An AI lab for new ways of working. We experiment with frontier models and applications, then turn what works into how organizations work next.
Where experiments become practice.
Built to experiment
“We build to learn, and we ship what teaches us the most.”
Dancing Robots is an AI lab. We run hands-on experiments with frontier models, agents, and intelligent automation in search of the patterns that change how organizations work.
The best experiments graduate: we develop them into prototypes, playbooks, and applications that partner teams put to work every day.
What we experiment with
Capability studies
Hands-on evaluations of new frontier models the week they land: what they make possible, where they break, and what it means for real work.
Agentic workflow prototypes
Autonomous systems that plan, act, and hand off to people, built and stress-tested against real operations.
Embedded experiments
Prototypes placed inside partner teams to observe how work actually changes, and what people do with the new capability.
Playbooks & operating patterns
The experiments that win graduate into playbooks, tools, and patterns organizations can adopt at scale.
From question to working pattern
Question
Pick a real workflow and a sharp hypothesis
Prototype
Build the smallest system that can teach us something
Field test
Run it inside the real workflow, with real people
Learn
Measure what changed, and why
Scale
Turn wins into playbooks and production systems
What comes out of the lab
Prototypes
Working agentic and multimodal demos grounded in your real workflows
Playbooks
Repeatable patterns your teams can adopt and own
Evidence
Clear reads on what actually works, measured in real workflows
Momentum
Teams that have seen the future of their own work, and know how to build toward it
Let’s run the first experiment.
Tell us which part of work you’d most like to reinvent. We’ll design a pilot with frontier models inside your real workflows, then report back what we learn.