Classified

Q Almost Daily Briefing

Q Branch — AI Intelligence Summary 2026-05-11

Sit down. Several organisations have decided the thing missing from AI was professional services. Naturally, there is also more.

OpenAI Launches Its Deployment Company, After Anthropic’s Services Venture2026-05-11

OpenAI launched a majority-owned Deployment Company with more than $4 billion in initial backing from consulting, systems integration, and investment partners, and agreed to acquire Tomoro for its forward-deployed engineering staff. Anthropic, meanwhile, announced its own AI services company with Blackstone, Hellman & Friedman, Goldman Sachs, and others on 2026-05-04. The labs have discovered that selling intelligence is easier when you also sell the people who know where to put it. OpenAI Anthropic

Claude Platform Is Now Generally Available on AWS2026-05-11

Anthropic made Claude Platform on AWS generally available, giving AWS customers native Claude Platform features with IAM authentication, CloudTrail logging, AWS billing, commitment retirement, Managed Agents, code execution, web search, Files API, Skills, MCP connector, prompt caching, citations, and batch processing. For enterprise teams, this lowers the procurement and governance friction around Anthropic’s agent platform without pretending procurement friction is spiritually meaningful. Anthropic

Cursor 3.3 Adds Native PR Review, Parallel Agents, and Split-to-PR2026-05-07

Cursor 3.3 shipped an integrated PR review workflow, a Build in Parallel mode that dispatches independent plan tasks as async subagents, and a quick action to split staged changes into logical PRs. The editor is absorbing more of the review and branch-management workflow, which is precisely where coding agents become less like chat boxes and more like operating environments. Cursor

OpenAI Releases Three Realtime Audio Models for Voice Agents2026-05-07

OpenAI introduced GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Whisper for live voice agents, speech translation, and streaming transcription. GPT-Realtime-2 adds GPT-5-class reasoning, 128K context, adjustable reasoning effort, parallel tool calls, and better recovery behaviour, which makes voice agents rather less like a parlour trick and rather more like infrastructure. OpenAI

Claude Code Adds Agent View and Goal-Driven Runs2026-05-11

Claude Code v2.1.139 added an agent view for managing multiple background sessions and a /goal command that lets Claude continue until a completion condition is met. A coding assistant that can keep several tasks moving without requiring constant supervision: dangerous, useful, and therefore inevitable. GitHub Anthropic

Testaify 2.0 Expands Autonomous QA for AI-Era Release Validation2026-05-11

Testaify released version 2.0 with a portfolio dashboard, visual workflow mapping, configurable test depth, AI learning between sessions, CI/CD API integration, release thresholds, replay, and developer-focused reports. As code generation accelerates, autonomous validation is becoming less optional and more like the bill arriving at the table. PR Newswire

Nvidia’s AI Equity Commitments Reportedly Top $40 Billion in 20262026-05-09

CNBC reported Nvidia has committed more than $40 billion to AI-related equity investments this year, including a $30 billion OpenAI investment and new commitments to Corning and IREN. Supplier financing, customer financing, infrastructure financing — at some point the circular economy stopped being a metaphor. CNBC TechCrunch

Anthropic Adds SpaceX and Akamai Capacity as Claude Code Limits Rise2026-05-08

Anthropic announced a SpaceX capacity deal adding more than 300 megawatts and over 220,000 NVIDIA GPUs, and Bloomberg reported a seven-year Akamai cloud deal worth $1.8 billion. Claude Code users are already seeing the practical side: higher five-hour limits and the retirement of peak-hours reductions. Capacity planning, it turns out, is still a product feature. Anthropic CNBC

Baidu Releases ERNIE 5.12026-05-08

Baidu released ERNIE 5.1, a more efficient successor to ERNIE 5.0 with substantially reduced total and active parameters, lower pretraining cost, and reported strength across search, reasoning, agentic, creative writing, spreadsheet, and exam benchmarks. It is a reminder that frontier-model competition is not confined to the usual American dashboard tabs. Baidu

Anthropic Brings Claude to Microsoft 365 and Outlook2026-05-07

Claude for Excel, PowerPoint, and Word is now generally available, while Claude for Outlook is in public beta for paid plans. Anthropic is pushing Claude into the documents, spreadsheets, email, and calendars where enterprise work actually happens, which is less glamorous than a demo and considerably more important. Anthropic

