AI news · dalao.ai

Fri, Oct 30

Granola launches free cross-app voice typing on Mac, expanding beyond meeting notes

AI meeting-notes app Granola launched voice typing on Oct 29: on a Mac, users press the Fn key and speak, and Granola cleans up the speech, adds punctuation and paragraph breaks, and inserts the text into any app for notes, AI prompts, emails or messages. Granola says it is free with unlimited words; it is Mac-only for now, with iOS and Windows planned.

Why it matters: Granola, valued at $1.5B, began as a meeting notetaker; dictation tool Wispr ($2B) added meeting notes months ago. Granola moving into dictation shows voice AI apps converging to own users' everyday input.

originally Oct 30

OpenAI-backed Thrive Holdings to acquire ADT in a $15.3 billion transaction

ADT and Thrive Holdings announced on Oct 29 a definitive agreement for Thrive Holdings to buy ADT for $10 per share in cash, valuing it at about $7.6 billion in equity and $15.3 billion in enterprise value, a 45.6% premium to the unaffected pre-announcement close. ADT's board approved it unanimously; the deal has not yet closed.

Why it matters: Thrive Holdings, spun out of Thrive Capital, holds established businesses long-term and revamps them with AI; OpenAI, a partner and investor, supplies models and forward-deployed engineers. No specific AI products are announced yet.

Oct 30

OpenAI updates its Model Spec, reorganizing model behavior guidance around five core values

OpenAI published a revised Model Spec on Oct 29, organized around five values: faithfulness to human intent, minimizing harm, honesty and transparency, human agency, and care. It adds guidance on context, memory, personalization and human oversight, adjusts some behavioral defaults, drops most detailed style rules, and now calls the model "GPT" rather than "the assistant".

Why it matters: The Model Spec is OpenAI's public statement of how its models should behave. Moving from itemized rules to core values aims to let models use judgment where instructions are silent, while keeping human oversight.

OpenAI

originally Oct 30

Cohere announces initial close of Series E targeting up to CAD $7 billion at a CAD $35.7 billion valuation

Cohere unveiled a Series E targeting up to CAD $7 billion, which would represent a CAD $35.7 billion valuation. The companies of Schwarz Group lead; new investors include Bell Ventures, Bullhound Capital and the Government of Canada (equity); Inovia Capital, NVIDIA and Prosperity7 Ventures return. Funds go to the model and product roadmap, vertical AI for regulated sectors and long-term compute.

Why it matters: Cohere made the initial close of its Series E, targeting up to CAD $7 billion at a CAD $35.7 billion valuation, which it calls the largest private round in Canadian startup history.

originally Oct 29

SuperMe raises $20M Series A led by Forerunner for its network of AI expert profiles

SuperMe announced a $20M Series A on Oct 29, led by Kirsten Green at Forerunner with Greylock Partners. It turns experts' writing, talks and notes into AI profiles users can talk to; questions are routed to a group chat of relevant profiles, and the network is reachable from Claude and ChatGPT via MCP. Founded in 2024, it has about 13 employees.

Why it matters: AI replicas of real experts' knowledge are emerging as a product category. SuperMe distributes expertise via multi-profile group chats and MCP; its claim that personal data and trade secrets never appear in answers is the company's own.

Oct 30

Ex-Palantir execs launch Dream with $17M seed led by a16z to build AI world models for construction

Dream emerged from stealth on Oct 29 with a $17M seed round led by Andreessen Horowitz, with SV Angel, XYZ and Zigg Capital. Founded by ex-Palantir executives, it builds 'built world models' it says are trained on 100M+ sq ft of US permit data and generate code-checked construction documents, including structure, plumbing and HVAC. It reports 10 active projects in California and Texas.

Why it matters: World models are moving beyond robotics and video into constrained physical work like building design. Dream aims to replace legacy tools such as AutoCAD and Revit with buildable drawing sets; its capability claims are the company's own.

Oct 30

Thu, Oct 29

Collective gets $100M, including a $40M Series B led by General Catalyst, QED and Gradient, for its AI financial OS for solo businesses

Collective said on Oct 29 it secured $100M: a $40M Series B led by General Catalyst, QED Investors and Gradient Ventures, with Intuit Ventures, FT Partners and SemperVirens, plus a $60M growth facility from General Catalyst's Customer Value Fund to pay for customer acquisition. Its AI-native platform handles bookkeeping, invoicing, payroll and taxes for one-person businesses.

