Pragma Research
Pragma Research is an independent frontier mathematics company/lab solving physics and AGI from first principles. Its central research program is repair networks: local parts compare shared information and repair disagreement until a system settles into a coherent state.
At a Glance
- AI researchers and developers
- Robotics and embodied-agent builders
- Physics and mathematics researchers
- Smart-contract security teams and Web3 developers
- +1 more
AI Tools by Pragma Research
(1)Cadence
Python Library for Equilibrium Learners
Discussions
No discussions yet
Be the first to start a discussion about Pragma Research
Latest News
Observers Are All You Need: the Theory-of-Everything that explains literally everything
All Forces are Repair: Unifying gravity with the strong, weak, and EM forces
Paradigm Shift: Deriving the Entire Universe from Zero Parameters
Everything Clicks Into Place All at Once
Products & Services
Open-source Python research library for a continuing equilibrium brain: a graph of local rate-neuron patches settles under observations, reads actions from the equilibrium, and learns synapses from differences between settles without a backward pass or separate training/inference mode.
Physics research program with preprints, machine-checked Lean proofs, and executable observer models that reconstruct physical structure from records finite observers can share.
Interactive, step-by-step tutorial and toy prover/verifier for STARK proofs, covering execution traces, constraints, polynomials, Merkle commitments, FRI, and zero knowledge.
AI code auditor that builds relation-first knowledge graphs while it investigates.
Market Position
Pragma Research positions itself as an independent frontier-mathematics lab rather than a conventional software vendor, combining first-principles observer-based physics with an AI architecture based on local repair and equilibrium settling. Cadence differentiates from feed-forward and recurrent backpropagation networks through reciprocal settling, continuing state, local plasticity, and learning during life; its adjacent security work competes with symbolic-execution and AI code-auditing tools.
Leadership
Founders
Bernhard Mueller
Security researcher and founder. He worked in vulnerability research at SEC Consult in Vienna (2005–2013), was Principal Security Consultant at Vantage Point Security in Singapore (2014–2017), and was a code auditor/software engineer at ConsenSys Diligence (2017–2020). He initiated OWASP MASTG and MASVS, created Mythril, and received a 2009 Pwnie Award for innovative Symbian research.
Executive Team
Bernhard Mueller
Founder
Independent security researcher and AI/physics researcher; formerly at SEC Consult, Vantage Point Security, and ConsenSys Diligence. He created Mythril, initiated OWASP MASTG/MASVS, and leads Pragma Research's repair-networks, physics, and AI work.
Founding Story
Mueller began the Pragma Research program in 2021 to pursue repair networks—local components that compare what they share and repair disagreement until the whole settles. The same idea was applied to physics through Observer Patch Holography and to AI through Cadence; the work is published as preprints, machine-checked Lean proofs, and open-source code.
Business Model
Revenue Model
The public site presents Pragma Research as an independent research lab and invites investment or partnerships; its public work is distributed through preprints, open-source repositories, tutorials, and research software.
Target Markets
- AI researchers and developers
- Robotics and embodied-agent builders
- Physics and mathematics researchers
- Smart-contract security teams and Web3 developers
- Zero-knowledge proof learners and engineers
- Researching and building agents that learn continuously from live experience
- Embodied AI and robots that sense, act, learn from outcomes, and repair failures
- First-principles physics and observer-based modeling
- AI-assisted smart-contract and code auditing
- Web3 vulnerability hunting and bounty tracking
- Learning zero-knowledge/STARK proof systems interactively