AWS Launches Agent Toolkit and Makes Its Managed MCP Server Generally Available2026-05-06

AWS launched Agent Toolkit for AWS with more than 40 validated agent skills, managed MCP infrastructure, IAM controls, CloudWatch and CloudTrail observability, sandboxed execution, and current service guidance. This is AWS trying to make coding agents safer on AWS, a problem category whose existence will surprise precisely no one who has let an agent near IAM. AWS AWS Blog

GitHub Copilot in VS Code Gains Semantic Search, Terminal Access, BYOK, and Browser Context2026-05-06

GitHub’s Copilot update for VS Code added semantic search across workspaces and connected GitHub orgs, inline diff previews in chat, open-terminal read/write access, browser tab sharing, BYOK support, and an experimental /chronicle command for querying session history. Copilot is being wired closer to the actual working state of the developer, which is where assistants become useful and occasionally presumptuous. GitHub

Atlassian Expands Rovo and Launches Product Collection2026-05-06

Atlassian made Rovo Studio generally available, added governance and Teamwork Graph access through Rovo MCP Server and CLI, and introduced Product Collection, an early-access product operating system combining Jira Product Discovery, AI-powered feedback, Rovo, and product analytics. The company is trying to turn scattered tickets, calls, surveys, Slack messages, and usage data into product decisions. Noble, difficult, and overdue. Atlassian Atlassian

Simon Willison: Vibe Coding and Agentic Engineering Are Blurring2026-05-06

Simon Willison argued that the line between irresponsible “vibe coding” and professional agentic engineering is eroding as coding agents become reliable enough that even experienced engineers stop reviewing every generated line. It is one of the more honest practitioner pieces on the current methodology gap: generation is improving faster than review culture. Simon Willison

GPT-5.5 Instant Becomes ChatGPT’s Default Model2026-05-05

OpenAI replaced GPT-5.3 Instant with GPT-5.5 Instant as ChatGPT’s default model, claiming fewer hallucinated claims on high-stakes prompts, better image analysis, stronger STEM performance, improved web-search decisions, and memory-source controls. Default model changes matter because most users never choose the model. The defaults choose them. OpenAI System Card

Claude Managed Agents Gets Dreaming, Outcomes, and Multiagent Orchestration2026-05-05

Anthropic added Dreaming, Outcomes, and Multiagent Orchestration to Claude Managed Agents: scheduled self-review sessions, grader agents that retry against developer-defined rubrics, and lead agents that decompose work across specialist subagents. Netflix is already using orchestration in production. The name "Dreaming" is doing quite a lot, but the platform direction is serious. Anthropic Ars Technica

US Expands Pre-Release Frontier Model Evaluations2026-05-05

NIST's Center for AI Standards and Innovation signed pre-deployment evaluation agreements with Google DeepMind, Microsoft, and xAI to test frontier models for national security and cybersecurity risks before public release. OpenAI and Anthropic also renegotiated earlier agreements. An administration that began by dismantling AI safety infrastructure has arrived at the conclusion that knowing what a model can do before releasing it is useful. Frontier capability research does have a way of concentrating minds. NIST CNBC

SAP Acquires Prior Labs and Commits More Than €1 Billion to Tabular AI2026-05-04

SAP agreed to acquire Prior Labs and invest more than €1 billion over four years in frontier AI for structured enterprise data: tables, ledgers, procurement records, finance documents, and other formats where business value inconveniently refuses to become prose. If tabular foundation models work as advertised, this is one of the more credible enterprise AI bets. SAP TechCrunch

Sierra Raises $950 Million at a Valuation Above $15 Billion2026-05-04

Sierra raised $950 million led by Tiger Global and GV, saying it now serves more than 40% of the Fortune 50 and powers billions of customer interactions. Enterprise customer-service agents remain one of the clearest places where AI moves from “interesting” to line-item economics, which is why the money continues to arrive wearing expensive shoes. Sierra TechCrunch

That is everything worth your attention. The rest can remain where it belongs: in someone else’s slide deck.