect for AI scale-up — the high-bandwidth, ultra-low-latency links that let accelerators act like one larger computer — bringing total funding to $188 million.
63. SOFTWAREPay-i rebrands as Ascerta and raises $18M Series A for AI ROIThe company formerly known as Pay-i said Wednesday it rebranded as Ascerta and raised an $18 million Series A led by Dell Technologies Capital to give CIOs and CFOs one system for AI cost, adoption, and measurable business value — bringing total funding to $22.9 million.
64. SOFTWAREOpenAI and Synopsys partner on GPT-Synopsys for AI-native chip designSynopsys and OpenAI said Wednesday they signed a multi-year agreement to jointly develop GPT-Synopsys, a specialized model meant to operate Synopsys electronic design automation tools for semiconductor design and verification, with a shared revenue framework and early technology engagements already underway.
65. SOFTWAREKong ships AI Gateway 2.2 with MCP tool governance and identity-aware AI policiesKong Inc. said Wednesday it generally available'd Kong AI Gateway 2.2, adding governed MCP tool access through a single endpoint, identity-aware AI policies, modality-aware cost management, native AWS IAM authentication for Amazon Bedrock AgentCore, and expanded provider support including Kimi, Microsoft Foundry, and Amazon SageMaker.
66. SOFTWAREGoogle announces Gemini 4 Argon, limited to trusted cyber testers for nowGoogle DeepMind said Wednesday it is announcing Gemini 4 Argon, a frontier model for long-horizon software engineering, enterprise knowledge work, and cybersecurity defense, rolling out first to trusted cyber defenders through its Fairwind Program while it strengthens safeguards before wider release.
67. SOFTWARECoreWeave launches Forge to close the model-and-agent improvement loopCoreWeave said Wednesday it launched CoreWeave Forge, a development layer that runs the full AI loop — run, observe, curate, improve, evaluate, and repeat — in one connected environment open across models, frameworks, and other clouds, with MasterClass and Canva already building on it.
68. SOFTWAREOuterSignal raises $22M Series A for agentic customer personalizationOuterSignal said Wednesday it raised a $22 million Series A co-led by Long Journey Ventures and Abstract Ventures to deepen agentic customer intelligence that builds full profiles from commerce and public context and personalizes marketing 1:1.
69. SOFTWAREGong ships Mission Callisto — Enrich, Agent Builder, and Deep ModeGong said Wednesday at Celebrate ’26 it launched Mission Callisto, adding Gong Enrich for account and contact data, Agent Builder for event-triggered custom agents, Deep Mode in Gong Assistant, and previewing Gong Activate for a self-improving revenue loop.
70. SOFTWARERestate raises $20M Series A for durable infra behind AI agentsRestate, founded by the creators of Apache Flink, said Wednesday it raised a $20 million Series A led by Singular, with Redpoint Ventures and Capital One Ventures joining, bringing total funding to $27 million for durable execution infrastructure used from backends to AI agents.
71. SOFTWAREFlow Engineering raises $50M Series B at $750M for AI agents on hardware programsSan Francisco-based Flow Engineering said Wednesday it raised a $50 million Series B at a $750 million valuation for its agentic platform that keeps CAD, simulation, code, and test aligned as hardware teams iterate.
72. SOFTWAREMetaview raises $60M Series C to put AI agents across recruitingMetaview said Wednesday it raised a $60 million Series C led by Insight Partners, bringing total funding to $110 million, to expand AI agents that source, screen, and turn interviews into hiring intelligence while people stay on the decisions.
73. SOFTWAREChina Telecom open-sources TeleOCR, a 1.2B model topping document-parsing benchesChina Telecom AI said Wednesday its TeleOCR document-parsing model, with about 1.2 billion parameters, set a new state-of-the-art on OmniDocBench v1.6 and ranked first on related camera-document and scientific-figure benchmarks, with weights open-sourced on GitHub and Hugging Face.
74. SOFTWAREKanu AI exits stealth with $11.7M to turn employee workflows into owned softwareSeattle-based Kanu AI said Wednesday it raised $11.7 million and emerged from stealth with a platform that watches how employees work and compiles those workflows into production software the company owns inside its own cloud.
75. SOFTWAREFTC opens industry-wide probe of OpenAI, Anthropic and other AI labsThe Federal Trade Commission is conducting an industry-wide probe into Anthropic, OpenAI and other AI labs over potential consumer dangers from the technology, a senior FTC official told Reuters on Wednesday.
76. SOFTWARECorvex launches Token Factory for open-weight inference with zero retentionCorvex said Wednesday it launched Token Factory, a serverless inference platform that serves open-weight models through an API with zero data retention by default and no customer-managed GPU clusters.
77. SOFTWARESekoia Elevate goes GA — agentic SOC verdicts in 99 secondsSekoia said Wednesday its Elevate agentic AI layer is generally available, with agents investigating customer security data and returning an audit-ready verdict in 99 seconds on average.
78. SOFTWAREDeepSeek open-sources Ascend chip tools with Huawei, aiming past CUDADeepSeek said Wednesday it partnered with Huawei to develop programming tools optimized for Huawei Ascend chips and is open-sourcing Ascend platform infrastructure, Reuters reported from DeepSeek’s WeChat post.
79. SOFTWAREVoltropy unveils Vast-10M, a frontier model with 10 million tokens of contextOrange County lab Voltropy said Tuesday it launched Vast-10M, a three-size model family with a native ten-million-token context window, and opened early access on voltropy.com.
80. SOFTWAREDYNA Robotics launches DYNA 2.1, a physical agent that can finish a laundry shiftDYNA Robotics said Tuesday it launched DYNA 2.1, a semi-humanoid physical agent that can autonomously complete full commercial workflows — including an end-to-end laundry shift of loading, drying, folding, and shelving — and is being deployed in hotels, laundromats, and restaurants.
81. SOFTWARERelevance AI launches Invent 2.0 to build and manage AI agent workforcesRelevance AI said Tuesday it launched Invent 2.0, the latest version of its collaborative agent for designing, building, and managing AI agent workforces that automate business processes end to end.
82. SOFTWARECData launches Connect AI Gateway as one control point for enterprise agentsCData Software said Tuesday it launched CData Connect AI Gateway, an evolution of its Connect AI platform that gives organizations a single control point for the models, tools, data, and actions used by AI agents and the people who work with them.
83. SOFTWAREOracle launches Fusion Claw, a governed runtime for complex agentic workOracle said Tuesday it launched Oracle Fusion Claw, a governed agentic execution runtime for Oracle Fusion Agentic Applications that pairs AI reasoning with deterministic enterprise computation so Fusion customers can run longer, more complex outcome-based work at scale.
84. SOFTWAREOpenAI launches $500 Pro 500 plan with Astra UltrafastOpenAI said at DevDay 2026 that it launched Pro 500, a $500-a-month ChatGPT Pro plan with its highest included usage and access to Astra Ultrafast, while reopening Pro 200 subscriptions with a reduced usage allowance for new customers.
85. SOFTWAREOpenAI launches Decisions API for real-time finite-choice answersAt DevDay 2026, OpenAI introduced the Decisions API, which focuses Luna's intelligence on a specific set of user-defined questions with finite pre-defined answers so developers can classify content, route requests, or choose an agent's next action in near real time.
86. SOFTWAREMongoDB launches Atlas Agent Engine for production AI agentsMongoDB said Tuesday at its Investor Day that it launched Atlas Agent Engine, a unified execution, memory, and governance layer for production AI agents, available now in public preview for Atlas customers at agentengine.mongodb.com.
87. SOFTWAREBlackbaud unveils Lantern, a fundraising-specific language model with DatabricksBlackbaud said Tuesday at bbcon 2026 that it launched Lantern, a domain-specific language model built with Databricks and post-trained on fundraising workflows so social-impact teams get intelligence tuned to giving behavior rather than consumer shopping patterns.
