// Lindwurm · Research program

Research program & FAQ

The funded 16-month timeline — and answers to the questions we hear most about the program.

The 16-month research program

Months 1–4 — Substrate scaling & input dynamics

Scaling the neural fabric toward the target structural complexity; research on dynamic input-filter behaviour under load.

Months 4–9 — Learning-loop robustness

Stability of homeostatic learning at scale; generalization studies on held-out task families.

Months 9–14 — Independent benchmarking & evaluation methodology

Public-benchmark campaigns; development and publication of the complementary evaluation methodology; external audit access.

Months 14–16 — Consolidation

Reproducibility, documentation, audit-readiness, and preparation of the first partner pilots.

Frequently asked questions

Who owns the results?

Method, architecture and code developed in the program are documented and protected; on incorporation, all IP transfers to the company. Public communication covers benchmark results and methodological principles — implementation details stay confidential.

Why grants instead of venture capital?

Because the current priority is measurement, not growth. Non-dilutive funding lets us optimize for honest results; once those results are independently verified, any later investor conversation starts from evidence, not promises.

What if the benchmark results plateau?

Even partial, independently verified results have value: they show where the approach stands and what to improve next. The method and the evaluation infrastructure remain — and every future decision rests on facts.

Why is the code closed?

Because sample-efficient, continuously learning systems are dual-use by nature — capabilities of this class could be misused, including in autonomous weapons systems. Until adequate safety and ethical mechanisms exist, the code stays closed. Our openness runs through a different channel: public benchmarks and independent audit access. Ethics →

How can we work with you?

Research and industry partners join through the application directions (Applications →); independent auditors — academic groups, accredited labs, ML-evaluation organizations — through the audit protocol. Both start at the contact form.

A hard problem, a pilot domain, a collaboration?

Working on a hard reasoning problem, or want to follow the program? Let's talk.

Request a consultation