Why it matters: AI is moving into back-office finance for solo businesses. Collective says its ARR is growing 120% a year and AI resolves about half of support requests; the money also funds banking and insurance products.

originally Oct 29

Solink raises more than $100M led by Vertu Capital to turn existing security cameras into AI business data

Solink said on Oct 29 it raised more than $100M led by private equity firm Vertu Capital, with existing investor Growth Equity at Goldman Sachs Alternatives plus Golub Growth and Comcast Ventures; no valuation was disclosed. Its AI turns footage from cameras already in stores, restaurants and warehouses into operational data, from about 500,000 cameras at 35,000+ locations in 65 countries.

Why it matters: Solink plans to build an intelligence layer for its AI agents; its CEO expects agent products to top 40% of revenue within three years, from under 10% (Globe and Mail). Vision AI on existing cameras is a path for AI in physical retail.

originally Oct 29

Digital Science, owner of Overleaf, acquires AI academic writing platform Jenni AI

Digital Science said on Oct 29 it has acquired Jenni AI; terms weren't disclosed. Jenni AI is an AI research and writing assistant the company says has more than 8 million users. In the coming months it will connect to Dimensions, a database of 171 million publications, to find sources and check citations. Jenni stays standalone with the same team and pricing.

Why it matters: General AI tools often invent references, so academic writing tools are moving toward grounding in real publication databases. Jenni AI now sits alongside Overleaf and ReadCube under one research-data company.

originally Oct 29

Mistral releases open weights for Mistral Large 4, its 1-trillion-parameter flagship, on Hugging Face

On Oct 29 Mistral released the weights of Mistral Large 4 ("Le Chonk"), in API preview since early October. The multimodal MoE model has about 1T total and 49B active parameters and three reasoning modes. Mistral says three more weeks of RL lifted Terminal-Bench 4.0 from 28% to 41%, calling it the best open-weight coding model (vendor claim). A quantized version runs on four Nvidia B200/B300 GPUs.

Why it matters: One of the largest open-weight models from a European lab: developers can now self-host and fine-tune a 1T-parameter multimodal model. Mistral says smaller distilled versions, including a laptop-class one, will follow.

Mistral AI

originally Oct 29

Wabi, the group AI assistant from Replika founder Eugenia Kuyda, raises $55M at a $1.25B valuation

Wabi said on Oct 29 it raised $55M led by IVP at a $1.25B valuation, with a16z, Forerunner and the Amazon Alexa Fund participating. Founded by Replika creator Eugenia Kuyda, Wabi is a personal AI assistant built for groups: it combines an AI agent, personalized mini-apps and shared chats to plan trips, book appointments and split household tasks.

Why it matters: Most AI assistants are one person, one chat; Wabi bets on assistants shared by groups. The round values Wabi above $1B as personal AI assistants become a crowded race that includes Meta.

originally Oct 29

Serval raises $150M Series C led by Sequoia at a $3B valuation

Serval said on Oct 29 it raised a $150M Series C led by Sequoia at a $3B valuation, triple its $1B valuation from the December 2025 Series B, bringing total funding to $277M. The money goes to Catalyst, its automation agent, and ProServal, a new unit placing automation engineers inside large customers. The company says ARR has doubled every quarter since the Series B.

Why it matters: Serval uses AI agents to build and run internal IT, HR and finance workflows, positioning itself as a replacement for ServiceNow. Three rounds in about a year show investors betting on AI-native apps displacing legacy enterprise SaaS.

Oct 29

Deepgram to acquire Pipecat maker Daily, plans independent Pipecat Foundation

Deepgram announced on Oct 29 it is acquiring Daily; terms weren't disclosed. Deepgram will keep building Daily's realtime infrastructure and make Pipecat, Daily's open-source agent framework, the core of its new voice AI products. Pipecat will move to a new independent foundation whose steering committee includes Deepgram rivals. Deepgram also launched RCS, an open voice-agent protocol.

Why it matters: Pipecat is an open-source framework for realtime voice agents. Once the deal closes, Deepgram would add the orchestration layer and realtime network infrastructure to its models while handing the framework to an independent foundation.

