Code with Claude London Lands Today — Day-1 Keynote Streams Live Across Three Tracks, With an Extended Edition for Indie Builders and Founders
Anthropic’s developer event arrives in London, and the Day-1 keynote plus breakout sessions livestream today. The agenda is built around three tracks: Research (what current models can do and where they’re going), Claude Platform (how teams ship production-grade agents), and Claude Code (running long-horizon, multi-repo, parallel-agent work at scale). It’s the European stop on the 2026 tour that opened in San Francisco on May 6 and continues to Tokyo in June.
The framing matters: this year’s tour has been about agent infrastructure, not a headline model launch. San Francisco shipped or previewed managed multi-agent orchestration, Claude Code routines, a new Advisor tool, Remote Agents, and CI auto-fix — so London is the venue to watch for the next layer on each, plus any GA promotions. A separate “Extended London” track is pitched at independent developers and early-stage founders who want hands-on workshops and office hours rather than enterprise keynotes. If you build on Claude, the livestream is the thing to have open today.
KPMG and Anthropic Sign a Global Alliance — Claude Goes Into Digital Gateway Across 276,000 Staff, and Anthropic Becomes a Preferred Private-Equity Partner
KPMG announced a global alliance with Anthropic that embeds Claude directly into Digital Gateway, its AI-enabled client-delivery platform. The deal gives KPMG’s 276,000-person workforce access to the Claude suite and lets clients build agentic workflows in real time. It builds on two years of Claude adoption inside KPMG’s Advisory, AI and Data Labs, and U.S. enterprise teams.
The initial focus is tax clients and private-equity firms — and notably, Anthropic is naming KPMG a preferred partner for private equity, with the two committing to build new Claude-powered products for PE portfolio companies. The alliance also folds in cybersecurity, risk, and AI-assurance work. It slots into a steady run of enterprise stamps this month (PwC, the legal vertical, small business), and the consistent shape is the same: the model layer gets monetized through deep partner integration, not just front-end chat.
Anthropic Says Claude Stays Ad-Free — A Public Stance That Ad Incentives Are Incompatible With a Genuinely Helpful Assistant
Anthropic published its position that Claude’s products will remain ad-free, arguing that advertising incentives pull against the goal of a genuinely helpful assistant and erode user trust. The post lays out how the company plans to widen access without leaning on ad monetization.
It’s a deliberate contrast play as other consumer AI products experiment with sponsored placements. Worth a careful read on the wording: the commitment is framed around Anthropic’s own products, and the company is explicit that it doesn’t prevent developers building on Claude from running ads in their own apps. For anyone evaluating Claude for customer-facing deployments, it’s a useful trust signal to cite.
Anthropic Acquires Stainless — the SDK and MCP Engine Behind Every Official Claude Library; Reported Above $300M, Hosted Products Wind Down
Anthropic is acquiring Stainless, the company that has powered the generation of every official Anthropic SDK since the earliest days of the Claude API. Founded in 2022, Stainless turns an API spec into SDKs across TypeScript, Python, Go, Java, and more — and crucially also generates MCP servers, the connectors that link AI agents to external APIs. The deal has been reported at more than $300 million. The team moves to Claude Platform to push developer experience and agent connectivity.
Two things to act on. First, the wind-down: Stainless says it will shut down its hosted products, including the hosted SDK generator, as the team refocuses on Claude Platform. Existing customers keep the right to modify and extend SDKs they’ve already generated — so if you depend on the hosted service, plan your migration now rather than later. Second, the strategic read: Stainless listed OpenAI, Google DeepMind, Perplexity, Groq, and Cloudflare among its customers. Anthropic created MCP, and now owns a chunk of the tooling that competitors used to ship their own SDKs and connectors. That’s a tightening of the dev-tooling and connectivity layer, landing the same week as the London platform keynote — not a coincidence.
