AI in the News — Weekly Brief

Anthropic grew revenue 80x in a single quarter, and about 80% of it is enterprise revenue — while OpenAI, still chasing consumer scale, loses roughly $1.22 for every dollar it earns. The market just voted: enterprise-safe AI is beating consumer hype.

LinkedIn post or short video: “The AI company betting on businesses just became more valuable than the one betting on chatbots.” Use the 80%-enterprise-revenue stat as the opening slide of a workshop on where real AI ROI actually lives.

SoftBank's cost to borrow for this bet jumped from 6.7% to 8.2% in nine months — that's the market pricing in real doubt about the AI trade, even while the headlines still say ‘trillion-dollar valuation.’

Short video breaking down ‘cost of capital’ in plain English for execs, paired with a 5-question AI-vendor financial-health checklist as a lead magnet or workshop handout.

When chip and model exports sit on the same table as tariffs and rare earths, ‘which AI vendor should we use’ stops being an IT question and becomes a geopolitical-risk question.

Workshop exercise: a short ‘AI vendor geopolitical exposure’ scenario map (US-based vs. China-linked dependencies) execs fill in for their own stack. Doubles as a LinkedIn carousel: ‘3 questions to ask before you pick your next AI model provider.’

The AI spending story just flipped — businesses aren't buying less AI, they're buying it 41% cheaper, and the smart ones are pocketing the difference instead of chasing the newest frontier model.

LinkedIn post or client email: ‘Your AI Bill Should Be Going Down.’ Use the per-employee figure to challenge SMB clients who assume more spend equals more sophistication; pair with a 3-question overpayment audit.

Siemens went from qualifying leads by hand to engaging 100% of inbound leads with AI — that's not a productivity tweak, that's a headcount-shaped hole where triage used to be.

Short video breaking down the two-agent handoff (engagement agent to qualification agent) as a replicable mid-market pattern — most SMB leaders can't deploy Siemens-scale tech, but the workflow design is a one-hour whiteboard exercise.

In 2019 a small business had maybe 70 ways to hire for AI. Today it has 600 — the skills gap isn't coming, it's already showing up in your job postings.

Workshop exercise: have execs list their current job titles, then mark which ones now implicitly require an AI-fluency layer. Strong hook for a post about hiring for roles that did not exist three years ago.

Every AI rollout I've seen fail wasn't a tools problem — it was a manager who never said out loud that using it was allowed. Gallup just put a number on that: 8.7x.

LinkedIn post or keynote clip: ‘Your AI Strategy Is a Management Strategy.’ Use as the opening hook for a workshop module on manager enablement.

AI in wealth management just stopped being a bolt-on chatbot — it's wired into custody, CRM, and portfolio systems for 16,000 RIAs at once. If your firm doesn't have an AI governance policy yet, your custodian just made that urgent.

LinkedIn post: ‘AI in wealth management just went mainstream.’ Pair with the Vanguard time-reclamation stat below. Strong opener for a wealth-management workshop module on what to ask your custodian about AI.

72% of advisors say they want AI to buy back time for prospecting and relationships — not to replace them. Compliance uncertainty, not fear of the tech, is the #1 reason it's not moving faster.

Short video or live workshop poll: ‘What's stopping you — the tool, the training, or the compliance sign-off?’ Surfaces the real objection in the room fast.

Only 6% of insurance customers trust AI alone — but 61% want their human agent using it. Clients aren't asking you to automate yourself out of the relationship, they're asking you to get faster and more precise inside it.

LinkedIn carousel: three stats that kill the ‘AI will replace my agents’ myth, using the 87% / 6% / 61% breakdown as three slides. Strong keynote opener for an insurance-vertical talk.

The insurance AI story just moved past ‘assistant that drafts an email’ to ‘agent that owns the whole renewal file end-to-end’ — and Patra backing that with E&O coverage on the AI's own work is the real signal of confidence.

Workshop exercise: give execs a real back-office workflow such as renewal processing and have them map task-level AI vs. full-workflow AI, using Patra as the case study for what risk controls should come with full-workflow automation.

YC isn't asking founders to build a better compliance dashboard — they're asking founders to replace the compliance team. If the smartest capital in startups thinks that headcount is automatable, assume your vendors are already building toward it.

LinkedIn post: ‘YC just told a room full of founders to go build software that replaces your compliance department. Here's what that means if you run one.’ Pair with a 60-second video walking through three compliance tasks that are now fundable startup ideas.

When venture capital funds two direct competitors selling insurance for AI agents in the same season, that's the market pricing in the fact that agents will make mistakes — and someone will need to pay for it.

Workshop exercise: ‘The Agent Liability Audit’ — walk through every AI agent or automation a client has deployed and ask who's liable if it's wrong, and whether that's covered. LinkedIn hook: ‘Two new startups just got funded to insure your AI agents.’

Everyone assumed AI would hit underwriters first because that's the ‘data job.’ YC is funding the opposite thesis — that the relationship-driven broker desk is the more valuable place to automate first.

LinkedIn carousel or talk clip: ‘The job everyone said AI couldn't touch is the one VCs just funded four startups to automate.’ Use as a bridge to any relationship-based sales role — brokers, account managers, advisors — that needs a readiness plan now.

You are not trying to compete with AI in producing outputs — that's a losing game. The gap isn't in the AI's capability, it's in how little of that capability most people and companies are actually using.

Build a short talk clip or workshop self-audit titled ‘The Capability Overhang’: for each of Mollick's four human advantages — deep knowledge, wide knowledge, taste, agency — ask where you have it and whether you're using AI to amplify it or just asking it to do your job.