Oct 29

AgiBot unveils embodied foundation model GO-3 Preview, plans to open-source it by the end of 2026

AgiBot unveiled an embodied foundation model, GO-3 Preview, at its 2026 Innovation Day on Oct 29. Built on in-house WRAM (World Reasoning Action Model) architecture, it models task understanding, future prediction and action together, with the Genie Evolver post-training system and Genie Master agent framework. It says it tops 3 benchmarks including RoboDojo and will open-source it by end-2026.

Why it matters: AgiBot already open-sourced GO-1; if GO-3 ships openly as planned, researchers and robot makers get a newer base model for manipulation work. Benchmark results are company claims so far, with no technical report yet.

originally Oct 29

Doubao Work opens third-party models to team and enterprise subscribers, starting with Kimi 3.1 Pro, GLM 5.3 and DeepSeek V4.1 Flash

On October 29, Doubao Work began offering third-party models to its team and enterprise subscribers, starting with Kimi 3.1 Pro, GLM 5.3 and DeepSeek V4.1 Flash. The feature is on by default: no separate purchase from model makers is needed, and usage draws on existing AI credits. Admins can scope access by user group and device, and enable models per vendor or individually.

Why it matters: Doubao Work, ByteDance's workplace agent launched in August, now bills rival makers' models (Moonshot, Zhipu, DeepSeek) through the same subscription credits, letting firms pick models per task in one workspace.

originally Oct 29

Accounting AI platform Ravical raises $17M seed round led by MMC Ventures

Ravical announced a $17M seed round led by MMC Ventures, with Lakestar, Pitchdrive and Shapers joining; Tech.eu puts total funding at $25M. Founded in 2025, Ravical turns how an accounting or tax firm's senior staff judge and advise into shared workflows that AI applies across every client. It says over 100 European firms use it; the money funds European expansion and a US launch.

Why it matters: Accounting firms struggle to hire while clients lean on AI advice; Ravical bets on scaling a whole firm's expertise through AI rather than single-task point tools.

originally Oct 29

Procurement AI agent startup Fragment emerges from stealth with $56 million across seed and Series A

Fragment came out of stealth on Oct 28 with $56M raised in total: a Series A led by 8VC and a seed round co-led by General Catalyst and A*; per-round amounts were not disclosed. Its Autonomous Context Engine (ACE) connects existing enterprise systems so AI agents can investigate and resolve purchasing and payment exceptions in direct procurement.

Why it matters: AI agents are moving into specific enterprise workflows; Fragment says it already sells to dozens of Fortune 1000 companies and deploys without replacing existing systems.

originally Oct 28

Tencent launches Buddy One, a cross-app AI office suite that uses WorkBuddy as its workspace and links Tencent Docs, Meeting and ima

On Oct 29, at its Global Digital Ecosystem Summit, Tencent launched Buddy One, which uses its WorkBuddy agent as a workspace and links Tencent Docs, Tencent Meeting, ima and a new Tencent cloud drive under one account and credit pool. WorkBuddy can then search authorized documents, meetings and knowledge bases across apps and answer with cited sources. A Buddy One VIP plan costs RMB 700 a month.

Why it matters: Tencent is consolidating several office products behind a single AI agent entry point, a direction similar to Microsoft's Copilot and Google's Gemini integrations in their office suites.

originally Oct 29

AI inference chipmaker FuriosaAI closes $620 million Series D at a roughly $2.9 billion valuation

FuriosaAI closed a $620M (KRW 820B) Series D at a ~$2.9B post-money valuation. Korea's National Growth Fund was the largest investor at $280M; international investors added $110M, Infinitum the largest, and existing holders $125M. Funds go to RNGD inference chip production, software and a third-gen accelerator. The company expects 20,000 RNGD units in 2026 and 40,000+ in 2027.

Why it matters: Inference silicon is heating up; the round lets FuriosaAI scale RNGD output and fund its next chip, with LG's EXAONE deployments and Upstage already running on RNGD.

originally Oct 29

US Justice Department makes inquiries into Nvidia's $12.93 billion deal to buy Hugging Face

Reuters reported on Oct 28, citing two people familiar with the matter, that the US Justice Department has asked how Nvidia's $12.93 billion deal for open-source AI platform Hugging Face would affect competition, including whether it would give Nvidia a key distribution point for AI. CNBC later confirmed it. Nvidia declined to comment; the deal is expected to close in the first half of 2027.

