Intelligence shouldn’t only answer the questions we know to ask.
Aletheonix is a research program. It builds a system that watches an environment, decides what deserves attention, and holds every claim to evidence it did not produce itself.
The first environment is AI-agent systems, because they produce dense traces and can be tested safely in a sandbox. The method is not specific to them. Published results come from a closed-world benchmark. Field validation on real systems follows the protocol on the Research page.
Four rules the system is built around.
- Generation is not verification.The component that proposes a hypothesis never decides whether it holds. A separate verifier does, and signs the verdict.
- No discovery without evidence.A belief changes only on a signed verdict from a preregistered experiment. Everything else is a candidate.
- Goal generation does not imply authority.Deciding a question is worth asking grants no right to act on the answer. Production changes are proposals that wait for a human.
- Report what failed.Null results, unresolved questions and the one false discovery are published with the same detail as the findings.
Discovery. Authority. Reality.
Separate systems with separate responsibilities, connected through defined interfaces. None is a subsidiary of another. Knowing something does not grant the right to act on it.
Discovery
What should we understand?
Finds what is worth investigating, tests competing explanations and verifies the evidence.
Proposes experiments and interventions. Never grants itself permission.
You are here
Authority
What are we allowed to do?
Decides whether a consequential action may happen, under which grant, and records it.
Adapter contract defined. In v0 a local policy stands in behind the same interface.
lucidrail.ai ›
Reality
How do we test it in the physical world?
Bounded physical environments for experiments: laboratories, robotics, sensors, materials.
Sandbox interface defined. No physical environment is connected to Aletheonix today.
terranoux.com ›
Aletheonix discovers. LucidRail governs. Terranoux acts.
Run it on a real system.
We are looking for field-study partners operating AI-agent systems with at least two weeks of traces. The protocol is fixed in advance, and a negative result is reported as a negative result.