NadIntellect Lab

Control of Future Superintelligence

NadIntellect Lab is working on solving the problem of controlling and safely working with future superintelligence.

  1. Theory
  2. Digital experiments
  3. Live testing
  4. Scaling
  5. Physical reactor
Where we are now Fund the research
Next: live E3-A experiment
Mission

We are looking beyond the horizon of today’s LLM capabilities — toward systems whose intelligence may significantly exceed human intelligence.

Our goal is to move from theoretical models and digital experiments to a physical experimental reactor for controlled work with superintelligence.

The technical architecture of NadIntellect is not currently public. The project is working on intellectual-property protection and patent strategy.

Where we are now

Moving to the live E3-A experiment

E3-A is a calibration experiment for one of the measurement methods used by the research program.

  • 17 Sep 2026E3-A specification frozen
  • 197 / 197Offline engineering checks passed
  • 51 / 51Pre-pilot readiness checks passed
  • 600Live model calls planned for the next experimental run
Next step
  1. First live E3-A run
  2. 600 model calls
  3. Results analysis
  4. Next experimental stage
Research trajectory

Where we are going

We are building a program of sequential experiments in which each next stage builds on the results of the previous one.

  1. 01

    Theoretical construction

    Formalization of key principles, identification of weak points, and technical testing of assumptions.

  2. 02

    Digital experiments

    Testing individual mechanisms and measurement methods under controlled digital conditions.

  3. 03

    Live testing

    Experiments with real, powerful language models.

  4. 04

    Scaling

    More complex conditions, stronger models, and more demanding tests.

  5. 05

    Physical reactor

    Transition to a physical experimental system for controlled work with superintelligence.

Research log

Progress

  1. E3-A specification frozen

    The experimental protocol was fixed before the live run.

  2. E3-A implementation completed and accepted

    Full offline engineering test suite: 197 / 197

  3. Pre-pilot preparation completed

    Readiness checks: 51 / 51

Funding

3 funding applications submitted

  • Manifund

    $32,000

    Public NadIntellect Lab application.

    Minimum funding level: $8,000.

    View NadIntellect Lab application on Manifund

  • BlueDot Rapid Grant

    $8,000

    Application submitted.

    Awaiting decision.

  • Transformative AI Research Grants

    $32,000

    Application submitted.

    Awaiting decision.

Current round
$32,000Funding goal
$8,000Minimum funding level

Funding is intended for live experiments, technical development, compute, technical review, and progression to the next engineering stages.

Get involved

NadIntellect is looking for

  • Funding

    For live experiments, technical implementation, and the next stages of the program.

    Fund the research

  • Technical reviewers

    Experts in AI safety, formal methods, systems engineering, and high-assurance systems.

    Review the research

  • Researchers & engineers

    For the development of the experimental program and the future transition from digital testing to physical implementation.

    Join the research

NadIntellect at a glance

Problem
How to control future superintelligence.
Goal
Build a system for controlled work with it.
Now
Transitioning to the live E3-A experiment.
Next
600 live model calls.
Then
More demanding digital and engineering tests.
Long-term engineering goal
A physical experimental reactor.
Funding
3 applications submitted. Current public funding goal: $32,000.

NadIntellect Lab

Yevhen Haluschak, Founder / Research Program Lead

Evgeniygaluschak1@gmail.com

Funding, technical review, research and engineering.