{{first_name | Hi}}, your strategic AI update is here.
IBM had its worst day in 60 years, General Compute raised $400M on inference chips, and Moonshot released the largest open-weight model ever. Buying the tools is the easy part. Grant Thornton surveyed 1,000 C-suite leaders: 51% say strategy drives AI ROI, 79% don't have one.
"In the Age of AI, organizations need to build their radar."
— Dan Mapes, Founder & CEO, Verses AI
In today's lineup
Robots
Strategy drives AI ROI. Only 22% have one.
Real ROI from Gen AI
Test your agent before it touches a live workflow.
People
Sassie Duggleby rewrites rocket propulsion from the ground up
Love
Build the AI program backward from the CFO's number.
Playing to Win, Roger Martin and A.G. Lafley
Reading time: 3 min.
ROBOTS 🤖
How are robotics and AI changing industries? We break down the latest news, tools, and innovations for you.
Top Insights
IBM's worst day in 60 years as customers pulled software budgets to fund AI infrastructure, sending shares down 25%.
General Compute raises $400M using inference chips as collateral, the first deal of its kind, as inference becomes the real AI cost line.
Moonshot AI launches Kimi K3 with 2.8 trillion parameters, the largest open-weight model ever released.
Xi proposes a World AI Cooperation Organization in Shanghai, splitting global AI governance in two.
Z.AI hits $1B in annual sales becoming the first Chinese AI company to reach that mark.
AI's productivity gains are not reaching workers and the backlash is growing.
Strategy is the #1 driver of AI ROI
Grant Thornton's 2026 AI Impact Survey of nearly 1,000 C-suite and senior business leaders: 51% say strategy is the top driver of AI ROI. 79% of operations leaders do not have a fully developed one.
The revenue data shows what that costs. Companies with fully integrated AI are four times more likely to report growth: 58% versus 15% for those still piloting. In manufacturing, 48% are still in pilot, versus 34% across industries.
The one in four who are ahead:
Lock a Finance-validated baseline before deployment begins.
Measure at the workflow level, where a process change reaches the P&L.
Run 12-week cycles with Finance validating the outcome before the next one starts.
Where does your organization stand on AI strategy right now?
Real ROI from Gen AI
How human-centric AI delivers business results
PepsiCo used Siemens and NVIDIA to build AI digital twins of its US plants, testing every production change before touching the floor and delivering 20% throughput gains with 10-15% lower capital expenditure on first deployments.
American Express scaled AI-assisted development tools to 11,000 engineers, cutting coding cycle time by more than 30%, per CEO Steve Squeri's 2026 shareholder letter.
Honeywell and MIT modeled five years of AI deployment across LNG operations, projecting $15 billion per year in production cost reductions for operators who implement AI-enabled process controls.
Tool Spotlight
Your weekly briefing on tools that create competitive leverage
Before you trust your agent with a real workflow, test it here
Bespoke Labs raised $40M on July 6, 2026, backed by Anthropic, OpenAI, and Meta insiders, building RL environments that mirror your actual enterprise systems so agents are trained and tested before they touch production data.
Best for: Chief AI Officers and technology leaders deploying agents into allocation, order management, or finance workflows where a wrong agent decision costs money.
Choose if you...
Are about to hand a live workflow to an agent and need proof it works before your CIO approves the rollout.
Work in manufacturing, energy, or healthcare where agent errors reach the P&L, not just a log file.
Want the agent trained on your actual data and systems, not a vendor's sample environment.
Patronus AI raised $50M on June 25, 2026, led by Greenfield Partners, building simulated environments that stress-test agents against your industry's specific failure scenarios and generate an audit trail before any live deployment.
Best for: AI and innovation leaders in regulated industries who need to show the board or compliance team how agent behavior was tested before go-live.
Choose if you...
Need to test how the agent handles edge cases specific to your operations, not a generic lab scenario.
Are deploying in energy, healthcare, or financial services where every agent decision needs an auditable record.
Are being asked to show how you validated the agent before it ran autonomously on live decisions.
3 other tools to explore
Arize AI suits teams that need agent monitoring and drift detection once agents are in production.
Confident AI suits teams that want open-source LLM evaluation with CI/CD integration.
Weights & Biases Weave suits teams already on W&B who want agent evaluation without adding a new platform.
PEOPLE 👥
Meet the innovators turning bold ideas into real-world impact.
Transformation Champion
Jenny Lee gives America's aging families an AI care team
Sassie Duggleby co-founded Venus Aerospace in 2020 after spotting a design flaw 70 years of rocket engineering had never fixed: the traditional combustion chamber wastes propellant. Her Rotating Detonation Rocket Engine ran 600 ground tests before its first flight in May 2025. Venus closed a $91M Series B earlier this month, led by Mercury Fund with Lockheed Martin Ventures, for defense and space deployment.
LOVE ❤️
Practical wisdom, growth tactics, and a must-read book that will challenge the way you think.
Build the AI program backward from the CFO's number
Seventy percent of AI transformation initiatives fail. The technology works. The sequence is wrong.
Platform approved. Pilots run. Adoption dashboards go green. Eighteen months later, the P&L hasn't moved. The AI was real. The transformation was theater.
The proof: one company invested $2.5M in AI and generated zero measurable value. Two steps were missing: a Finance-validated baseline and a named cost or revenue line for each capability deployed.
Start with the outcome Finance will validate. Then build the program to deliver it.
Get the full framework in my new book, 100x.
Transformative Reads
One book, handpicked from my conversations with friends, industry leaders, and tech innovators:
In Playing to Win, Roger Martin and A.G. Lafley define strategy as two specific choices: where to play and how to win. Budget allocations and technology roadmaps are not substitutes for those choices. Most AI programs skip straight to deployment without making them. That is why 70% never reach the P&L.
Perfect for: Gamora-CIO and Star-Lord-CFO who need to articulate why "we're doing AI" and "we have a strategy" are not the same statement.
In Culture
THANK YOU!
Lighthouse takes us many hours to create, and we don’t run ads or charge anything. We are dedicated to making it the best AI transformation newsletter possible.
We just ask for one thing: tell one friend about Lighthouse today!
Want daily AI & Innovation insights? Connect with me on LinkedIn or X.
Got suggestions? Just reply to this email!
Much Love,
Matt and the Future Works team
