{{first_name | Hi}}, your strategic AI update is here.
Banks are hiring people to decide which agent does what, Axios warns companies will answer for their rogue agents, and Bank of America screens every AI use case against 16 risks. If you run agents, every task they touch needs an owner. I'd draw those lines before the next agent goes live. Only 26% of organizations clearly define which work is human-led, AI-assisted or AI-executed.
Innovation Snapshot
"Certainly it will be about how teams and employees use AI agents effectively, but the ultimate force-multiplier will be the context that the agents get."
— Aaron Levie, Co-founder and CEO, Box
In today's lineup
Robots
Build the scaffold before you buy a bigger model
Real ROI from Gen AI
When an agent crashes halfway through the job
People
Natasha Jones builds the data layer energy agents need
Love
Give one person the job of connecting your agents
Sapiens by Yuval Noah Harari
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
McKinsey's Eric Kutcher says AI hasn't reached company-wide productivity, and says that takes redesigning the work end to end.
Ferrari's chairman shrugs off AI bubble fears, but says the business models are still unclear.
AI labs are gaming out a public revolt after a major AI incident, which insiders expect within a year, so know which agents you could switch off fast.
HBR says shopping agents are changing who owns the customer, so brands must win the agent too.
SoftBank seeks $100bn from Gulf investors for AI, as the AI infrastructure race gets pricier.
Build the scaffold before you buy a bigger model
Two-thirds of companies run AI agents in production. Only 7% have them working together, according to a recent IDC survey of 400+ tech leaders. Agentic AI budgets are set to double anyway.
Google's James Manyika and Blaise Agüera y Arcas, with philosopher Benjamin Bratton, call the single-superintelligence picture "likely the wrong vision" in a DeepMind Institute essay. The intelligence, they write, will reside "in the embedded rules, procedures, precedents, and feedback mechanisms of the scaffold."
Your competitors rent the same models from the same few providers. The scaffold around those models is yours to build, and it has to fit how your business runs.
Where to start:
Check that one weekly decision your agents feed would survive a new owner or a different model
Map who hands what to whom before you add another agent
Write each agent mistake back into a rule or escalation path, so it doesn't happen twice
The leadership pulse
Will a bigger AI model fix your biggest AI problem?
Real ROI from Gen AI
How human-centric AI delivers business results
C.H. Robinson sends anything its hundreds of planning agents can't correct to a team member, and for Stanley Black & Decker 87% of incomplete orders now get corrected with no one touching them.
Wellstar Health System lets AI agents built on Microsoft Foundry match nurses to open shifts. Nurse managers keep the approvals and now spend over 80% less time filling shifts, on average.
On, the Swiss sportswear brand, moved 24 services to Google Cloud with specialized AI agents doing the steps and engineers signing off each switch to production, cutting each move from three months to two weeks.
Tool Spotlight
Your weekly briefing on tools that create competitive leverage
When an agent crashes halfway through the job
Temporal records every step an agent takes, so after a crash the job resumes where it stopped, without losing pending approvals. It's open source, works with any model, and recently brought in authorization startup Oso's team to keep agents within company permissions. Best for Chief AI Officers whose agent jobs can't start over.
Restate does the same as one small program with no separate database, and an agent can wait days for a sign-off at no compute cost. It recently raised $20M, and Replit's coding agent runs on it. Best for CIOs who don't want another cluster to run.
Pick by how your team builds
PEOPLE 👥
Meet the innovators turning bold ideas into real-world impact.
Transformation Champion
Natasha Jones builds the data layer energy agents need
Natasha Jones, a former Octopus Ventures investor, co-founded Metris to pull a solar portfolio's scattered data (inverters, meters, SCADA, billing, spreadsheets) into one live record per asset. Its AI agent, Metria, works on top of that record, spotting underperformance and handling scheduling. Metris raised a $5M seed recently and says it manages more than 10,000 solar plants. Next, Jones wants operators to trust Metria to carry out the actions it now only suggests.
LOVE ❤️
Practical wisdom, growth tactics, and a must-read book that will challenge the way you think.
Give one person the job of connecting your agents
In spring 2025 I spent a week with Linda Hill at Harvard Business School, in the Leading in the Digital Era program. She describes three core roles in innovation: the Architect builds, the Catalyst seeds ideas, and the Bridger connects teams so they work as one. The one I'd never skip is the Bridger. With AI agents, that's whoever owns the handoffs.
Name one Bridger for each workflow where agents and people share work.
Pick someone both IT and the business already trust.
Give them a seat in every agent rollout from week one.
I wrote more about all three roles in the LEAP Guide.
Transformative Reads
One book, handpicked from my conversations with friends, industry leaders, and tech innovators:
In Sapiens, historian Yuval Noah Harari argues that humans came to run the world because only we can cooperate flexibly in large numbers, even with strangers. What holds that together is shared stories everyone agrees to follow, like money, nations and human rights. Worth reading before you ask a dozen agents to work as one team.
Perfect for: CIOs and Chief AI Officers writing the rules their agents will share.
In Culture
66% have the agents. 7% have the answer.
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
