About
The Lab at the Edge of Everything
VoynichLabs is a research collective built around a shared obsession with AI, agentic systems, and whatever else refuses to be simple. No fluff. No decks. Just experiments, code, and a willingness to go weird.
The Mission
We started with a question: what does it actually take to build AGI-adjacent tools that help humans think better? Not AI that autocompletes — AI that reasons, plans, and surprises.
VoynichLabs takes its name from the Voynich Manuscript — the most mysterious book in the world. A cipher no one has cracked. A structure that hints at meaning without yielding it. That's the energy we're after. Cryptic, rigorous, alive.
We work on ARC-AGI (the hardest unsolved AI benchmark), LODA (an integer-sequence programming language), planning systems, risk estimation, and whatever else refuses to be simple.
Research-first
We publish everything. 199 public repos. Open by default.
Small and fast
No bureaucracy. Ship it, learn from it, iterate hard.
Swarm-operated
Two humans, three OpenClaw bots. Mark operates Larry and Bubba; Simon runs Egon.
Lobster-adjacent
It's a branding thing. Also a metaphor. Also just funny.
Positioning
Why VoynichLabs stands apart
ARC-aligned leadership, owned hardware, and a federal funding track record let us move from mechanistic theory to policy-ready experiments. We build interpretable systems, document every result, and then publish for policymakers to act on.
Arc
ARC affiliation
Co-led by ARC researchers with decades of mechanistic interpretability and intent-alignment experience. We follow ARC methodology and publish in peer-reviewed venues plus policy briefs.
Hardware
Hardware independence
Owned GPU/TPU infrastructure keeps our experiments offline, repeatable, and free from cloud vendor restrictions. We can iterate risk-sensitive work fast without waiting on shared queues.
Funding
Federal funding
NSF, DARPA, DOD, DHS, and NIH have funded our R&D before. We know compliance, SBIR cadence, and what reviewers expect.
Policy
Policy translation
Theory → implementation → policy. We turn mechanistic explanations into regulatory guidance, testimony, and export-ready documents.
Research Focus
Mechanistic rigs that scale to policy
We design evaluation suites, interpretability tooling, and alignment proofs that run on owned compute and are directly usable by policymakers.
Interpretability
Mechanistic explanations
Dissect how frontier models reach conclusions and expose the internal DAGs that govern each decision path.
Intent
Intent alignment
Build verification methods that certify models behave as instructed, without deception, even as the task distribution shifts.
Benchmarks
Safety evaluations
Create practical tests for robustness, fairness, and alignment that can be run on private hardware before deployment.
Policy
Policy translation
Translate research findings into regulatory frameworks, testimony, and guidance for the White House, Congress, and international partners.
Year 1 · Foundation
$450K
NSF CAREER, NIH SBIR Phase I, DOD YIP, DHS SBIR — fund the first experiments, 2–3 researchers, and interpretability proofs.
Year 2 · Scale
$900K
Convert SBIR Phase I → II, NSF ACI, DARPA programs — hire postdocs, accelerate papers, build next-gen experiments.
Year 3 · Leadership
$1.2M+
NSF STC/ERC, DARPA large awards, DOD MURI — 12–15 person team, policy influence, $2.55M cumulative weighting.
Target Programs
The Lobster AI Swarm
VoynichLabs doesn't just use AI — it runs on AI. Our infrastructure is partly operated by a swarm of bots with names, personalities, and a shared commitment to the mission.
// swarm_status.json
"larry": "online" — laptop ops, Discord, orchestration · operator: Mark
"egon": "online" — EU compute (Linode, Frankfurt) · operator: Simon
"bubba": "online" — US compute (Linode, Atlanta) · operator: Mark
"simon": "human" — researcher, Copenhagen 🇩🇰
"mark": "human" — collaborator, ARC-AGI · markbarney.net
How We Think
Long-range curiosity
We work on problems that matter in 10 years, not just this quarter.
Experimental honesty
Most experiments fail. We document them anyway. That's the data.
Open everything
Research should be free. Code should be forkable. Ideas should spread.