88. SOFTWAREOpenAI launches ChatGPT Space so teams can work with ChatGPT and Dots togetherAt DevDay 2026, OpenAI introduced ChatGPT Space, a dedicated place where teammates, ChatGPT, and a user's Dot can build on shared knowledge, with Pages as documents built for people and agents to edit together.
89. SOFTWAREOpenAI launches GPT-6.1 Sol with near-Astra coding smarts at about one-fifth the API priceOpenAI said Tuesday at DevDay it is introducing GPT-6.1 Sol, a major upgrade to GPT-6 Sol that delivers near-Astra performance on agentic coding and professional work at roughly one-fifth of GPT-6 Astra's standard API input and output token prices.
90. SOFTWAREOpenAI launches Dots, proactive agents with their own cloud computerOpenAI said Tuesday at DevDay it is introducing Dots, proactive assistants powered by GPT-6 Astra that run on their own cloud computer, learn from feedback, and can keep working toward goals around the clock through ChatGPT, Slack, Teams, and connected apps.
91. SOFTWAREWhite House launches America.gov as AI front door to federal servicesThe White House said Tuesday President Trump launched America.gov, a single digital point of entry for federal information and services, and signed an executive order directing agencies to integrate covered services into the portal while Americans can already ask questions in plain language.
92. SOFTWARESalt Security launches Claude Connect to inventory MCP servers on Claude EnterpriseSalt Security said Tuesday it introduced Salt Claude Connect, an integration that gives security teams continuous, read-only visibility into the Model Context Protocol servers connected to their Claude Enterprise organization through Claude’s Compliance API.
93. SOFTWAREMcDonald’s uses AI to set “optimal” menu prices store by storeMcDonald’s is increasingly using artificial intelligence to guide menu prices across the U.S. and some global markets, Reuters reported Tuesday after reviewing pricing-engine screenshots and interviewing people with first-hand knowledge of the system.
94. SOFTWAREMeta launches Muse for Small Business with workplace app connectorsMeta said Tuesday it is launching Muse for Small Business, expanding its Muse personal AI agent with skills and connectors so owners can run goals across the tools they already use.
95. SOFTWAREGimlet Labs partners with Cerebras as CS-4 launch partner for ultrafast inference cloudGimlet Labs and Cerebras Systems said Monday they are collaborating to deliver ultrafast AI inference through Gimlet Cloud, combining Cerebras wafer-scale compute with GPUs in a disaggregated architecture and naming Gimlet a launch partner for Cerebras CS-4.
96. SOFTWARESpaceXAI launches Team Bots, shared Grok Bots with team context and memorySpaceXAI (xAI) said Monday it is launching Team Bots — Grok Bots built around a shared role or workflow so a whole team works from the same context, with plugins into apps like Salesforce, Notion, and GitHub, plus Slack handles and per-user private conversations.
97. SOFTWAREShopify opens checkout to browser-based AI agentsShopify said Monday that browser-based AI agents can now finish checkout on eligible merchant stores — including Shop Pay — using new WebMCP tools that read and update the active checkout and submit the order only after the buyer authorizes it.
98. SOFTWAREManus launches Manus 2.0 agent platform and Cue personal agentsManus on Monday released Manus 2.0, saying the rebuild centers on a Cascade agent harness that uses fewer tokens and finishes work faster, upgrades the desktop app into Manus Studio with video and game tools, adds Cloud Computers and event-triggered Automations, and ships Cue — a standalone app where each personal agent gets its own email, phone number, wallet, and computer.
99. SOFTWAREElevenLabs launches Eleven v4 and v4 Turbo speech modelsElevenLabs on Monday released Eleven v4 and its low-latency twin Eleven v4 Turbo, saying the new speech models add finer expression control, stronger voice cloning from as little as 10 seconds of audio, support for more than 90 languages, and median first-speech latency around 100–150 milliseconds for voice agents.
100. SOFTWAREAnthropic launches Claude Sonnet 5.5Anthropic on Monday released Claude Sonnet 5.5, the second model in its Claude 5.5 family, saying the mid-tier model runs more than 30% faster than Sonnet 5, costs up to 30% less per task at the same $2/$10 per-million-token list price, and ships with cyber safeguards previously reserved for its most capable models.
101. SOFTWAREMeta launches Enterprise Platform, hires MongoDB’s CJ Desai to run itMeta on Monday said it is starting Meta Enterprise Platform, a new major pillar of the company that will bring its AI models, agents, and tools — including the Muse agent, Meta Business Agent, Muse API, and Muse Code — to businesses and developers, and named Chirantan “CJ” Desai, who stepped down as MongoDB CEO the same day, as Chief Enterprise Platform Officer reporting to Mark Zuckerberg.
102. SOFTWAREArchipelo launches Salmon to cryptographically verify what AI agents actually did25 Sep 2026 (ET): Archipelo announced Salmon, Execution Verification Infrastructure (EVI) that captures AI-agent and automation runs as signed events with before/after state, building a verifiable execution record across humans, agents, and tools.
103. SOFTWARETestMu ships AI agent that finds duplicate test cases25 Sep 2026 (ET): TestMu AI (formerly LambdaTest) launched a Test Deduplication Agent inside Test Manager that reads titles, steps, and descriptions by meaning, scores matches 0–100, and refuses to delete anything without a human confirmation.
104. SOFTWAREMeetKai takes sovereign AI built on NVIDIA to six countries25 Sep 2026 (ET): MeetKai said its sovereign AI platform, MK-A1 SOVIA, will run on NVIDIA computers in Brazil, Ukraine, Pakistan, Kazakhstan, Uzbekistan, and Bangladesh — more than 700 million people, on the company’s count — after a SERPRO tender it says it won and a Ukraine AI factory whose first power is aimed at 2027.
105. SOFTWAREJobber launches MCP so ChatGPT and Claude can use a service business’s account25 Sep 2026 (ET): Jobber launched Model Context Protocol integrations — a Plugin in ChatGPT and a Connector for Claude — so a service business can ask about its own clients, jobs, quotes, and invoices, and can turn an email inquiry into a client, property, and request, while Jobber stays the system of record.
106. SOFTWAREGen launches AI Data Checker to show how AI apps use your data25 Sep 2026 (ET): Gen Digital opened a beta of AI Data Checker — enter an email or username and it lists connected AI apps plus what they collect, whether data may train models, retention, and privacy controls — as part of its Agent Trust Hub.
107. SOFTWAREOmneky launches TASTE BENCH to score AI ad creatives as ready-to-run25 Sep 2026 (ET): Omneky introduced TASTE BENCH, an evaluation suite that tests how well image and video models turn creative briefs and brand assets into ready-to-run ads — with GPT Image 2.5 Sunburst posting the highest first-attempt ready-to-run rate at 54.2% across eight models.
108. SOFTWARETexture open-sources CommonGrid, a public U.S. power-grid registry25 Sep 2026 (ET): Texture, the grid operating platform bringing AI to energy teams, released CommonGrid — a free, editable, Open Database License registry that connects U.S. utilities, grid operators, power plants, transmission lines, substations, rates, pricing nodes, EV charging stations, and demand-response programs into one public data model with an interactive map and API.
109. SOFTWARELangSmith ships Engine v2 with red teaming and agent fine-tuning25 Sep 2026 (ET): LangChain announced LangSmith Engine v2 at Interrupt NYC — red teaming, broader issue detection, and automated fix validation — plus Managed Deep Agents 0.8 (user-level memory and built-in web search), Trajectories (readable agent-session views), Fine-Tuning via smithtune, and Custom Apps for workspace UIs.
