vision lab raises $6m in fundinglearn more

the industrialdata layerfor physical rsi.

backed by
+ more
angels from

Project Watermelon
Frontier Lab #1

atomic action with clip-level task diversity, encompassing more than 1,000 environments

Project Strawberry
Frontier Lab #2

frame-perfect dense annotation and hand pose extraction

Project Cherry
Frontier Lab #3

atomic action annotations with full expert validation

Project Pear
Frontier Lab #4

industrial data SOP capture with engineer validation

Project Watermelon
Frontier Lab #1

atomic action with clip-level task diversity, encompassing more than 1,000 environments

Project Strawberry
Frontier Lab #2

frame-perfect dense annotation and hand pose extraction

Project Cherry
Frontier Lab #3

atomic action annotations with full expert validation

Project Pear
Frontier Lab #4

industrial data SOP capture with engineer validation

Project Watermelon
Frontier Lab #1

atomic action with clip-level task diversity, encompassing more than 1,000 environments

Project Strawberry
Frontier Lab #2

frame-perfect dense annotation and hand pose extraction

Project Cherry
Frontier Lab #3

atomic action annotations with full expert validation

Project Pear
Frontier Lab #4

industrial data SOP capture with engineer validation

client identities withheld under NDA

high-density action labelling with proprietary VLMs and human in the loop.

pick an industry to play
laboratoryhandling culture plates in sterile isolator
engineering data

from first sketch to finished part,engineering, captured as data.

the knowledge that builds the physical world, from the first idea to the finished product, becomes a data layer machines can finally learn from.

  • 01

    design

    how ideas become geometry, and the intent behind every decision.

  • 02

    simulation

    how designs meet the laws of physics before they exist.

  • 03

    manufacturing

    how digital intent turns into a physical part.

  • 04

    materials

    how matter behaves, from the lab bench to the factory floor.

  • grounded in real engineering practice
  • verified by working engineers
  • structured for machines to learn from
talk to us