Foundation policy
Cross-embodiment grasping out of the box.
Platform
The full visual + code manipulation platform
Train, deploy, and orchestrate manipulation from one platform. Visual tooling for teams, source-grade control for engineers.
Built on NVIDIA Isaac, GR00T, Jetson & Omniverse
NVIDIA InceptionIsaac LabGR00TJetson ThorOmniverseCosmos
Core capabilities
What the platform gives you
Everything from data generation to on-robot control, in one place.
Touch prediction
Closed-loop control grounded in contact.
Edge runtime
Real-time inference on Jetson.
Visual fine-tuning
Adapt policies without deep ML work.
Source & GitHub
Export configs and integrate with your CI.
Experts standing by
Hands-on help from robotics engineers.
Visual editing
Full visual editing for manipulation
Compose skills, set safety limits, and monitor fleets from a visual console — no scripting every contingency.
- Drag-and-drop task graphs
- Live grasp telemetry
- Fleet rollout controls
Code & GitHub
Source code and GitHub when you need it
Drop to config and code when you want full control. Version, review, and ship like any other software.
- Exportable policy configs
- CI-friendly integration
- Reproducible builds
Workflow
Idea to deployed dexterity
1
Generate data
Simdex builds contact-rich datasets in simulation.
2
Train & fine-tune
Dextra learns your hand and objects.
3
Deploy
Reflex ships it on-robot with safety limits.
4
Orchestrate
Sequa sequences and recovers multi-step tasks.
100+
Integrations & object types
4
Layers, one platform
24/7
Fleet monitoring
Loved by teams
Built for builders
“One platform replaced a pile of brittle scripts.”
“Visual editing got our non-ML engineers shipping.”
“GitHub export made it fit our existing workflow.”
Platform FAQ
Good to know
Is it cloud or on-prem?
Training runs on GPU cloud or DGX; inference runs on-robot at the edge via Reflex.
Do I need ML expertise?
No — visual fine-tuning covers most teams; engineers can drop to code anytime.
Which robots are supported?
Any arm or hand with supported sensors; cross-embodiment by design.
Why GripSim
Built for production robotics
The manipulation layer teams ship with, not a research demo.
Cross-embodiment by design
One policy runs across many grippers and hands — no per-robot rewrite.
Closed-loop with touch
Tactyl predicts slip and force so grasps react before they fail.
Runs on the robot
Reflex executes on Jetson at control-loop latency, with safety limits.
A moat that compounds
Every deployment feeds the shared model through the data flywheel.
Stay in the loop
Manipulation research, in your inbox
Benchmarks, sim-to-real notes, and product updates. No hype, no spam.
See the platform in action
Start free or book a guided walkthrough with our engineers.