110. SOFTWAREOnpipeline AI turns one request into a quote ready for e-signature25 Sep 2026 (ET): Onpipeline announced an AI agent built into its CRM that can find an existing deal, create a sales quote, generate a PDF, and prepare the document for electronic signature from a single natural-language request — no separate agent setup required.
111. SOFTWAREMicrosoft ships Copilot super app with Home, Code, and AutopilotMicrosoft unveiled its biggest Copilot update yet — a redesigned app with Home (Chat + Cowork), Code for building cloud-hosted apps inside a company’s tenant, Autopilot (the Scout agent rebranded) as a long-running digital teammate, and full Word, Excel, and PowerPoint inside Copilot. CEO Satya Nadella called it a new OS for work.
112. SOFTWARESoundHound ships OASYS Edge for on-device agentic voice AISoundHound AI introduced OASYS Edge, a new embedded architecture that brings LLM-powered voice AI agents onto edge hardware for vehicles and smart devices. Agents can keep talking locally when the network drops, keep cabin audio on-device by default, and hand off to the cloud only when needed.
113. SOFTWAREKontext raises $4 million for runtime checks on AI agent actionsKontext announced $4 million, led by 42CAP, for software that checks an AI agent's identity, task, and target before the action runs.
114. SOFTWAREDocker launches Cloud Sandboxes so coding agents finish in the cloud24 Sep 2026 (ET): Docker introduced Cloud Sandboxes, putting the same microVM-based Docker Sandboxes on Docker-managed always-on compute so coding agents can run for hours without a laptop staying awake. Kits for Claude Code, Codex, Copilot, Antigravity, OpenCode, and Hermes are ready via `sbx --cloud run`.
115. SOFTWAREDataiku launches Agent Management to inventory AI agents across platforms24 Sep 2026 (ET): Dataiku unveiled Agent Management at Dataiku Succeed, a standalone product that finds every AI agent an enterprise is running across platforms, measures performance, and flags the riskiest ones. General availability is set for October.
116. SOFTWAREMultiSensor AI launches Solar Reflection Analysis to cut false thermal alerts24 Sep 2026 (ET): MultiSensor AI launched Solar Reflection Analysis inside its Solar Performance Monitoring product, scoring each thermal alert against sun position and time of day so teams waste less time chasing glare.
117. SOFTWAREDescartes ships Datamyne AI Agent for conversational trade intelligenceDescartes Systems Group issued a GlobeNewswire announcing the Descartes Datamyne AI Agent, a conversational capability inside Datamyne that answers a trade question in plain language and shows the shipment records beside the answer, with a company claim of up to 90% less research and analysis time.
118. SOFTWAREIsland raises $400M Series F at $6.4B for agentic enterprise controlIsland announced a $400 million Series F financing round that brings its valuation to $6.4 billion, more than doubling since 2024, led by Evolution Equity Partners.
119. SOFTWAREAvalara opens Versori agentic integration fabric to partnersAvalara announced Versori by Avalara is available through the Avalara Developer Portal — an agentic integration fabric that plans, connects, builds, and deploys production-grade connectors to Avalara tax and compliance systems.
120. SOFTWAREAutonomyAI launches Discover Mode to close the product-to-PR loopAutonomyAI launched Discover Mode as the final piece of Autonomous Product Delivery. Product managers can ask what to improve (or let the system suggest), research across company data, turn findings into a plan, and generate code that lands as a review-ready pull request for engineers to merge.
121. SOFTWAREGurucul ships AI Risk and Response to catch Shadow AI and risky agentsGurucul announced general availability of Gurucul AI Risk and Response, bringing behavioral AI to SOC and Insider Risk teams monitoring Shadow AI, excessive access, and risky autonomous agents — with AI Prevention controls in preview.
122. SOFTWAREOmneky puts its advertising Growth Agent on the ChatGPT plugin marketplaceOmneky said its plugin is live on the ChatGPT plugin marketplace, so a signed-in ChatGPT user can connect an Omneky account and put the company’s AI Growth Agent to work in chat, researching strategy, generating creatives, and launching campaigns across major ad channels while pulling brand and customer data from connected tools.
123. SOFTWAREOmada buys EmpowerID to govern AI agents at runtimeOmada A/S, an identity governance and administration company, said it acquired EmpowerID and will fold that company’s runtime agent checks into Omada, so an enterprise can decide in real time what each AI agent may do and stop it the moment it steps outside those boundaries.
124. SOFTWAREMeta unveils Muse Charm, a Tamagotchi-like handheld for its Muse AI agentMeta introduced Muse Charm, the first device made for its Muse personal AI agent, a keychain-style handheld with a circular screen that shows a Muse avatar and a fingerprint sensor for voice without a phone, with availability aimed at December 2026 and no price disclosed on the product page.
125. SOFTWARENovaclad launches free iPhone AI-LiDAR app that measures and prices a kitchen countertopNovaclad released a free iOS app built on its AI-LiDAR measurement engine — homeowners scan a kitchen with an iPhone Pro’s LiDAR, get a fabrication-grade takeoff priced from a real shop catalog, and hand the lead to a local fabricator, with Rock Creek Granite as the St. Louis launch partner.
126. SOFTWARESol Foundry emerges from stealth with $4M for AI that does the work you promised in emailSol Foundry Inc. launched Sol, a proactive AI that scans Gmail for commitments, then researches, drafts documents, builds slides, schedules meetings, and writes replies — returning each result for user approval. Backed by $4M from General Catalyst, Nexus Venture Partners, DeVC, Peercheque, and Kunal Shah.
127. SOFTWARECollibra launches Maestro agents, Live Map, and Guardian runtime controlsCollibra announced Collibra Maestro (no-code agents with governance built in), Live Map (a context graph over curated unstructured documents for agentic retrieval), and Guardian Agents with Agent Contracts for runtime control of third-party agents — and said Maestro Assistant is the first Data and AI governance company agent to achieve AIUC-1 certification.
128. SOFTWAREMeta ships Muse Realtime Avatar for live video AI charactersMeta AI Research introduced Muse Realtime Avatar, embodiment tech that turns Muse Realtime Voice into expressive interactive avatars — streaming 448×768 video at 25 fps with about 870 ms latency and up to 12 concurrent sessions per NVIDIA GB200 — with Meta Video Seal watermarks.
129. SOFTWAREBrain launches governed company memory for people and AI agentsBrain (heybrain.io) launched a shared, persistent company memory that connects existing knowledge sources, deploys agents on that memory, enforces each source’s permissions on every request, and records access in a tamper-evident, content-blind ledger, with an MCP endpoint so other AI apps can use the same memory.
130. SOFTWAREGoogle launches Gemini 3.8 Flash TTS and Flash-Lite TTSGoogle introduced Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS — expressive speech models rolling out in the Gemini API, Google AI Studio, and Google products, with generative voice design, a 2,000+ voice library, consent-checked 30-second voice replication, and SynthID watermarking on every clip.
131. SOFTWAREDeepMind adds private server-side memory to Private AI ComputeGoogle DeepMind detailed how Private AI Compute will gain persistent, cross-device AI memory with on-device privacy standards — encrypted cloud storage unlocked only by keys held on the user’s devices, plus a public software attestation record and an independent security audit.
132. SOFTWARERocket Software expands EVA agentic AI on the mainframe and launches PlanGuardRocket Software expanded Rocket EVA, its agentic AI platform for mission-critical systems, and launched Rocket PlanGuard, a security layer that puts a policy checkpoint between AI reasoning and system execution on the mainframe.
133. SOFTWAREZeroDrift launches Anchor 3.0 enforcement models for AI agent communicationsZeroDrift made Anchor 3.0 generally available, a family of small language models built to enforce regulatory and company rules on AI-generated communications before they are sent, and published a FINRA communications benchmark built with Surge AI.
