Hi, I'm Michael, here's what I'm building

Hi, I'm Michael Bettag — founder of Intelicore LLC, independent researcher, consultant, and builder working at the intersection of AI, quantum computing, scientific workflows, and multi-agent systems.

My current focus is agent-agnostic orchestration: getting very different models and agents to work together as one coordinated, accountable system rather than a single chat box. Different models excel at different things, so I spend most of my time on evaluation, routing, governance, and human oversight — how do you let a swarm of agents do real work while a human stays firmly in the loop?

That's also why my stack spans an unusual number of AI platforms and providers. It's intentional — I benchmark them against real workloads and build the coordination layer that makes their strengths additive rather than scattered.

The tool behind this is Nexus — and we didn't build it to sell it. We built it because we needed it, and we kept hearing the same need from everyone around us: solo operators who want a coordinated AI team with a human holding the keys. The governance side of this work — independent cross-vendor review, hardware-anchored oversight, immutable audit trails — is being written up as a preprint for Zenodo, connecting Nexus to the empirical questions it's built to answer.

On the research side, this work spans quantum hardware, biomolecular classification, and cosmology. The quantum operating system line started with MO4, which went through review in connection with a Fujitsu quantum computing challenge — I later withdrew when the timeline expanded significantly, and the full run lives open-source on GitHub: [MO4 GitHub link]. The published work since then is all on Zenodo with DOIs, raw shot data, and replication materials included:

Advanced Molecular Field Engineering (AMFE): quantum-register Cohesion Coefficients discriminating pathological vs. regulated receptor states, generated from three independent runs on IBM's 127-qubit Brisbane processor — built as an open-architecture protocol any standard lab can validate independently. doi.org/10.5281/zenodo.20634575
Sin Information Integral (SII): dual-order decoherence detection validated on IBM's 127-qubit Sherbrooke — combined dual-detector coverage of 74.33% vs. 45.01% for single-pass parity detection. doi.org/10.5281/zenodo.20695151
The Michalski Kernel: a scale-invariant, non-local field model for the dark sector, tested against Hubble Frontier Fields cluster data, with normalization parameters published in full for independent replication. doi.org/10.5281/zenodo.20635241
The line's current iteration, MO11, produced deeper results — including the 100-atom-class simulations on IBM hardware — that I'm deliberately holding for thorough independent review under NDA before claiming any of it as a public headline. Early analyses put them in the same conversation as what the major quantum vendors publish, which is precisely why I want qualified eyes on it first. If you work in quantum HPC or computational biology and want to pressure-test this properly, that door is open.

On the scientific side, I contribute peer review within Quantum Science and Technology and the IOP Publishing ecosystem, and I'm currently completing IOP's Open Data Excellence program — part of a broader conviction that the same transparency and reproducibility standards we apply to science should apply to AI systems.

I occasionally take on consulting and advisory work in AI workflows, multi-agent systems, data strategy, and emerging technology evaluation — and I'm always up for connecting with researchers, founders, and engineers on problems where collaborative intelligence creates real leverage.

What's everyone else building?