Claude Code Week 20 Updates — Agent View, a /goal Command That Works Until Done, Fast Mode on Opus 4.7 by Default, and “Summarize Up to Here” in Rewind
The Claude Code train kept moving through the May 11–15 window. The headline addition is Agent view: one screen for every Claude Code session, showing what’s running, what’s blocked on you, and what’s done — the kind of cockpit you want once you’re running multiple agents at once. Alongside it: a new /goal command that keeps Claude working across turns until a completion condition holds, fast mode now defaulting to Opus 4.7, and a Rewind menu option to “Summarize up to here” that compresses earlier context on demand.
The release also bundled a broad reliability pass: better background-session handling, improved model and plugin management, faster MCP and SDK startup, and a long list of fixes for terminal glitches, crashes, and session recovery across desktop and Windows. If you’ve been holding off on parallel-agent workflows because the session sprawl was hard to track, Agent view plus /goal is the combination that makes it manageable. Update and pin your version before the London builder sessions.
Claude Mythos and Project Glasswing Keep Reshaping the Security Conversation — Thousands of Zero-Days Found, Working Exploits on the First Try in 83% of Cases
Anthropic’s Claude Mythos — the frontier model built to autonomously discover and chain zero-day vulnerabilities — is still the security story the industry can’t stop talking about. Anthropic has used the preview to identify thousands of previously unknown flaws across every major operating system and browser, reproducing vulnerabilities and building working exploits on the first attempt in over 83% of cases. One example, CVE-2026-4747, is a 17-year-old unauthenticated remote-code-execution bug in FreeBSD’s NFS server, with an exploit Mythos built autonomously.
Anthropic withheld Mythos from commercial release and instead stood up Project Glasswing — a defensive coalition including AWS, Apple, Microsoft, Google, CrowdStrike, and Palo Alto Networks, with access extended to roughly 40 more organizations. The thesis analysts keep repeating: this collapses the window between vulnerability discovery and weaponization from weeks to hours, likely for good. If you own production systems, the practical takeaway is to assume that gap is closing and prioritize patch velocity accordingly. (Reminder: defensive security is the lane here — the offensive capability is exactly why it wasn’t shipped openly.)
The Vertical-Bundle Push Keeps Compounding — Small Business and Legal Connector Packs, Plus Claude Reaching Into Creative Software Like Adobe and Blender
Anthropic’s ecosystem build-out continues on the verticals it opened earlier this month. Claude for Small Business put Claude inside QuickBooks, PayPal, HubSpot, Canva, Docusign, Google Workspace, and Microsoft 365, with ready-to-run workflows for payroll, invoicing, sales, marketing, and month-end close. Claude for Legal shipped 20+ MCP connectors and 12 practice-area plugins spanning research, contracts, discovery, matter management, and legal aid.
The newer thread is creative tooling: reporting points to Claude reaching into design and 3D software like Adobe and Blender, so agents are starting to touch design files, mockups, and scripts — not just text and code. Read together, it’s the same flywheel the MCP ecosystem has been spinning all year: the value isn’t the model name on the endpoint, it’s the depth of the connector graph a team has already wired up. That’s the lock-in that’s hard for any competitor to copy quickly.
Wednesday Read — London Is a Platform Keynote, Not a Model Keynote, and the Stainless Deal Tells You Why
Step back and the day has a clear shape. Anthropic timed the Stainless acquisition for the same week as the London keynote, and that pairing is the tell. This tour has never been about a shiny new model — the current lineup is still Opus 4.7 and Sonnet 4.6, and there’s no confirmed Sonnet 4.8 despite the rumor mill. The story Anthropic is choosing to tell is about agent infrastructure: the platform, the tooling, the connectors. Buying the company that generated every official SDK and a big slice of the MCP servers in the wild — including ones competitors relied on — is a move to own the plumbing, not the headline.
The KPMG alliance is the demand-side bookend to that supply-side move. A 276,000-person firm embedding Claude into its client-delivery platform, plus a preferred-PE-partner designation, is exactly the kind of distribution that justifies investing in tooling depth over model flash. And Claude Mythos lurking in the background is a reminder that the frontier-capability story is still very much alive, just deliberately gated. If you build on Claude, today’s practical homework is short: watch the London livestream for GA promotions on the managed-agent stack, map your dependency on any Stainless hosted service and plan the migration, and assume the connector graph — not the model number — is where the durable advantage now sits.