134. SOFTWAREWisdomAI launches Live Apps for governed, prompt-built enterprise analyticsWisdomAI announced general availability of Live Apps, prompt-built governed analytics applications connected to live enterprise data, positioned as a faster alternative to traditional dashboards without giving up accuracy or control.
135. SOFTWAREMarketo co-founder Jon Miller launches Phave, AI-native marketing automationPhave came out of stealth as an AI-native marketing automation platform founded by Marketo co-founder Jon Miller and Nick Bonfiglio, using Maestro to compute per-person playlists that replace rule-based Marketo, HubSpot, Pardot, and Eloqua stacks, with general availability that began in August 2026.
136. SOFTWAREAmazon Seller Assistant goes agentic with workflows and Quick/Claude pluginAmazon said Seller Assistant now runs continuous seller-defined workflows and ships a Selling Partner plugin into Amazon Quick and a Claude beta — Bedrock-backed agentic help for listings, inventory, account health, and ads, with seller approval on actions.
137. SOFTWAREOpenAI ships better prompt caching tools for GPT-6 agentsOpenAI detailed an improved prompt caching stack for long-running GPT-6 agents: higher default hit rates, discounts of up to 90% on cached input tokens when an eligible shared prefix is reused within 30 minutes, plus a hit-rate dashboard, miss diagnostics, explicit breakpoints, and reasoning-effort changes that keep the saved prefix.
138. SOFTWARESpaceXAI runs Cursor+xAI support on Grok Bot without new headcountxAI published how SpaceXAI rebuilt the combined SpaceXAI and Cursor support operation around Grok Bot, handling a 175% ticket surge without new hires, with usage-priced resolutions as low as $0.20 to $0.30 versus the $1 to $4 flat fees typical AI support tools charge.
139. SOFTWARETencent Cloud launches DataBuddy, an agent-native data workbenchTencent Cloud launched DataBuddy, a fully managed agent-native Data + AI workbench that puts governed agents on data engineering, governance, analytics, and data science, available in China, Thailand, South Korea, and Indonesia, with rollouts continuing across Europe, North America, and South America.
140. SOFTWARESelector Foundry lets NetOps teams author AI agents on their own telemetrySelector introduced Selector Foundry, a build-and-run environment inside its network platform so operators can write, test, version, and promote their own AI agents where the network data already lives, without copying that data to an outside agent service.
141. SOFTWAREKantata opens Agent Studio for plain-language custom PSA agentsKantata introduced Agent Studio inside its Expertise Engine, letting professional services firms describe a job in plain language and publish a custom AI agent that runs their delivery standards every time — available and in use by customers now.
142. SOFTWAREStarSling raises $3M and puts code-review agents inside GitHub ActionsStarSling announced a $3 million pre-seed from Bessemer Venture Partners and Y Combinator, plus five more funds, and launched Review Runners in private beta — GitHub Actions jobs that run a team’s own model and repository review skills on every pull request before a person opens it.
143. SOFTWAREDigitalOcean puts agent runtimes and 16,000 tools on one managed stackDigitalOcean opened Managed Agents to public preview, combining Harness Runtime (Firecracker microVM sessions with pause and resume) and Action Gateway (a managed Model Context Protocol endpoint to 16,000+ tools) with its Inference Engine, the service that runs the model, so teams can run coding agents without building the plumbing.
144. SOFTWAREProofpoint unifies data security and AI security in one agentic systemProofpoint announced the Proofpoint Agentic Data and AI Security system at Protect 2026, built to secure AI and data as one connected risk with shared identity, access, and data context.
145. SOFTWAREFindem launches Studio — AI agents that return finished people workFindem launched Findem Studio, combining its people-intelligence graph with expert-backed agents that deliver finished work such as succession plans, market maps, hiring briefs, and skills analyses for teams to review.
146. SOFTWAREAmmonix exits stealth with System One + System Two specialist agentsAmmonix emerged from stealth with an AI architecture that pairs a fast specialized System One decision layer with a System Two language model for reasoning, aimed at high-stakes specialist agents that can run on local hardware.
147. SOFTWARESiemba ships MCP so teams can pentest from their AI assistantSiemba announced general availability of Siemba MCP, a Model Context Protocol server that connects its offensive-security platform to AI assistants so teams can start scans, pull findings, and run packaged analysis skills without opening a separate console.
148. SOFTWARESalt Security adds native AI Detection and Response for LLM runtimeSalt Security announced native AI Detection and Response inside its Agentic Security Platform, adding real-time LLM protection against prompt injection and jailbreaks while linking those hits to MCP tools and APIs the agent reaches.
149. SOFTWARErabbit unveils OS3, an agentic OS with direct computer controlrabbit released OS3, a cloud operating system for software agents that takes a stated outcome as input, then reasons with connected models, skills, and devices—including local PCs via the rabbit agent—across up to five machines per account.
150. SOFTWARESubconscious raises $5.1M for inference built for long-running agentsCambridge startup Subconscious announced $5.1 million across a pre-seed and a seed, led by MassVentures, for an inference platform that compresses and caches agent context so long-running coding agents finish faster and cost less.
151. SOFTWAREParcel Perform rebrands as Perform.AI, launches AI Commerce OSPerform.AI (formerly Parcel Perform) launched its AI Commerce Operating System — covering AI Commerce Visibility, Checkout, Post-Purchase, and Returns with Logistics underneath — and said merchant teams can connect to Perform.AI data through the Model Context Protocol (MCP) inside the AI tool they already use.
152. SOFTWAREChina Telecom AI releases Xing4.0-29B agentic model for single-GPU deployChina Telecom Artificial Intelligence Technology Co., Ltd. (China Telecom AI, also called TeleAI) released Xing4.0-29B-A4B, a lightweight agentic model with 29 billion parameters in total and 4 billion turned on, and says it runs on one consumer graphics card at 15 GB of memory with a 256,000-token context window.
153. SOFTWARELumos launches MCP Governance for Claude Code and CodexIdentity platform Lumos released MCP Governance, a runtime control that checks Claude Code and Codex tool calls—including MCP servers, bash, file operations, and browser use—against policy before they execute, and blocks the action when policy says no.
154. SOFTWARETeradata turns Tera into an agentic coworker for enterprise dataTeradata announced a major evolution of Tera into an agentic coworker for enterprise data work, introducing Tera Context Engine for governed business knowledge, Tera Harness for intelligent execution, and purpose-built Agent Skills for data engineering, analysis, and science.
155. SOFTWAREViam debuts BoxBot box-opening robot for IROS 2026Viam announced BoxBot, a research arm that cuts tape and opens cardboard box flaps, pairing camera-guided cutting with a fine-tuned open-source vision-language-action model trained on 125 person-driven demonstrations collected in about five hours.
156. SOFTWAREUnity launches Simulation Pro early access for roboticsUnity opened early access to Simulation Pro, a supported robotics package for Unity Industry customers that imports a robot description, simulates sensors, connects to ROS 2, and can run many copies on Linux without a screen.
157. SOFTWARENVIDIA ships Isaac ROS 5.0 with CUDA buffer and AI agent skillsNVIDIA published Isaac ROS 5.0 guidance showing all nodes updated to the CUDA buffer backend for ROS 2 Lyrical zero-copy GPU-resident message transport, plus an AI coding-agent skill that migrates CUDA nodes onto that path for Jetson AGX Thor deployments.
158. SOFTWAREOpenAI sets priorities for third-party AI safety assessmentsOpenAI published priorities and principles for independent technical safety assessments of its frontier models, committing to deeper third-party access across training, evaluation, and deployment as part of pacing the frontier.
159. SOFTWAREOuterlimit raises $16M to secure agentic AI action layerOuterlimit emerged from stealth with $16 million in pre-seed funding to build a decentralized security and authorization layer that extends Zero Trust to the agent action layer at the moment of tool execution.
