Hey

Welcome to The AI Report — the #1 AI newsletter for business leaders.

You don’t need to be technical. Just informed.

• 1. 🤨 “What’s missing from today’s risk models?” with IBM
• 2. ⚡️ AWS ships 35% faster AI compute
• 3. 💼 Your Business Briefing
• 4. 📈 Build your perfect chatbot with NETGURU
• 5. ✍️ Today’s Policy Corner
• 6. 🗞️ The News Bulletin

Are you ready to launch AI transformation for your company?

TOGETHER WITH IBM

What’s missing from today’s risk models?

Enterprises are built to manage known risks—financial volatility, regulatory pressure, technology failure. The greater challenge is what isn’t modeled: lagging insight, hidden misalignment and systems that fail under scale. These blind spots are where AI initiatives lose momentum. For the C-suite, the priority is no longer just strategy. It’s visibility, alignment and resilience under pressure.

 Surface gaps in real-time insight across the business
 Align decision-making before scaling exposes fault lines
 Build systems that perform consistently under pressure

Leaders who address these hidden risks early are the ones who turn AI potential into sustained advantage.

Read more in C-Suite Takeaways ‘Where executive decisions meet real‑time pressure’

Latest in AI

AWS ships 35% faster AI compute

🚨 Our Report

AWS has launched new cloud servers built explicitly for agentic AI workloads. The M9g and M9gd instances, powered by Graviton5, run workloads with up to 25% better compute performance than the previous generation, with Meta already deploying Graviton at scale starting with tens of millions of cores to support its agentic AI efforts.

🔓 Key Points

AWS says Graviton5 runs AI tasks up to 35% faster and databases up to 30% faster than the previous generation. ClickHouse reported a 36% performance boost with zero code changes; Honeycomb saw 36% better throughput per core across a 6-month A/B test; HubSpot saw MySQL query duration drop by up to 60%.

The new instances also introduce a security layer that makes these the first cloud servers whose isolation has been verified with mathematical proof. Graviton5's expanded processing capacity also cuts the steps where AI agents sit waiting on the server, keeping the rest of your infrastructure moving.

🔐 Relevance

This is cheaper and faster AI compute with two levers you own. First, cost: running AI workloads 25-35% faster means lower compute spend per task. Second, risk: mathematically proven isolation clears the security bar in healthcare, financial services, and government.

Before you renew your cloud contract, have your team price your AI workloads on these new servers. M9g is now generally available in the US and Europe.

The AI Report Podcast

Microsoft and Mayo Clinic Build an AI for Healthcare + Europe Pushes Back on US Tech | AI News in 5

THE BUSINESS BRIEFING: FINANCE (powered by Upscaile)

NG.CASH, a Brazilian Gen-Z digital bank with over a million customers, was running customer support on a decision-tree platform that required constant maintenance and left the majority of inquiries unresolved. A recent acquisition scaled the customer base dramatically. The old model could not keep up. That had to change fast.

Tool used: Decagon, generative AI support agents built around Agent Operating Procedures (AOPs).

Result: Autonomous resolution rose from 13% to 70%. Absorbed a 3-4x volume spike with no added support headcount. Avoided hiring 35+ support agents.

The lesson: AI support works when your agents carry enough structured context to handle open-ended questions -- not just route them. NG.CASH's decision-tree system failed because it could only follow pre-built paths. Decagon's AOPs replaced that rigidity with agents that could resolve issues the old system would have escalated every time.

Steal this: This week, pull your support escalation data. Find the ticket categories your current system fails most often. Document the volume and cost per escalation, then use that number as your business case to pilot an AI agent on those specific categories before touching anything else.

TOGETHER WITH NETGURU

Build your perfect chatbot with custom power & SaaS speed

You don’t need to choose between SaaS & custom chatbots. 

With SaaS, you move fast. But you can't tailor the solution to your precise needs, and your data lives somewhere you don’t fully control. Custom software gives you the freedom, but building from scratch takes time and money.

We've put together something that sits in the middle: An open-source, production-ready chatbot platform you can fully customize.

THE POLICY CORNER

Florida Attorney General sues OpenAI over ChatGPT mental-health harms.

