Vision

There's a room in Bangkok with a fleet of robot arms, 3D printers, a wall of ESP32 boards, AI glasses, and a cyclorama we lit with Godox strobes and wired up ourselves. Nothing in that room is required to run the product we sell. It's still the most important room in the company.

This page explains why.

The same primitives run an inbox and a robot arm

The product does its work against business systems — inboxes, ERPs, databases, repos. Every one of them shows up the same way: as a cell in a scene, a window onto something real, with an agent that acts through gates and a run that gets recorded.

Here's the thing we believe: nothing in that sentence is about business software. A temperature sensor is an asset the same way an inbox is. A robot arm is an asset whose tools move metal instead of records. "Update the row" and "move the arm four centimeters left" are the same shape — a proposed action, a gate if it's consequential, a recorded outcome. The episode log of a grasp attempt and the audit trail of an accounting close are one format.

If that holds, the runtime we sell for operations is also the runtime for orchestrating physical work — and the systems people build bespoke today, from warehouse automation to the self-driving labs emerging in chemistry, biology, and materials science, are one architecture, not a hundred.

Physical intelligence needs operations software

Physical intelligence without operations software is nothing. A robot that can grasp is a demo until something orders its spare parts, tracks what arrived, schedules the next run, and writes down what happened. Real labs and real factories are mostly operations with some actuation at the end — which is why the interesting thing to build is not robotics plus some software on the side, but one runtime where both live.

We test that combination on ourselves. The lab's own operations run through Daslab: parts orders from Taobao, what shipped, what arrived broken, what to reorder before the next build. The lab feeds us messy operational problems every day, and the product has to eat them. The product doesn't need the room; the room runs on the product.

Reality is the best forcing function

"If that holds" is carrying a lot of weight up there. The lab is where we make it hold.

Software that only ever works in conversation develops a kind of confidence that nothing has tested. A robot arm doesn't extend that courtesy. The arm reached the cup or it didn't; the print finished or it warped; the glasses streamed or they dropped frames. Those verdicts are free of interpretation, which makes them the most useful feedback we get. So we take the primitives we ship for business systems, point them at hardware, and see what breaks.

The lab today is a workshop

Nobody in Bangkok is running autonomous experiments yet. It's a workshop: we solder boards, calibrate arms, print fixtures, write scenes that drive deterministic motions, and film what happens on the cyclorama. The occasional bench note lands in the changelog.

The north star is a self-driving lab

The end state we want: a lab that does research on physical reality by itself. An agent proposes the next experiment, the runtime schedules it, arms and instruments run it, sensors record what actually happened, and the result decides the experiment after that — overnight, in parallel, every run recorded, every consequential action gated the same way an invoice reminder is gated today. The operations around the science — parts ordered, samples tracked, machines scheduled — run through the same scenes, because operations is most of what a lab is.

Our customers are on the same road. The businesses we serve run machines, warehouses, inventory, shipments — operations that already end in physical reality. The gates that protect their ERP today will run their shop floors tomorrow, in the same scenes, not in a separate product.

Research in chemistry, biology, and materials is bottlenecked on how many experiments a team can run and how honestly the results are recorded. Both are things a runtime can change. That's the thing we want to exist. Everything else on this page is how we get there.

Updated 2026-07-18