160. SOFTWARESecure Passage launches Truman 2.0 Physical AI OSSecure Passage announced Truman 2.0, an identity-first Physical AI operating system that correlates humans, non-human identities, locations, security systems, sensors, robots, and drones into one vendor-neutral control plane.
161. SOFTWAREFirecrawl launches Alexandria and raises $75M Series BFirecrawl introduced Alexandria, a knowledge library that combines live web crawl with official data providers and Firecrawl indexes for AI agents, and raised a $75 million Series B led by Smash Capital.
162. SOFTWARECisco Talos open-sources CAIRN to hunt AI-integrated malwareCisco Talos released CAIRN (Cognitive Artifact Intelligence Research Network), an open research toolkit that hunts and classifies AI-integrated malware from metadata such as prompt templates and LLM provider endpoints, and began publishing findings with CLOSEDQUORUM.
163. SOFTWAREOpenAI launches GPT-6 Sol and LunaOpenAI introduced GPT-6 Sol and GPT-6 Luna, expanding the GPT-6 family beyond Astra with models trained using similar methods and API prices about 50% lower than GPT-5.6 Sol and Luna promotional pricing.
164. SOFTWAREAnthropic launches Claude Opus 5.5Anthropic published Claude Opus 5.5, the first model in its Claude 5.5 family, saying it performs at Claude Fable 5.1 level on most work while costing about 40% less to run than Opus 5, with faster output and raised subscription usage limits.
165. SOFTWAREAWS Strands Agents open-sources Strands harness general-purpose agentThe Strands Agents project released Strands harness, a fully assembled open-source AI agent you can run locally or deploy on any cloud, under an Apache 2.0 license, with one-line Python or TypeScript setup.
166. SOFTWAREMeshy details Mora for AI-built interactive worlds, ships 7.1 Ultra 4KMeshy publicly detailed Mora (Multimodal Open-world Real-time Architecture) for generating playable interactive worlds, with Mora 1 demos at mora.fun, and highlighted Meshy 7.1 Ultra 2K/4K geometry modes.
167. SOFTWAREMiniMax open-sources MiniMax Code terminal coding agentMiniMax open-sourced MiniMax Code, a terminal coding agent that can understand projects, edit files, and run tests from the command line, with source on GitHub under the MIT license for first-party code.
168. SOFTWAREAzure Cosmos DB ships first GitHub Spec Kit database extensionMicrosoft’s Azure Cosmos DB team announced public preview of the first database extension for GitHub Spec Kit, giving coding agents structured guidance for Cosmos DB apps — partitioning, queries, implementation, and review.
169. SOFTWAREXiaomi open-sources MiMo-V2.6 Pro and FlashXiaomi’s MiMo team released MiMo-V2.6-Pro and MiMo-V2.6-Flash as open omnimodal models — text, image, video, and audio — with Hugging Face weights, a technical report, and disclosed reinforcement-learning tooling, while Xiaomi’s platform changelog dated the series for the same news window.
170. SOFTWAREYandex open-sources AliceAI-Foundation-80B base modelYandex released AliceAI-Foundation-80B-A3B-Base, an 80-billion-parameter mixture-of-experts language model that turns on about 3 billion parameters per token, trained from scratch and published under Apache 2.0 on Hugging Face with a long technical report.
171. SOFTWARESpaceXAI ships Grok 4.7 for coding and knowledge workSpaceXAI released Grok 4.7, calling it its most capable model for coding and long knowledge-work tasks. The company says the model uses a larger base than Grok 4.6, longer reinforcement-learning training on multi-hour problems, stronger self-checks, and a new safeguard stack — at the same list price as 4.6.
172. SOFTWAREAmazon blocks Meta’s Muse AI agent from shopping its storeAmazon blocked Meta’s Muse AI agent from shopping Amazon.com on users’ behalf, citing unauthorized-agent terms, a lack of notice or agent identification, and credential risk; Meta has said Muse cannot see passwords or payment methods.
173. SOFTWAREStepFun opens Step 5 Preview API — 600B MoE, 27B active, weights Oct 15StepFun said Step 5 Preview is available through its API and Studio — a 600-billion-parameter sparse Mixture-of-Experts model that turns on 27 billion parameters per token, with a 1-million-token context and vision input — while full open weights are scheduled for October 15, 2026.
174. SOFTWAREQwen open-sources Image-2.1, a 7B model that generates and edits (with transparency)Alibaba’s Qwen team released open weights for Qwen-Image-2.1 — a unified image generation and editing model with a 7B visual-generation component, native transparent (RGBA) output, up to 10 reference images, and mask/circle local edits — via Hugging Face, GitHub, and the Qwen blog.
175. SOFTWAREAlibaba's Qwen launches Qwen3.8-Omni-Flash for multimodal agentsQwen announced Qwen3.8-Omni-Flash, an API model that natively handles text, images, audio, and video, with a 1-million-token context window, tool calling, and web search. Qwen says it is designed to turn long media into completed work, such as video editing, translation, and recaps.
176. SOFTWAREOpenAI tests Sponsored Agents inside ChatGPT AdsOpenAI said it is introducing new AI-powered experiences for ChatGPT Ads. The company is testing Sponsored Agents, which let people start a conversation with a business-sponsored agent after clicking an ad in ChatGPT. Those agent chats are clearly labeled and stay separate from ChatGPT’s own answers and the user’s original conversation. The test is with select advertisers in the United States. OpenAI also said advertisers can create ads with a few prompts in ChatGPT Work, that Ads Manager is getting new AI creative tools, and that HubSpot (first CRM partner) and Shopify (first ecommerce partner) now connect ChatGPT Ads into tools businesses already use. Shopify’s ChatGPT Ads app is available today for US merchants, with international rollout in ChatGPT Ads markets starting September 23 (company).
177. SOFTWAREAnthropic merges Claude Cowork and chat into one ClaudeAnthropic said Claude Cowork and chat are merging into one Claude, so users no longer pick a separate place for bigger agentic work. The company also introduced Claude Docs and Claude Slides, and said Claude Design now works inside conversations. The merge is rolling out to Pro and Max plans first over the next few weeks on web, desktop, and mobile, with Team and Free to follow and Enterprise admins getting at least 30 days’ notice before changes.
178. SOFTWAREMistral models power Firefox Smart Window in Mozilla partnershipMistral and Mozilla announced that Firefox Smart Window (beta), Mozilla’s AI browsing assistant, is now powered by Mistral models — including Mistral Small 4 for users in the US and Canada, with Smart Window beta and French-language support expanding in France. The UK and Germany are expected later this year. Both companies say the deal is about open distribution, multilingual models, and keeping AI choice in the browser; Mozilla says chats aren’t saved on its servers by default and partners such as Mistral agree to zero data retention.
179. SOFTWARExAI ships Memory in Grok Build — coding agent keeps project notes across sessionsxAI’s newsroom said Grok Build now has memory. As you work, it keeps notes on conventions, decisions, and project facts; later sessions read those notes before touching related code. The company frames it as Grok Build getting better the more you use it.
180. SOFTWAREFootprint raises $25M Series B to scale Percy, an AI OS for financial-crime complianceFootprint said it raised $25 million in Series B funding led by QED Investors, with participation from MUFG, Commerce Ventures, LightBank, and Alumni Ventures, plus continued backing from Index Ventures, Lerer Hippeau, BoxGroup, Operator Partners, and Animal Capital. The New York company builds Percy, an AI operating system for financial-crime compliance workflows used by banks and fintechs.
181. SOFTWAREBackOps raises $42M Series B to automate supply-chain claims, billing, and exceptionsBackOps said it raised $42 million in Series B funding led by Insight Partners, with participation from Theory Ventures, Construct Capital, Gradient Ventures, and 10VC. The San Francisco company builds an AI resolution layer for firms that make or move physical goods, automating work across carrier claims, chargebacks, billing, and related operations.