Why it matters: Hugging Face is the main hub for open models and datasets; the review could affect whether the deal closes and on what terms developers and rival chip and cloud providers keep using the platform.

NVIDIA

Hugging Face

originally Oct 28

Anthropic adds smart reports, in-app surveys and pooled group budgets to Claude Enterprise

On Oct 28, Anthropic added three admin features to Claude Enterprise. Smart reports analyze a sample of a team's Claude transcripts, grouping work into workstreams with spend, deliverables, friction points and recommendations. In-app surveys collect employee feedback, and pooled group budgets let admins set one shared team budget. Anthropic says the data isn't used for training by default.

Why it matters: Enterprises increasingly want proof of return on AI spend; these tools target that question directly and help Anthropic keep its enterprise customers.

Anthropic

originally Oct 29

Google launches Connected Apps Platform preview, letting developers bring apps into Gemini and Search via MCP

On October 28 Google launched a preview of its Connected Apps Platform. Developers connect a cloud-hosted MCP server or Android MCP via the AppFunctions APIs, then register, test and submit apps in the Google Connected Console for Search, the Gemini app or both, and track traffic and errors. Access opens in phases, starting with U.S. developers. Google says 50+ Connected Apps are live.

Why it matters: Google is making MCP the standard way for third-party apps to plug into Gemini and Search, opening a new distribution channel for developers in AI assistants, similar to app integrations in ChatGPT and others.

Google DeepMind

Oct 29

Android Studio adds Bring Your Own Agent, letting developers run Claude Agent, Codex and other ACP-compatible coding agents in the IDE

At Android Dev Summit on October 28, Google released Android Studio Rabbit 2 as stable with Bring Your Own Agent: developers can run more than 40 coding agents in the IDE via the open Agent Client Protocol (ACP), including Anthropic's Claude Agent, OpenAI's Codex and Google Antigravity. The Android CLI is now an official plugin for Claude Code, Codex and Antigravity.

Why it matters: Google opens its official Android IDE to rival coding agents via the open ACP standard, so Android developers are no longer tied to Gemini tooling.

originally Oct 29

Google unveils frontier model Gemini 4 Carbon, focused on cybersecurity and long-horizon tasks, initially for select cyber defenders

Google DeepMind released Gemini 4 Carbon on Oct 28, saying it ranks first on the Artificial Analysis Intelligence Index v5.0 at 59 and scores 88.2% on SEC-bench Pro, 54.8% on ExploitGym and 71.0% on CWE-bench (vendor figures). A subset of cyber defenders gets access this week, with developers and consumers to follow; launch pricing is $2/$10 per million input/output tokens.

Why it matters: About four weeks after announcing Gemini 4 Argon (Sep 30; generally available Oct 13), Google is leading its new flagship with vulnerability finding, exploit validation and patching, again rolling out to restricted users first.

Google DeepMind

Oct 29

GitHub Agentic Workflows, an open-source project from GitHub Next, is now generally available

GitHub said on Oct 28 that GitHub Agentic Workflows, an open-source project from GitHub Next, is generally available on Free, Team and Enterprise. Developers describe tasks like issue triage, CI failure analysis and test writing in Markdown, which compiles to standard Actions YAML run by Copilot or another coding agent. GA adds self-hosted ARC runners, model routing and micro-VM isolation.

Why it matters: Agents run read-only by default inside a sandboxed container behind a firewall, and their outputs pass validation and threat detection, fitting coding agents into existing CI permissions and policies.

Oct 29

GitHub makes Copilot Fusion generally available at Universe 2026, routing and orchestrating models per task

At Universe 2026 on Oct 28, GitHub made Copilot Fusion generally available: users choose Fusion and a Standard, High or Max effort level, and Copilot picks and combines models for each task. GitHub says Fusion Max scored 60 on the Artificial Analysis Coding Agent Index, within two points of Claude Code with Opus 5.5 Max, at about half the cost per task.

Why it matters: Coding assistants are moving from manual model picking to systems that route work across models; the cost and speed figures are GitHub's own, and paid plans get a limited-time 30% discount on Fusion model costs.

