Skip to main content
The technodollar preserves purchasing power through how the basket is allocated, rather than through the safety of any single asset. Programmable rails carry the instrument; the allocation engine does the work of holding value as conditions change. An Investment Committee, chaired by Dr. Nouriel Roubini and composed of experienced asset managers, sets the hypotheses and the rules. Beneath it sits Atlas AI, the research and execution architecture that runs the experiments the Committee directs and returns reports, results, and recommendations. The Committee decides. Every macro framework decays. The logic that worked through the disinflation of 1985 to 2020 fails under fiscal dominance and rising real-rate volatility. The engine exists to keep the framework current as markets adapt. The engine runs in four layers, each with a distinct role.
1

Research

Sweeps the academic literature for findings on regime identification and risk transmission.
2

Execution

Validates the data, builds the regime-classification and factor-decomposition models, and tests them out of sample.
3

Oversight

Runs every result past two review committees, one for scientific soundness and one for investability under the mandate.
4

Decision

The Investment Committee weighs the reports against the mandate and makes the final call.
Constitutional rules hold the layers together. Pre-registration, append-only revision logs, and supersession checks prevent any result from being quietly overwritten.
Screenshot 2026 06 02 At 6 59 43 PM
The division of labor is deliberate. The allocation itself runs on classical statistical and machine-learning methods, each pinned to a fixed version per experiment. Large language models sit one layer up, where they run the directed research, validate results, maintain the audit chain, and generate the macro and geopolitical scenarios that inform defensive positioning.
No AI has autonomous allocation authority. The Investment Committee decides every allocation. AI’s role is expanding by stage, from execution and operations today toward research, and only under future disclosed mandates toward broader allocation work. The Committee’s approval of every allocation does not change.
Cadence is fixed. The Committee reviews and approves every monthly allocation before it is deployed, and any off-cycle rebalance runs through a defined fast-path that requires the same sign-off. The engine classifies conditions along a spectrum that runs from muddle-through to disorderly, and the basket is built to answer each one. In each case the construction serves one goal: to hold value when holding it is hardest.