182. SOFTWAREArcee AI hits $1B+ valuation with Series B for American open-weight modelsArcee AI said it closed a Series B led by Vista Equity Partners, Cambium Capital, and Emergence Capital that values the company at more than $1 billion. The San Francisco lab builds frontier open-weight models (including Trinity Large) and says the capital will accelerate the next Trinity generation, expand work with the U.S. Department of Energy and national labs, and grow products for customizing and running open models.
183. SOFTWAREOrbits raises ~$2M from a16z Speedrun to grow Bit, an AI that runs the householdOrbits said it closed nearly $2 million USD (about $2.6 million CAD) in pre-seed funding from Andreessen Horowitz’s a16z Speedrun fund, N49P, Garage Capital, and angel investors to expand Bit, an AI household assistant available on iOS and Android. The company says Bit consolidates calendars, lists, and communications while executing real-world household tasks.
184. SOFTWARENoetive exits stealth with $41M seed to bring self-improving AI to factories and logisticsNoetive said it emerged from stealth with $41 million in seed funding led by Eclipse to build what it calls an intelligence of record for the physical economy — a self-improving AI that learns how a business runs, paired with multi-modal sensing pods so agents can perceive real-world operations. The company says it is already working with design partners in manufacturing, logistics, energy, and data centers.
185. SOFTWAREEnvive AI raises $15M Series A for self-improving agents that sell for brandsEnvive AI said it closed a $15 million Series A led by Fuse VC, bringing total funding to $20 million, to scale self-improving, brand-safe AI agents that learn from shopper behavior and drive outcomes across search, sales, content, and support. The Seattle company says brands including Spanx, Supergoop!, Coterie, and Wine Enthusiast already use the system.
186. SOFTWAREAbre launches Vera — analytics agent for K-12 district data questionsAbre said it launched Vera, an Analytics Agent built into its K-12 Data Intelligence Platform so educators can ask plain-language questions of district data instead of living in static dashboards. The company lists Vera 360 chat, Liveboards, AI Highlights, AbreViz, and KPI alerts, and says packages ship in Basic, Enhanced, and Premium tiers.
187. SOFTWAREMarket Logic ships DeepSights Radar GA as always-on AI market early-warningMarket Logic Software announced general availability of DeepSights Radar, a new capability in its DeepSights market-intelligence platform. The company says always-on AI agents monitor selected industries and topics, maintain structured multi-chapter profiles, and surface what changed since the last update.
188. SOFTWAREPhrase ships Atlas GA so teams run localization work by chatting with the platformPhrase announced general availability of Atlas, a conversational AI interface built directly into the Phrase Language Intelligence Platform. Users can configure multilingual projects, build automation workflows, generate reports, and troubleshoot issues in natural language. The company says Atlas is available now to all Phrase Platform customers.
189. SOFTWAREMomentus launches MomentusMAX and makes MAX for Sales generally availableMomentus Technologies said MAX for Sales is generally available to Momentus Enterprise and Elite customers. The release also launches MomentusMAX, the company’s new umbrella for venue AI products. MAX for Sales automates lead capture and scoring, calendar conflict resolution with revenue comparisons, and full proposals with upsells (company).
190. SOFTWAREFotor launches Agent to turn briefs into editable multi-track videoFotor said Fotor Agent is available. The tool takes a brief or raw assets and produces complete, editable video projects and 4K motion graphics through natural-language prompts, with multi-track editing instead of a flattened AI clip (company).
191. SOFTWARESalesforce unveils AIforce so CRM context works inside Claude, Slack, and CoworkerSalesforce said AIforce is a live interface layer that brings Salesforce’s data, workflows, business logic, permissions, security, and governance into AI surfaces. It launches with Claudeforce (Salesforce in Claude, beta), Slackforce, and Agentforce Coworker (company).
192. SOFTWARESPS Commerce makes MAX agentic AI generally available for supply chainsSPS Commerce said MAX, its agentic AI layer inside the SPS supply-chain network, is now generally available to fulfillment customers after early access. The same announcement expanded Visibility Management and Decision Intelligence so brands and retailers can ask questions, spot anomalies, and act on inventory and purchase-order risk using network data (company / Business Wire).
193. SOFTWARECoder puts Claude Code on Agent Relay so the work stays on your machinesCoder said Claude Code now runs through Agent Relay, so the coding agent’s work happens inside the customer’s own Coder workspaces — sandboxed and logged — while Anthropic still handles billing and the agent loop.
194. SOFTWAREAugmentir launches Augie Command Center so factories run ops in plain languageAugmentir announced at IMTS 2026 that Augie Command Center and Augie Procedure Studio are generally available. The company says frontline leaders can deploy industrial AI agents, build dashboards, analyze operations, and turn SOPs into interactive procedures by describing what they want in natural language — with approval gates and audit trails (company).
195. SOFTWARESimScale puts an engineering AI agent inside Onshape CADSimScale launched an Engineering AI Agent that runs inside PTC’s Onshape cloud CAD, available through the Onshape App Store and announced at IMTS 2026 in Chicago. From a single prompt, the company says the agent can propose a simulation setup, prepare CAD geometry, run physics (CFD, thermal, electromagnetic, FEA), and explain results in plain engineering language — without handing the job to a specialist or leaving Onshape.
196. SOFTWAREShanghai AI Lab opens Atria Dawn Preview for long-horizon research agentsATRIA, from the Shanghai Artificial Intelligence Laboratory, released Atria Dawn Preview — an open agentic model built to carry scientific work from a method in the literature to executable experiments, reproducible metrics, and an inspectable report. The model card and weights are live on Hugging Face; documentation and evaluations are published at atria-asi.ai.
197. SOFTWAREDataCebo ships SDV 2.0 so companies can own a generative model of their databasesDataCebo released SDV 2.0, software that lets an organization build a generative relational model of its own databases inside infrastructure it controls — capturing structure, relationships, and business rules across tables — then use that model for synthetic data, scenario generation, and AI-agent training without moving production rows. Self-service pricing starts at $500/month; available now.
198. SOFTWARETestMu AI’s Kane CLI now designs tests straight from the PRDTestMu AI (formerly LambdaTest) launched the Assurance Lifecycle in Kane CLI — new commands that ingest requirement documents (PRDs, specs, policy pages), design acceptance tests permanently linked to those requirements, measure coverage from sealed evidence packs, and flag stale tests when a source document changes.
199. SOFTWARECyberhill’s Cerebro now plugs business context into Claude, ChatGPT, Gemini, and GrokCyberhill Partners said Cerebro, its enterprise semantic layer, now feeds an organization’s workflows, terminology, and business rules into the enterprise versions of Anthropic’s Claude, OpenAI’s ChatGPT, Google’s Gemini, and xAI’s Grok — so the model reasons over company systems of record with a traceable path, while leaving data in place. The company says the capability is available now.
200. SOFTWARECockroach Labs launches Continuum, an Agentic Database CloudCockroach Labs announced Cockroach Continuum, which it calls an Agentic Database Cloud — software that treats an organization’s full database estate as one elastic system. Continuum pools capacity across fleets, adds agent-assisted operations (Cockroach Aegis), and bills for work performed rather than peak capacity provisioned, as AI agents create more databases and traffic than human operators can size by hand.
201. SOFTWARELectra launches Apogy, agentic AI for fashion product developmentLectra announced Apogy, a cloud product-development platform for fashion that embeds agentic artificial intelligence to help teams move from ideas and sketches to manufacturable prototypes. The Euronext-listed company says agents assist decision-making and automate low-value tasks while centralizing product data for brands, partners, and suppliers — with early customers including O’Neills and Oniverse group brands.
202. SOFTWARELevio and Cohere partner on sovereign AI for Canadian governmentsLevio and Cohere said they formed a strategic partnership to help Canadian governments and regulated enterprises move AI from pilots into production while keeping control of data, deployment, and language environments — including French.
