The foundation for physical AI.
El Segundo, CA
Joined May 2020
- Tweets361
- Following13
- Followers197
- Likes5
A sensor stops reporting at 2am. The dashboard keeps showing its last reading as if nothing changed. A person might catch it. An agent has no reason to doubt it.
How does your team decide when a reading is too stale to act on?
#PhysicalAI #WorldModel
More on how we apply this to location data in Jose de Castro’s Sensor Fusion blog: na2.hubs.ly/H08d3KM0
A hot rack, the technician who could check it, and the inspection robot usually show up in three different systems.
In episode 2 of our new One Day AI video series, they’re in one 3D view of a simulated 1,000-rack data center 🎥
Watch (link in reply)
#PhysicalAI #OneDayAI
We’re at CREtech New York! Find Shaun Cooley, Robert Goodman, and Donovan Fitch at Booth #326 at the Javits Center today and tomorrow. Come say hi 👋
#PhysicalAI #CREtech
We gave Claude a model of a simulated airport. The ops dashboard took 10 hours. Then came an assistant inside it, and a simulated facilities manager to test it ✈️
Jose de Castro walks through the build (link in reply)
#PhysicalAI #OneDayAI
Barely unpacked from Detroit and we’re headed to NYC 🔥
Catch us at CREtech next week, Booth #326 at the Javits Center. Shaun Cooley, Robert Goodman, and Donovan Fitch will be there to talk about what AI-ready operational data makes possible.
#PhysicalAI #CREtech
Mapped retweeted
Sensor Fusion in action.
Multiple sensors. Different coordinate frames. Different confidence levels. One continuously updated live world model of where people, vehicles and robots are — and how they’re moving.
@Mapped is the spatial foundation Physical AI needs.
Mapped retweeted
.@Mapped now has Sensor Fusion: one live world model for every person, vehicle and robot in a site.
UWB, BLE, Wi-Fi, cameras, GNSS, odometry, IMUs and more become a single canonical pose — giving Physical AI a real-time understanding of the world around it.
No single sensor tells the whole story. That's why we built Sensor Fusion: one live view of where people, vehicles, and robots are, and how much to trust it.
Our CTO Jose de Castro breaks it down below. At NexusCon? Get a first look at Booth #12 👀
#PhysicalAI #WorldModel
We're in Detroit for NexusCon! Excited for the next few days of conversations and connections. If you're here, come meet the Mapped team at Booth 12 👋
Every evaluation has a moment where a data engineer asks whether this becomes another place their data lives.
It doesn't. We get the data out, turn it into a structured model of the site, and stream it to Power BI, Snowflake, Databricks, Fabric or Grafana.
#PhysicalAI
Our CTO walks through what this looks like with Microsoft Fabric: go.mapped.com/mapped-microso…
The retention question in a deal rarely comes from the person you've been talking to. It comes from their security team, as a policy.
Three months by default. Ten years for audit. Days if the policy is strict.
Say yes to the policy. We'll match it.
#PhysicalAI #WorldModel
Controlling a building system safely usually comes down to one agreed point you're allowed to set, like an occupancy mode.
You set it, the building's own logic handles the rest.
That pattern works anywhere. The protocol underneath changes at every site. That's what we handle.
Send a command to a building system and it works. A second later, the value is back to what it was.
The building's control logic saw a number it didn't set and put it back.
The fix isn't commanding harder. It's commanding a point that logic watches.
#PhysicalAI #WorldModel
Data center operators are choosing infrastructure for a portfolio that doesn't exist yet, for customers they haven't signed, in buildings they haven't seen.
The decision gets made at one or two sites, then it's load-bearing for the next five years.
#PhysicalAI #WorldModel
More on how this works for multi-site operators: mapped.com/industries/datace…