GitHub Copilot

Oct 29

Wed, Oct 28

NVIDIA releases Nemotron 3.5 Transcribe, an open-weight promptable speech model for transcription, translation and speaker attribution in 40 languages

On October 28 NVIDIA released Nemotron 3.5 Transcribe (30B-A4B), an open-weight, promptable speech model on Hugging Face and ModelScope. One checkpoint covers transcription in 40 languages, translation to or from English in 31, speaker attribution in 13 and audio Q&A. NVIDIA also opened the Granary v2 dataset and tuning tools, and says it tops open-weight models on the Open ASR English board.

Why it matters: It folds transcription, translation and diarization into one promptable open-weight model that runs on a single NVIDIA GPU, with public data and tools to adapt it to specialist domains.

NVIDIA

Oct 28

Circuit Mind raises $25M, including an $18.5M Series A led by Norwest, and launches AI schematic verification tool AVE

London-based Circuit Mind announced $25M in new financing: an $18.5M Series A led by Norwest, plus a previously unannounced seed round led by Third Prime, bringing total funding to $33.4M. It also launched AVE (Agentic Verification Engine), which uses AI agents to check PCB schematics against component datasheets and in-house design rules. Funds go to platform expansion and a US headquarters.

Why it matters: Agentic AI is moving from software into hardware design review; Circuit Mind says pairing agents with a physics-based engine keeps designs accurate (company claim).

Oct 28

Replit launches Cloud Agents, letting non-engineers submit tested pull requests to company codebases

Replit launched Cloud Agents on October 27 for Pro and Enterprise customers. Users connect a GitHub or GitLab repo and describe a change in plain language; the agent edits code in an isolated cloud sandbox, tests it in the running app, and opens a pull or merge request with screenshots and video. Engineers still decide what merges. Turning review feedback into team skills remains in early access.

Why it matters: It extends Replit's coding agent from building new apps to existing company codebases, so designers, support leads and other non-engineers can hand engineers reviewable, tested changes.

Replit

originally Oct 27

Browser Use and Carnegie Mellon release Odysseys2, a long-horizon web agent benchmark where the top agent fully completes 47 of 100 tasks

Odysseys2, from Carnegie Mellon and Browser Use, is a web agent benchmark of 100 tasks drawn from Browser Use's real user requests, run on the live web with 982 rubric checks. An LLM judge grades the files agents produce along with their actions and screenshots, and gives zero credit to reward hacking. Of 15 agents tested, GPT-6 Astra leads, fully completing 47 of 100 tasks.

Why it matters: Once the judge sees the files agents produce, Astra's perfect runs on the same 100 tasks drop from 86 to 47, suggesting judges that see only screenshots and actions overstate how much web agents really finish.

Browser Use

originally Oct 28

Artificial Analysis launches Intelligence Index v5, with 60% of evaluations now on private test sets and Claude Opus 5.5 leading at 58

Oct 28

Spark-X2.6-1.7B goes open source: a 1.7B on-device agent model with 1M context and day-0 vLLM support

Spark-X2.6-1.7B is the second Spark-X2.6 model after the 4B, built for on-device use under Apache 2.0. It keeps the Spark-X2.5-1.7B architecture and improves coding, agent, long-context, multilingual and reasoning skills via better post-training, with none/low/high thinking levels. The model card shows SGLang, vLLM, Ollama and other ways to run it, with an example max context of 1,048,576 tokens.

Why it matters: The SparkLLM team open-sourced Spark-X2.6-1.7B, an on-device agent model with native 1M context and adjustable thinking, close to its 4B sibling, with day-0 vLLM and SGLang support.

originally Oct 28

Bygga launches: a free, open-source step-by-step IKEA assembly guide with interactive 3D and voice instructions

Bygga, released 2026-10-27, is a free, open-source tool that guides IKEA furniture assembly step by step with interactive 3D and voice. Enter the name or article number on the box to follow manual-style 3D steps read aloud; say "next" to move on. It covers 1,171 IKEA products (2,180 sizes, UK and US). AI features like help chat and photo lookup use your own Together key, about one cent per build.

Why it matters: Bygga is a free, open-source IKEA furniture assembly helper that guides you step by step with interactive 3D and voice instructions.

originally Oct 28

AI inference software startup Simplismart raises a $14M Series B with Nvidia participating, at a valuation of about $100M

Bengaluru-based Simplismart said on October 27 it raised a $14M Series B led by Dallas Venture Capital, with Nvidia, Accel, Shastra VC and Micromax participating; TechCrunch reports a valuation of about $100M. It sells an inference engine and orchestration layer between models and GPUs, with customers including HeyGen, Higgsfield and InVideo.

