Persistent core
Slower-moving, diversified evidence establishes the strategic portfolio context across liquid asset classes.
Tensor is building an agentic research and operating system for systematic, multi-asset strategies. Specialist agents expand the research surface, structure evidence and support repeatable operations—inside explicit human mandates, controls and audit trails.
Tensor-Qube is the internal decision architecture at the centre of our research. It is designed to combine persistent cross-asset evidence, faster market-structure signals and explicit portfolio constraints without allowing one model or narrative to dominate.
Slower-moving, diversified evidence establishes the strategic portfolio context across liquid asset classes.
Market internals, liquidity and structural flows are investigated as bounded overlays with causal timestamps.
Cross-asset state and stress information changes how evidence is weighted—not the history used to discover it.
Exposure, concentration, liquidity and drawdown controls remain embedded in the architecture and subject to human limits.
The objective is controlled scale: more research breadth, more consistent operations and a clearer record of who approved what.
Tensor is designed around specialised agents rather than one opaque model. Research agents search, test and challenge ideas. Operating agents organise documents, prepare reporting and monitor workflows. A shared control plane records provenance, permissions and review status before anything can move forward.
The same discipline applied to research can support the operating layer: intake and classification of records, controlled document routing, scheduled reporting, exception monitoring and an auditable hand-off to responsible people. Agents prepare, reconcile and escalate; authorised humans review and decide.
Scan structural market change, translate observations into falsifiable hypotheses and preserve point-in-time evidence.
Separate exploration from production, lock test rules and surface decay, overlap and failure conditions.
Route documents, prepare recurring reports and monitor process exceptions under defined permissions.
Investment decisions, model promotion, external communication and risk limits remain with responsible people.
Every candidate starts with a mechanism, point-in-time inputs and a reason it could fail. A compelling backtest is not sufficient.
Agents operate within scoped permissions, recorded workflows and explicit review gates. Scale never removes accountability.
A deliberately focused liquid universe can still draw on wide, orthogonal evidence without accumulating fragile instruments or redundant signals.
Tensor Asset Management S.à r.l. is a newly incorporated Luxembourg company. It is not currently authorised, licensed or directly regulated by a financial supervisory authority and has not commenced any regulated activity. Any future regulated activity will begin only after applicable legal and regulatory requirements have been completed.
Share a concise professional profile and the purpose of your enquiry. This helps us route corporate, research and institutional conversations to the appropriate person.