Public timeline

Capability timeline

Xolver's action-intelligence work is evolving toward machines that can understand changing work, predict likely outcomes, and stay inside reviewed boundaries.

September
2024

The Foundation

  • Focused on how machines understand task context.
  • Explored semantic cues for physical work.
  • Improved the product story around responsive machine behavior.
November
2024

Perception & Data Scaling

  • Expanded thinking around multi-step industrial work.
  • Improved attention to visual details that matter in operations.
  • Studied how human demonstration can shape machine behavior.
  • Strengthened the emphasis on visual reasoning.
January
2025

Intelligent Reasoning

  • Explored how machines can consider likely outcomes before acting.
March
2025

Efficiency & Stability

  • Focused on more efficient reasoning.
  • Separated task understanding from physical behavior.
  • Improved the story around smooth, reviewable motion.
May
2025

Robust Planning & Agency

  • Improved plain-language task following.
  • Explored how complex requests can be broken into reviewable steps.
July
2025

Physical Grounding & Diffusion

  • Studied richer spatial understanding for physical tasks.
  • Used failure review to improve product direction.
  • Explored alternative approaches to proposing action sequences.
September
2025

World Modeling & Imagination

  • Studied how predicted object changes affect task planning.
  • Connected world modeling with reviewable machine workflows.
October
2025

Safety & High-Fidelity Simulation

  • Strengthened the product focus on reviewed boundaries.
  • Used simulation thinking to improve confidence before physical work.
November
2025

Real-Time High-Frequency Control

  • Explored responsive machine behavior at industrial speeds.
  • Studied how changing conditions affect operating context.
December
2025

Advanced Logic & Verifiable Success

  • Focused on faster, clearer reasoning.
  • Studied physical constraints as part of task understanding.
  • Improved handling of simple and complex workflows.
January
2026

Memory & Industry Standards

  • Expanded long-horizon operating context.
  • Improved the narrative around smoother machine behavior.
  • Aligned product language with industrial expectations.
February
2026

Brain-Inspired Spatial Math

  • Explored more efficient spatial representations.
  • Studied ways to improve learning data quality.
March
2026

Reflection & Vision Upgrades

  • Connected memory with adaptive task review.
  • Studied recovery from spatial uncertainty.
  • Improved visual reasoning direction.
  • Expanded research into contact-aware workflows.
April
2026

Cloud & Hardware Interfaces

  • Studied how teams coordinate AI-assisted workflows.
  • Focused on local review experiences near physical work.
  • Improved partner-facing application review.
  • Strengthened operating-boundary messaging.
May
2026

Next-Gen Transformers & Learning

  • Explored more efficient visual reasoning.
  • Studied better ways to handle changing observations.
  • Made uncertainty a clearer part of review workflows.
  • Improved evidence and review storytelling.
July
2026

Deformable Manipulation, Tactile Dexterity & Live Contact Monitoring

  • Shared public research notes on deformable-object workflows.
  • Expanded the story around tactile and dexterous manipulation.
  • Connected contact-rich work to Xolver's review philosophy.
  • Explored changing object state as an operating challenge.
  • Positioned contact-rich manipulation as a key research direction.
View research note →

Return to Architecture

View Technology Layers

FAQ

What does the capability timeline represent?

It is a public story of Xolver's product and research direction. It is not a feature matrix, launch schedule, or implementation guide.

How should simulation milestones be interpreted?

Simulation references describe how Xolver thinks about review and confidence before physical work. Detailed methods are shared directly during qualified conversations.

Why does the timeline include safety and runtime work alongside model work?

Xolver treats model capability, operator review, operating records, and facility context as parts of one product story.

Does every milestone apply to every robot or cell?

No. Each application depends on the work, equipment, facility, and team. The public timeline is intentionally high level.