Welcome to Vantage Point, the newsletter you’ve known as Level.UP. 

We’re rebranding with a new name, but we’ll be focusing on the same insights: what AI means for the people running the physical economy, and what’s actually worth doing about it.

This week we look at what industrial AI needs to work, starting with Schneider Electric’s plan to put AI inside plants. We also examine SAP’s plan to connect robots to the systems you already use to run your business.

Plus: a $215M chemistry startup, robots on a monthly bill, and $600M for autonomous excavators.

Think someone else needs this? Forward it to a friend or colleague navigating the same terrain.

Schneider Electric wants AI to help plant teams investigate equipment problems and manage energy use, using data that stays close to the operation.

The company announced a partnership with NVIDIA and HPE to bring AI agents into industrial facilities. The systems are designed to connect equipment alarms with maintenance history and operating conditions, helping teams identify what needs attention.

There’s an existing foundation. Pulp and paper producer CMPC Chile has used Schneider’s monitoring software for three years to track the health of its control systems. The new agents are a proposed next step.

That distinction matters. Getting a clear view of the operation comes before giving AI more responsibility.

A maintenance log can tell you a pump was repaired. An experienced operator can tell you why the same problem keeps coming back. Making that knowledge usable is a substantial part of building industrial AI.

An alarm might signal a failing component. It might also be expected during a changeover. The useful answer depends on what the plant is making, what happened on the previous shift, and what the maintenance team has already tried.

That context lives across software, operating procedures, and people. Connecting it takes engineers who understand the equipment and operators who know where the written process differs from daily practice.

This is where company-specific development earns its place. Standard AI software can provide a starting point. Someone still has to connect it to the plant’s systems, account for its exceptions and establish which decisions require a person.

Your plant already has people who know which alarms matter and which fixes hold. The opportunity is to make that judgment available across shifts and sites, without losing the experience behind it.

SAP wants the software that runs your warehouse to tell the robots what to do, then record what they actually did.

At TechEd on October 27, SAP introduced its physical AI approach: business systems assign work to machines, and the results flow back into the same business records.

SAP says the approach works across hardware and AI models. The promise to operators is a clearer connection between a work order and what happened on the floor.

A robot can finish its trip while the business still has unfinished work.

Picture this: The pallet reaches the loading dock, and the machine even reports success. But the inventory record never updates, so the next shift sends another robot to collect it. The physical task worked, but the handoff failed.

That is the useful part of SAP’s announcement. A company already running its warehouse or maintenance processes in SAP has a reason to connect machines to those records. Fewer separate dashboards could mean less reconciliation for the people running the operation.

But the failed task is the demo worth watching. What happens when the robot reports a delivery that a supervisor disputes? Who can correct the record or authorize another attempt? What survives a lost connection?

Those decisions also shape dependence on a supplier. Supporting several robot brands helps. Moving the rules for accepting work and resolving exceptions to another software system may still be expensive.

At the dock, a driver is waiting for inventory the robot says it delivered. The supervisor needs to find the pallet, correct the record, and release the truck. That is the integration worth paying for.

Ready to work smarter? Here are some tools we’re tracking, picked for this edition’s security story:

Warp Takes Chemistry as a Service. Warp Sciences emerged from stealth with $215M in funding on October 22, led by Lightspeed. The company combines chemists, AI, and robotics to make and test molecules for drug researchers. Customers retain their drug programs; Warp is building a chemistry service rather than its own drug pipeline. The signal: Buying completed experiments could be more useful to a research team than buying another AI tool. The funding establishes ambition. Repeatable chemistry and delivery times will determine how much work customers entrust to it. → Read more

Formic Puts the Robot on a Monthly Bill. Formic announced $50M in combined equity and debt financing on October 22. Its Series B was led by Mercato Partners. The company bundles equipment, deployment, financing, and service into a monthly charge. The pattern: Factory automation requires someone to carry the financing and maintenance burden. A capable machine still needs a business arrangement a factory can use. → Read more

Bedrock Raises $600M To Automate The Excavator You Already Own. Bedrock Robotics announced a $600M Series C on October 26, co-led by CapitalG and Emergence Capital. The company retrofits existing construction equipment and says its autonomous excavators began paid work on live commercial sites in August. Contractor Haydon is deploying the technology in Texas. The signal: Contractors can try autonomy without replacing the machines on their balance sheets. The next test is whether Bedrock can make deployment and field support repeatable across crews and sites. → Read more