Rogue Agents

Episode 11: Consultancy Capture, the Values Premium, and the Death of the Operating System

On May 14, PwC signed a deal to train 30,000 consultants on Claude. The Big Four firm whose entire job is implementing software just licensed the model that's coming for that job. Same day, Anthropic and the Gates Foundation announced a $200M global health partnership. Two days earlier, Bloomberg said Anthropic's in talks at a $950B valuation. Day before that, Google said Android is no longer an operating system. Call it the week the model providers stopped pretending they only sell models. This week on Rogue Agents — three big stories, five quick takes: Three Deep Dives:  1. Anthropic's Enterprise Triple Play — Claude for Small Business, the Gates Foundation deal, and PwC all inside 36 hours. The model provider absorbs the implementation layer. We're calling it Consultancy Capture.  2. Pentagon Excludes Anthropic (Mark Hinkle joins) — The War Department's eight-company AI deal labels Anthropic a "supply chain risk" for refusing "all lawful purposes" language. Mark unpacks the Red Hat parallel and the operating cost of saying no.  3. Google's Gemini Intelligence — Sameer Samat says Android is moving from an operating system to an intelligence system. Plus the Googlebook reveal. Build your MCP server. Agent SEO is your new acquisition channel. Five Shorts: OpenAI ships GPT-5.5-Cyber to EU regulators · Anthropic's $950B valuation is really a hyperscaler re-up · Wall Street pours PE money into model-provider consulting arms · the Trump admin starts testing Google, Microsoft, and xAI models · Googlebook lands in K-12 and your kid's homework just changed. Read the full stories at theaienterprise.io. Subscribe at rogueagents.io. — The AIE Network — theaie.net  The AI Enterprise Newsletter — theaienterprise.io  AI CIO — aicio.ai  AIOS Training — theaios.ai  All Things AI Conference — allthingsopen.ai

Episode 10: The Round-Trip, Wall Street's New Consultant, and Vertex Dies on a Tuesday

On May 5, Anthropic committed to spend $200 billion with Google over the next five years. One week earlier, Google committed to invest up to $40 billion back into Anthropic. Five-to-one. Money out is five times money in. That's not a deal — that's a closed loop. Call it the Round-Trip. A note from Mark: Episode 10 was supposed to ship last week. Our publishing pipeline went sideways and the audio never made it to the feed. There was no Episode 10. There was just silence. This is the makeup — we're shipping the May 4–9 news week as Episode 10 so the catalog stays clean. This week on Rogue Agents — three big stories, five quick takes: Three Deep Dives: 1. The Round-Trip — Anthropic and Google just locked themselves into a captive supplier arrangement that makes the OpenAI-Microsoft divorce look easy. 2. Wall Street's New Consultant (Mark Hinkle joins) — Anthropic shipped 10 finance agents on Claude Opus 4.7 the day after launching a $1.5B services JV with Blackstone, Hellman & Friedman, and Goldman Sachs. McKinsey has 90 days to decide what it does next. 3. Vertex Is Dead — Google killed Vertex AI on a Tuesday and replaced it with the Gemini Enterprise Agent Platform. Workspace Studio ships no-code agents to 3 billion seats. The era of "pick your model" is ending. Five Shorts: Anthropic absent again from the CAISI testing deal · ServiceNow's nine-second production database deletion and the kill switch · the agentic wars heating up across five vendors · Anthropic's quiet move into offensive cybersecurity (Project Glasswing) · the AI roll-up era your acquirer is already pricing. This is the first episode in our new format — three Deep Dives interleaved with five Shorts. Tell us if it works. Read the full stories at theaienterprise.io. Subscribe at rogueagents.io. — The AIE Network — theaie.net The AI Enterprise Newsletter — theaienterprise.io AI CIO — aicio.ai AIOS Training — theaios.ai All Things AI Conference — allthingsopen.ai AI Toolbox — theaitoolbox.ai

Episode 9: The Pentagon Veto, the Azure Divorce, and Wall Street's Anthropic Math

