Understand the scene.
Connect visual observations with objects, spatial relationships and task intent.
THE NEXT CHAPTER OF PHYSICAL AI
Intelligence,
embodied.
A world understood.
A future built with humanlike machines.
VEXKIN V1HUMANOID CONCEPT / 001
$150KRaised in the past 12 monthsCompany-reported · October 2026
01Unified visionHumanlike intelligence for real work
∞A world of possibilitiesResearch-led. Built to keep learning.
01 / OUR CONVICTION
The world wasn't built
for robots.
So we're building robots
for the world.
We believe the next leap in AI will happen beyond the screen. Vexkin's vision brings world models, embodied learning and humanoid systems together—so machines can reason about the physical world, then act within it.
02 / THE VEXKIN PLATFORM
One vision.
Three building blocks.
A body to act. A model to anticipate.
A learning loop to improve.
Our proposed product architecture.
VX—01 / EMBODIED SYSTEMCONCEPT RENDER
+Built around the way people work.
HUMANOID RESEARCH CONCEPT
Meet V1.
Our vision for an adaptable humanoid platform: coordinated movement, dexterous manipulation and task-level reasoning in spaces designed for people.
01
Whole-body coordination
02
Vision-guided manipulation
03
Human-supervised operation
Target capabilities. Hardware specifications and performance remain to be validated.
Explore intended applications03 / WORLD MODEL INTELLIGENCE
Before the first move,
imagine the next.
Our research direction: predict possible outcomes, choose an action, then update the plan as the world changes.
PRECISION PICK & PLACEOBSERVE / 01
STATE SPACEPREDICTED TRAJECTORIES
Selected pathAlternative
Perception identifies the component, the target and the surrounding workspace.
Reason about what's next.
Explore action-conditioned futures before committing to a physical movement.
Learn from what changes.
Use observed outcomes to refine plans and build a stronger learning loop.
04 / THE WORLD AT WORK
Human environments.
New possibilities.
We see the starting point in structured, repeatable work—with people in the loop.
Manufacturing
Component handling, machine tending and repeatable assembly support.
PROPOSED PILOT DIRECTIONLogistics
Item sorting, tote transfers and material movement in controlled workspaces.
PROPOSED PILOT DIRECTIONResearch labs
A platform for exploring embodied learning, manipulation and human–robot interaction.
PROPOSED COLLABORATION05 / STANDING ON SCIENCE
Frontier research.
Our starting point.
Following the ideas moving physical AI forward. Independent work that informs our direction, not Vexkin performance claims.
SEP 2026
FIGURE · TECHNICAL REPORT
GENERALIZATIONRead source AUG 2026Helix 2.5
Whole-body behaviors that transfer to previously unseen homes.
LIU ET AL. · PREPRINT
WORLD–ACTION MODELSRead paper MAR 2026StageWAM
Stage-level prediction for progress through manipulation tasks.
MUR-LABADIA ET AL. · PREPRINT
SELF-SUPERVISED LEARNINGRead paper JUN 2025V-JEPA 2.1
Dense visual representations for spatial and temporal understanding.
ASSRAN ET AL. · FOUNDATIONAL PAPER
PREDICTIVE WORLD MODELSRead paper V-JEPA 2
Learning from video to understand, predict and plan in the physical world.
Research selection updated 08 October 2026. Referenced authors and organizations are not presented as Vexkin partners.
06 / THE PEOPLE & THE AMBITION
Research at our core.
Impact on our horizon.
Building the connection between scientific ideas and machines that can do useful work.
ACADEMIC BACKINGAs provided by Vexkin
Prof. Truyen Tran
Prof. Tho Quan
Prof. Anthony
Tung Kum Hoe
OUR NEXT CHAPTER
From conviction
to physical intelligence.
We're preparing for the next funding conversation: turning a research-led vision into testable systems and focused industrial pilots.
EARLY CAPITAL / USD$150,000
Raised over the past 12 months.
Company-reported · October 2026THIS IS ONLY THE BEGINNING.