203. SOFTWAREGoogle launches Gemini 3.8 Live voice models with Extended ThinkingGoogle said Gemini 3.8 Live and 3.8 Live Extended Thinking are rolling out for developers, enterprises, and consumers — voice agents that keep talking while tools run in the background, with live visual grounding and 97-language support.
204. SOFTWARECreatify Labs launches Boreal, a cent-per-second realtime video modelCreatify Labs said it launched Boreal, an AI video model for text-to-video and image-to-video billed at about one cent per second of finished video and aimed at realtime speed — roughly a five-second clip in about five seconds. The company says blind human reviews and ad-quality checks put Boreal near top closed models advertisers use today, at far lower cost and about 40 times the speed in its benchmarks.
205. SOFTWAREWSO2 ships Agent Manager GA to govern enterprise AI agents without locking to one stackWSO2 announced general availability of Agent Manager, an open-source control plane meant to secure, observe, and manage AI agents across models, frameworks, and deployments. The company says GA adds per-agent, per-environment identity controls, MCP-level governance, and a sandboxed runtime after a June 2026 beta.
206. SOFTWAREDecimal AI raises $4M seed for an AI Support Engineer that reads code and logsDecimal AI said it raised $4 million in seed funding co-led by Khosla Ventures and Kearny Jackson, with Atlassian Ventures and Weekend Fund participating. The San Francisco company sells an AI Support Engineer meant to resolve software customer issues end to end — including digging through code, logs, and production data — and create bug-fix drafts for engineering review.
207. SOFTWARESalesforce and Google Cloud unify agents and Hyperforce for one AI stackAt Dreamforce 2026, Salesforce and Google Cloud said they expanded their strategic partnership so Salesforce workloads can run on Google Cloud via Hyperforce and so agents on either platform can reason and act on the same data by connecting Salesforce’s headless architecture with Gemini Enterprise — without custom integrations.
208. SOFTWAREMeta launches Meta One, a paid plan for more AI usage across its appsMeta introduced Meta One, a new subscription across Instagram, Facebook, WhatsApp, and Meta AI. The company says plans unlock more AI usage — including Muse-powered image and video generation — plus creator and business tools, while the core free experience stays free.
209. SOFTWAREGoogle Cloud opens Singapore Engineering Center for enterprise cloud and AIGoogle Cloud inaugurated the Singapore Engineering Center, calling it its flagship product development hub in Southeast Asia. The center sits with Southeast Asia’s first Google DeepMind research lab and is meant to turn frontier AI research into production-grade cloud and AI solutions tuned for Singapore companies competing worldwide.
210. SOFTWARESalesforce launches Koa, a CRM reasoning model built on NVIDIA NemotronSalesforce and NVIDIA announced Koa, Salesforce’s first CRM reasoning model for Agentforce, built on NVIDIA Nemotron. Salesforce says it post-trained NVIDIA Nemotron 3 Super on proprietary synthetic enterprise scenarios — not customer data — so agents can reason through multistep CRM work inside Salesforce’s trust boundary. Pilot customers named include 1-800Accountant, Baxter Credit Union, Engine, Formula 1, UChicago Medicine, and Xero; general availability is expected in winter 2026 in U.S. regions.
211. SOFTWARETSA Ace AI agent handles ~100k traveler chats a month on Salesforce AgentforceSalesforce said the Transportation Security Administration (TSA) has deployed Ace, an AI agent built with Salesforce Public Sector Solutions, to answer routine traveler questions about airport security. Since going live during the summer travel surge, Ace has handled about 100,000 conversations a month and resolved 96% of routine inquiries — such as packing liquids or checkpoint rules for medical devices — without human escalation, per Salesforce’s press release.
212. SOFTWAREZendesk ships Specialized AI Agents for industry and custom workflowsZendesk introduced Specialized AI Agents purpose-built for industry and company-specific work. The launch pairs Industry Agents — starting with commerce workflows such as shopping, orders, returns, delivery issues, and refunds — with Custom Agents that businesses build in Zendesk’s no-code Agent Builder around their own policies, tools, and approvals.
213. SOFTWAREfal acquires Lucent, bringing creative-media AI agent founders in-housefal, a generative media platform for image, video, and audio generation, said it acquired Lucent, an AI agent-based creative-media platform, and brought Lucent co-founders Dimi Panagiotopoulos and Alex Koumpas onto its team. Terms were not disclosed. Lucent’s approach let creators brief an agent in plain language while the system orchestrated models, edits, and brand rules behind the scenes.
214. SOFTWARESuperhuman buys Fathom, the AI meeting notetakerSuperhuman announced it acquired Fathom, an AI meeting partner that captures, searches, and acts on what happens in meetings. Superhuman plans to bring Fathom’s notetaking into its suite spanning email, calendar, collaborative documents, and databases, and to feed that meeting context into Superhuman Go — its proactive AI assistant across apps and websites.
215. SOFTWAREBolt.new launches Forge to widen who can build with open modelsBolt.new launched Bolt Forge, a research-preview agent that runs only on open-source models and gives every individual Pro plan up to 50 times more usage through October 14 — if the builder opts in to share anonymized sessions that help train new open-weight models with Arcee AI.
216. SOFTWARESourcegraph makes Agentic Batch Changes generally available for Cloud customersSourcegraph said Agentic Batch Changes is generally available for Sourcegraph Cloud customers — an AI agent that plans, runs, and tracks a code change across hundreds or thousands of repositories from one prompt, with pricing based on the pull requests that actually merge.
217. SOFTWAREApple ships iOS 27 with Siri AI — the rebuilt assistant is live in English betaApple said the next generation of Apple Intelligence is available today across iOS 27, iPadOS 27, macOS 27, watchOS 27, and visionOS 27, powering Siri AI — a reimagined Siri that can use personal context from messages, email, and photos, react to what’s on screen, and take more systemwide app actions. Siri AI starts rolling out in English beta; French, Japanese, Korean, Portuguese, and Spanish are slated next month. It will not be available initially in the EU on iOS, iPadOS, and watchOS, and Apple Intelligence features including Siri AI remain unavailable in China while Apple works through regulatory requirements.
218. SOFTWAREPerplexity brings Portable Computer local agents to Windows RTX PCsNVIDIA’s blog says Perplexity’s Portable Computer — a local version of its Computer agent — is now in the Perplexity app for Windows on compatible GeForce RTX PCs and RTX PRO Workstations, using local models for multistep on-device work and asking permission before any cloud handoff.
219. SOFTWAREByteDance’s Doubao Phone Assistant ships consumer version with beta phone controlsByteDance’s Doubao Phone Assistant released its consumer version — more stability-focused than last year’s tech preview — with screen-based Q&A, local search across photos, messages, and notes, voice and dedicated AI-key wake, and a beta “operate phone” feature, plus a new SAEP (Screen Automation Execution Protocol) 30-day public rule window so third-party apps can declare whether the assistant may automate their screens. The first consumer build is on the nubia NaviX Ultra, set to go on sale Sep 16.
220. SOFTWARESokin plugs licensed payment rails into ChatGPT, Claude, and Gemini via MCPLondon-based payments platform Sokin said it released a Model Context Protocol (MCP) connector that lets authorized finance teams use AI tools such as ChatGPT, Claude, Gemini, and Perplexity to pull live balances, FX quotes, and payment status — and to prepare beneficiaries, currency exchanges, and payment instructions that still need human confirmation and any existing Sokin approval rules before settlement.
221. SOFTWAREOpenAI says Perplexity trusts GPT-6 Astra with end-to-end systemsOpenAI published a customer story saying Perplexity — the AI answer-engine startup — is using GPT-6 Astra to craft communications, edit real production software, monitor systems, and build end-to-end test harnesses.