On May 1, the Pentagon signed classified-network AI contracts with 8 companies — SpaceX, OpenAI, Google, NVIDIA, Microsoft, AWS, Reflection, Oracle. Anthropic was excluded by name for refusing the lawful-purpose clause. On April 27, OpenAI and Microsoft formally dismantled the cloud exclusivity that bound OpenAI to Azure for seven years. And Fortune ran the math: half of Google's and Amazon's blowout Q1 AI profits came from their Anthropic stake — not operating revenue. Three stories splitting the AI economy along three fault lines: principles, platforms, and paper. Three Deep Dives, four Quick Hits, Mark Hinkle joins for the Anthropic math. 1. The Pentagon Veto — Anthropic's AUP refusal as the open source moment for AI vendors. GPL vs BSD as the philosophical parallel. Audit your vendor's TOS for five phrases by Friday. 2. The Azure Divorce — multi-cloud AI is now real. Inventory your OpenAI API spend, get a Bedrock quote, take it into your Azure renewal. Microsoft account economics in 2026 reward the move. 3. Wall Street's Anthropic Math — Yahoo Japan, ICG, SoftBank Vision Fund, and the Berkshire asterisk. Mark's discipline frame: separate operating AI revenue from valuation marks on AI investments. Three eras, same physics. Quick Hits: Mandiant's prompt-injection-via-public-web warning, Avoca's $1B vertical-agent unicorn (for plumbers), Microsoft Copilot Agent Mode GA in Word/Excel/PowerPoint, and Anthropic's $50B raise at $850-900B valuation. Read the full stories at theaienterprise.io. Subscribe at rogueagents.io. — The AIE Network — Everything in one place: theaie.net The AI Enterprise — Newsletter: theaienterprise.io AI CIO — AI for IT Leaders: aicio.ai AIOS — AI Training & Courses: theaios.ai All Things AI — Conference: allthingsopen.ai AI Toolbox — Recommended tools: theaitoolbox.ai

Episode 8: GPT-5.5 Lands, Copilot Goes Agent, and the 80K Layoff Receipts

Three next moves, three Deep Dives, four Quick Hits. OpenAI ships GPT-5.5 and doubles the price. Microsoft puts Copilot Agent Mode in every Office app. Adobe rebrands its entire Experience Cloud around AI agents. Anthropic admits Claude got dumber for six weeks. And the Q1 2026 layoff data finally caught up with the hype — 80,000 tech jobs gone, 48% attributed to AI. This week on Rogue Agents: 1. The frontier model fight — GPT-5.5 vs Claude 4.7 vs DeepSeek, why the routing layer is now the most valuable software in your stack 2. The agent layer goes production — Microsoft Copilot Agent Mode GA, Adobe CX Enterprise Coworker, MCP and A2A as the new LAMP stack 3. The labor math catches up — 80K Q1 layoffs, Goldman's 16K-jobs-per-month run rate, why freezing entry-level hiring now creates a 2031 leadership-bench problem (Mark joins this one) Plus four Quick Hits: DeepSeek's new flagship, AMEX acquires Hyper, the Google Cloud + NVIDIA superstack, and the Anthropic quality bug nobody wants to talk about. Mark closes with field notes from Raleigh-Durham Startup Week and a one-on-one with Erik Troan, founder/CTO of Pendo, on the new moat in a vibecoding era. Read the full stories at theaienterprise.io. Subscribe at rogueagents.io. — The AIE Network — Everything in one place: theaie.net The AI Enterprise — Newsletter: theaienterprise.io AI CIO — AI for IT Leaders: aicio.ai AIOS — AI Training & Courses: theaios.ai All Things AI — Conference: allthingsopen.ai AI Toolbox — Recommended tools: theaitoolbox.ai

Episode 7: Opus 4.7, Stanford Says 66%, and the Agent Sprawl Nobody's Fixing

This week on Rogue Agents: Anthropic dropped Claude Opus 4.7 on every major cloud on day one — and quietly used a model called Mythos to find thousands of previously unknown zero-day vulnerabilities in every major operating system and browser. Bruce Schneier called it the end of a twenty-year cybersecurity equilibrium. Meanwhile, Stanford says AI agents jumped from 12% to 66% on real computer tasks in twelve months. And 94% of enterprises running agents admit they don't know what those agents are actually doing. Three stories. Four altitudes. Same verdict: the capability is here. The governance gap is bigger than the deployment gap. 👉 Subscribe, rate, and share: rogueagents.io [https://rogueagents.io] Read the full stories * Claude Opus 4.7 drops on every major cloud day-one [https://theaienterprise.io] — The AI Enterprise * Project Glasswing and the end of the 20-year cybersecurity equilibrium [https://theaienterprise.io] — The AI Enterprise * Stanford 2026 AI Index: agents hit 66% on OSWorld [https://theaienterprise.io] — The AI Enterprise * OutSystems: 96% of enterprises run agents, 94% can't govern them [https://theaienterprise.io] — The AI Enterprise Quick Hits this week * OpenAI's GPT-5.4-Cyber program opens invite-only access backed by $10M in API credits * Gartner projects 40% of enterprise applications will have task-specific AI agents by year-end * Apple quietly ships enterprise agent support in iOS 26 with on-device private compute * Meta drops a new open-weights Llama mid-tier with built-in tool use for agentic workflows Learn more about AI for business * The AI Enterprise [https://theaienterprise.io] — Newsletter * AI CIO [https://aicio.ai] — AI for IT Leaders * AIOS [https://theaios.ai] — AI Training & Courses * All Things AI [https://allthingsopen.ai] — Conference * The AIE Network [https://theaie.net] — Everything in one place * AI Toolbox [https://www.theaitoolbox.ai/] — Recommended tools The reporting is human. The delivery is AI. rogueagents.io [https://rogueagents.io]