Florida AG James Uthmeier sued OpenAI and CEO Sam Altman on Monday, alleging ChatGPT is a dangerous product for users' mental health and public safety -- centering the suit on accusations that ChatGPT advised the suspect in a fatal Florida State University shooting.

The suit applies product-liability law, the same strategy plaintiffs are using in thousands of cases against social platforms and the same approach almost every U.S. state waged against tobacco companies in the 1990s, leading to multibillion-dollar settlements and cigarette-marketing restrictions.

Two legal shields that protected social media may not hold for AI: Section 230 may not apply because the chatbot, not a user, produces the disputed speech (UF professor Jane Bambauer: "there's just no other party to sue"); and First Amendment coverage of chatbot output remains unsettled.

Deadline: Litigation is active now. No compliance deadline. Exposure is ongoing.

Risk: A product-liability verdict against an AI developer could expose any company deploying a user-facing chatbot to similar suits. Section 230 may not be a reliable shield where the AI, not a user, generates the content in dispute.

Your move: Pull your legal team into a review of every customer-facing chatbot your company operates. Map the outputs it can produce and document existing safeguards - especially around mental health, minors, and safety-critical advice. If Section 230 is baked into your liability assumptions, complete a revised analysis against the product-liability theory Florida is now testing in court.

AI News

• 🏢 Oracle's AI-cloud backlog hits a record $638B: Q4 FY26 remaining performance obligations (RPO) grew $85B QoQ to $638B as Cloud Infrastructure revenue rose 93% to $5.8B and total revenue hit $19.2B (+21%). FULL STORY

• ⚖️ Taiwan weighs criminal ban on AI-chip exports to all of China: Bloomberg reports Taipei is considering legislation that would extend export curbs beyond blacklisted firms like Huawei to all Chinese customers and criminalize unauthorized AI-server shipments for the first time. FULL STORY

• 🏗️ Applied Digital signs a $5.2B AI data-center lease: The Dallas firm inked a 15-year, 210 MW take-or-pay deal at Delta Forge 2 with a US investment-grade hyperscaler, lifting shares ~10% and pushing its contracted lease portfolio to ~$36B across 5 campuses. FULL STORY

• 🔒 A Security exits stealth with $37M: The autonomous offensive-security startup, backed by Lightspeed and Cyberstarts with angels including Wiz CEO Assaf Rapaport, continuously finds and chains vulnerabilities to validate real attack paths before weaponized AI can exploit them. FULL STORY

Trending AI Tools (Sponsored by the AI Executive’s Pass)

A curated look at the AI tools quietly transforming how teams work.

1. Incogni*: Data brokers feed AI your info. Incogni removes you from 200+ of them — automatically.

2. Persuwise: AI email coach that enhances outreach and accelerates positive responses.

3. Credal: Enterprise-focused AI agent platform enhancing data security and compliance.

^{*indicates a sponsored tool, if any}

⚡️ Why pay for 4–6 separate AI tools at full price when the AI Executive’s Pass fixes that?

The Money: In-house legal teams are the new outside counsel budget

Two rounds announced days apart, both targeting the same wallet: outside counsel spend. Sandstone and Wordsmith are building the infrastructure that lets corporate legal teams keep more work internal, and both just raised to accelerate that displacement.

Deals to know:

• Sandstone (Series A, $30M, valuation undisclosed) -- AI platform for in-house legal teams that ingests intake from Slack, email, and Jira, then routes, drafts, and reviews work inside the department. Investors: Lightspeed Venture Partners, Mantis VC, SV Angel, Operator Partners, Kearny Jackson, Daybreak Ventures, Litquidity Ventures

• Wordsmith (Series B, $70M, valuation undisclosed) -- Receive/Route/Resolve/Record platform that handles routine legal requests end-to-end, escalating to lawyers only when judgment is required. Investors: Highland Europe, Index Ventures

Signal: Capital is no longer chasing AI copilots that help one lawyer draft faster. Instead, capital is backing systems that absorb entire request queues and keep them off the outside counsel invoice.

Refer a Friend

Latest episode: Microsoft and Mayo Clinic Build AI for Healthcare → Listen here
Want to reach 400k+ decision-makers? → Sponsor us

Until next time, Arturo and Liam.

P.S. Unsubscribe if you don’t want us in your inbox anymore.