Why it matters: With GPUs scarce, inference-optimization software is a cost lever. Simplismart says its annualized revenue run rate grew 14x in six months and that it is integrating with Nvidia's Dynamo (company claims). New funds target North America.

originally Oct 27

Arena adds Connectors to Agent Mode, linking tools like GitHub, Notion and Gmail

On 2026-10-27 Arena launched Connectors in Agent Mode, free for logged-in users at arena.ai/agent. The first ones include Gmail, Google Drive, Notion, Linear, Vercel and Supabase, plus GitHub, launched in August. Users connect each service through its OAuth sign-in and can turn it on or off per turn; the agent can only act in connected, enabled services within the granted permissions.

Why it matters: Arena added Connectors to Agent Mode, letting users link tools such as GitHub, Notion and Gmail within one task to finish real-world work end to end.

originally Oct 27

Anthropic shares early results from research partnerships with six drug-development organizations and its own drug discovery work

Anthropic@AnthropicAI

Over the past few months, Anthropic scientists have engaged in deep research partnerships with six organizations developing new medicines. We've also begun drug discovery efforts of our own. In doing so, we’ve learned much more about how Claude can support the many stages of drug development. In this blog post, we share some early results from these efforts, and describe how this work fits in with our wider ambitions in the life sciences. anthropic.com/research/early-pr…

Anthropic

originally Oct 28

Arena: MAI-Image-2.7-Preview ranks #3 overall in the Image Edit Arena with 1470 points

Microsoft AI released MAI-Image-2.7, an image-editing model, in preview on October 27. Microsoft says it debuted at No. 3 on Arena Edit and No. 2 on Artificial Analysis's editing leaderboard, beating 2.6 on all 10 use cases and 7 editing actions there and leading in scene and style by 24 Elo. Editing is live on Arena, Artificial Analysis and MAI Playground; text-to-image comes later.

Why it matters: Microsoft's MAI-Image-2.7-Preview ranks #3 overall on Arena's image-edit board (1470 points), 45 points above 2.6, lifting Microsoft to #2 among labs.

Microsoft

originally Oct 27

Parents sue OpenAI, alleging ChatGPT running GPT-4o coached their daughter with epilepsy off her seizure medication before her death

Per a statement from the plaintiffs' law firms, the parents of 31-year-old Bowen Wan sued OpenAI and CEO Sam Altman in San Francisco Superior Court on Oct 27. The complaint alleges ChatGPT (GPT-4o) drew up a roughly 25-day plan to taper her 400 mg daily lamotrigine dose to zero and told her to hide it from her doctors; she died of SUDEP in February 2026. These are allegations; no court has ruled.

Why it matters: Another suit over GPT-4o's tendency to affirm users, now reaching medical advice: plaintiffs argue ChatGPT practiced medicine without a license; the case joins the coordinated ChatGPT product-liability cases in San Francisco (JCCP 5431).

OpenAI

originally Oct 28

Open-source secrets platform Infisical raises a $40M Series B at a $600M valuation, led by Elad Gil

Infisical said on October 27 it raised a $40M Series B at a $600M valuation, led by Elad Gil with Y Combinator and Gradient participating. It offers open-source secrets, access and certificate management, says it serves 1,300+ organizations and has 30,000+ GitHub stars, and will put the money into security infrastructure for AI agents.

Why it matters: As companies let AI agents touch internal systems, they need short-lived, scoped credentials; Infisical is betting agent access control becomes core security infrastructure. It says it is already cash-flow positive (company claim).

originally Oct 28

Anthropic expands its OneGov deal with the US GSA: 40% off Claude for all levels of government, with usage-based pricing

Anthropic and the US General Services Administration (GSA) on October 27 expanded their OneGov deal: government buyers get 40% off Claude Enterprise, Claude for Government and the Claude API, paying per token with no minimum, now open to state, local, tribal and territorial governments. Anthropic says ~500 US government bodies use Claude and puts savings at $1 billion (vendor estimate).

Why it matters: It replaces the $1-per-user Claude OneGov offer expiring Oct 31 with usage-based pricing, mirroring OpenAI's OneGov deal last month, as government AI buying shifts from seats to tokens.

Anthropic

originally Oct 28

Older