Robots building robots: the world we want
A world where robots help decide what is needed, build it, and keep improving it.
What world is Gantry working toward?
We want a world where robots and agents can work out what is needed, define the work, and help build and improve the machines that do it. The goal reaches beyond a single robot or a single product. It includes making, using, repairing, and reusing what we build.
Where does the next idea come from?
From what already exists: how a machine performs, where it fails, what parts cost, and what can be repaired or reused. Agents can use this evidence to compare possible next steps. A new design is one answer. Improving an existing robot—or deciding not to build anything—may be better.
Why does this need a shared record?
Many agents can work at once, but their work must still fit together. A software change can alter a physical limit. A new part can make an earlier test irrelevant. The team needs to know which version was tested, what was decided, and why. Gantry brings software and hardware development into one shared history.
What should the platform connect?
Our wider vision connects design, digital twins, sourcing, manufacturing, delivery, operation, repair, and recycling. Digital twins belong throughout development and use, wherever a model can help test a choice. External tools do the specialist work; Gantry provides the shared context for agents to coordinate around it.
What makes the work improve over time?
Designs, trials, discussions, decisions, and results should inform the next task. Agents should be able to form useful roles and carry lessons forward. Those lessons must remain open to correction when new evidence arrives. We want to build this world together, through open source.
Written by the Gantry team. Published November 2, 2026.
Further reading: Gantry on GitHub · Ledger and Crew responsibilities. For setup and current supported behavior, use the repository documentation.