Lab note
The research agenda at formation
Four directions for an early lab: reasoning systems, multimodal intelligence, agents, and safety evaluation.
A broad mandate, an honest starting point
Capvise Labs is being formed as an AI research and engineering lab within Capvise Group. The mandate is broad by design. The current body of public work is not.
At formation, the lab has one public research system in development, VetoBench, and one applied venture, Marka. The rest of the site describes questions we intend to pursue, not completed programs or undisclosed models.
The four directions
Reasoning systems asks how plans, memory and verification can survive long tasks. Multimodal intelligence asks how models should combine evidence across text, images, audio, video and structured environments.
Agents and autonomy focuses on tool use, state, recovery and operator control. Safety and evaluation focuses on measurements that reveal unreliable or risky behaviour before deployment.
How the agenda will change
Projects will be added only when there is an owner, a defined question and work that can be described without inventing results. Research directions may narrow, merge or disappear as evidence accumulates. Public pages will retain clear status labels so ambition is not confused with release.
Document status
This page records the lab's position at publication. It does not claim an unreleased model, completed benchmark run or peer-reviewed result. Material changes will be dated.