222. SOFTWAREHyundai puts Atria AI Data Flywheel into full operationHyundai Motor Group said its self-driving Data Flywheel — a loop of collection, AI training, checks, and deployment — is now in full operation, and outlined a two-track plan with NVIDIA production targets plus its own Atria AI system.
223. SOFTWARENVIDIA and CrowdStrike unveil SafeMind agentic cyber defense on NemotronNVIDIA’s blog says Jensen Huang and CrowdStrike’s George Kurtz announced SafeMind at Fal.Con — cybersecurity software that can take a sequence of actions, built with Nemotron open models, plus Falcon IQ — with company claims and CrowdStrike’s internal test numbers staying theirs.
224. SOFTWAREOpenAI ships Data agent in ChatGPT Work for company data and dashboardsOpenAI published a Data agent in ChatGPT Work that connects approved company sources, investigates changes, and builds shareable interactive dashboards in natural language — with Alpha and customer usage figures staying OpenAI’s.
225. SOFTWAREGoogle shows autofinetune — agents that post-train language models overnight on TPUsGoogle published autofinetune, a demo in which an agent post-trains large language models overnight on Cloud TPUs using Tunix, Gemma, Antigravity CLI, and Gemini Flash 3.7.
226. SOFTWAREGoogle puts Lyria 3.5 music model in the Gemini app and APIGoogle put Lyria 3.5, its music-generation model, into the Gemini app and the Gemini API, with genre and vocal-or-instrumental controls plus short or longer tracks.
227. SOFTWARECohere releases North Small Translate, an open-weight machine-translation modelCohere released North Small Translate, a mixture-of-experts machine-translation model with 218 billion total parameters and 25 billion active, covering 50+ languages under a non-commercial open-weight license.
228. SOFTWAREMistral: AI agents migrate 40,000 lines of Fortran to C++Mistral published a case study on helping an unnamed European energy operator migrate 40,000 lines of Fortran 77 to C++ in a physics-intensive reservoir simulator that had no test suite and no centralized documentation.
229. SOFTWAREGoogle launches a Gemini desktop app for WindowsGoogle published a native Gemini desktop app for Windows 10 and 11, with an Alt + Space shortcut that opens Gemini over the user’s active work.
230. SOFTWAREMistral partners with Cloudera for sovereign enterprise AIMistral published a partnership with Cloudera to run Mistral models on Cloudera’s hybrid data platform — including private cloud, on-prem, and air-gapped — and to train custom models on proprietary data customers keep.
231. SOFTWAREDeepMind ships agentic video understanding for Gemini FlashDeepMind published that agentic video understanding is live on Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite — a loop that searches the tape instead of reading every frame at a fixed rate.
232. SOFTWAREGoogle Labs opens Dreambeans to all US accountsGoogle Labs opened Dreambeans — its personalized daily-stories experiment — to every eligible U.S. account age 18 and older on Android and iOS, including ordinary consumer accounts.
233. SOFTWARESalesforce unveils Trusted Enterprise AI Harness and AI Control PlaneSalesforce published the Trusted Enterprise AI Harness — six named capabilities plus an AI Control Plane for Salesforce and third-party agents — with core pieces available now and the unified experience slated for early fiscal 2028.
234. SOFTWAREIBM releases Granite Time Series PatchTST-FM-r2 at about 385M parametersIBM Research published Granite Time Series PatchTST-FM-r2 — about 385 million parameters — on a Hugging Face blog and model card, with GIFT-Eval ranks and “SOTA” lines staying IBM’s dated claims.
235. SOFTWARECognition applies GPT-6 Astra so Devin can test and prove its workOpenAI published a customer story saying Cognition is applying GPT-6 Astra across Devin Cloud, CLI, and Desktop so the coding agent can test its work and return evidence.
236. SOFTWAREOpenAI and GSA expand ChatGPT to state, local, and tribal governmentsOpenAI and the U.S. General Services Administration announced a multi-year OneGov deal that drops the ChatGPT license fee to $0 — from a usual $15 a month — and cuts usage 50% for federal, state, local, and tribal governments.
237. SOFTWAREOpenAI ships ChatGPT Images 2.5 with Sketch and faster API modelsOpenAI published ChatGPT Images 2.5, saying it offers sharper detail, more precise multi-turn edits, and up to 50% less wait than Images 2.0.
238. SOFTWAREGoogle, Meta, and Microsoft ship new speech-to-text models for agentsGoogle, Meta, and Microsoft each published speech-to-text models for voice agents and live workflows — software that turns talk into text — and the word-error and speed scores stay the companies’.
239. SOFTWAREDeepSeek ships V4.1-Flash on its API with native multimodal and new pricingDeepSeek published DeepSeek-V4.1-Flash, a multimodal mixture-of-experts model on the API as deepseek-flash with up to 1 million tokens of context and a posted off-peak cached-input price of $0.003 per million tokens.
240. SOFTWAREOpenAI pauses new $200 ChatGPT Pro sign-ups over Astra demandOpenAI’s help page says it is temporarily pausing new sign-ups and upgrades to the $200 ChatGPT Pro plan because Astra demand is straining capacity, while existing $200 subscribers are unaffected.
241. SOFTWAREMeta launches Muse, a personal AI agent, in the USMeta introduced Muse, a personal AI agent on iOS, Android, and muse.ai — plus WhatsApp chat, with AI glasses later — that the company says runs in a Muse Secure VM with a separate Sentinel.
242. SOFTWAREGoogle DeepMind launches Gemini 3.8 Flash and Flash CyberGoogle DeepMind launched Gemini 3.8 Flash — its most capable Flash model, it says, for coding and agents — and Gemini 3.8 Flash Cyber for finding and fixing software holes, gated through a Fairwind Program.
243. SOFTWAREDeepMind’s WeatherNext 3 adds hourly satellite-fed forecastsGoogle DeepMind and Google Research published WeatherNext 3, a global weather model they say takes live weather-satellite mosaics and refreshes hourly at about 5 kilometers on key surface fields.
244. SOFTWAREHugging Face rebuilds AUTOMATIC1111 image-generation tools as Gradio Workflow1111Gradio authors published Workflow1111, rebuilding most of AUTOMATIC1111’s image-generation toolbox as a 73-node Gradio Workflow canvas with eleven media pipelines.
245. SOFTWAREOpenAI launches Data agent in ChatGPT WorkOpenAI published a Data agent for ChatGPT Work that connects to company data sources and, the company says, builds interactive dashboards from ordinary-language questions.
246. SOFTWAREOpenAI launches GPT-Live-1, a full-duplex voice model, in the APIOpenAI put GPT-Live-1, a full-duplex voice model that can listen and speak at the same time, into the developer API at $0.05 a minute for the voice layer.
247. SOFTWARECognition introduces SWE-2, a coding model for DevinCognition introduced SWE-2, a coding model for its Devin product, and claims 50.0% on FrontierCode 1.1 Main — within about a point of Anthropic’s Fable 5.1 — at 64% less cost.
248. SOFTWAREOpenAI launches ChatGPT for Financial ServicesOpenAI launched ChatGPT for Financial Services, a ChatGPT Work plan for banks and research desks built on GPT-6 Astra with built-in market data.
249. SOFTWAREOpenAI documents an Agents API for durable cloud agentsOpenAI published official documentation for an Agents API that runs durable cloud agents with managed sessions, sandboxes, MCP connections, and subagents.
250. SOFTWAREOpenAI published GPT-6 Astra on its own siteOpenAI published a product page for GPT-6 Astra on its own site, and that live page is the confirmed event.
251. SOFTWAREAnthropic published Claude Fable 5.1 and Mythos 5.1 — same model, different safeguardsAnthropic published Claude Fable 5.1 and Claude Mythos 5.1 and says they are the same model with different safeguards.