Nesean Crofford
Operator background, production systems, and the research that came out of building them.
I spent seven years inside PE-backed portfolio companies, mostly operations. I've built and shipped production platforms since. Most operational cost turns out to be people reading documents, and that is the part I work on.
How I think about the work
Most software makes you choose between powerful and simple. That trade is a design failure, not a law of nature.
Intelligence you rent gets more expensive the more you use it. Intelligence you own gets cheaper.
The same input should produce the same output. Every time, provably, years later.
A system that guesses when it is unsure is worse than one that stops and asks.
Current focus
Two layers that make operational work run unattended and prove afterward that it ran correctly. Together they are the foundation of Parity.
Featured research
Empirical work examining failure modes and reliability gaps in autonomous systems. As AI systems move from generating outputs to taking actions, reliability becomes a systems problem, not a model problem.
March 2026
An 850-run study showing that outcome-based evaluation misrepresents agent behavior in workflow-realistic settings, failing to distinguish true failure, invalid execution, and correct remediation.
March 2026
A 3,900-run study demonstrating that inference-based systems fail to satisfy core execution properties (correctness, determinism, temporal fidelity, and governance) while compiled execution achieves perfect reliability by removing inference from the runtime path.
March 2026
An 850-run study showing that no existing system architecture (LLMs, structured systems, or agents) produces reliable execution, and that external governance, applied as a wrapper, enforces constraints but renders systems non-functional.
March 2026
A 750-run study demonstrating that runtime governance can operate as domain-invariant infrastructure, producing consistent enforcement, tamper-evident records, and fully reconstructable decision trails across policy domains and model providers.
April 2026
A 39-workflow study showing that natural-language enterprise workflow descriptions can be compiled into deterministic, governed execution artifacts with 100% compilation, execution, and replay determinism.
Essays
March 2026
Organizations do not run on reasoning alone. They run on guarantees—verification, repeatability, and policy enforcement—that ensure work behaves predictably at scale.
March 2026
Why guarantees in complex systems are enforced by infrastructure rather than reasoning systems themselves.
March 2026
How intelligent systems move from reasoning to reliable action through separation of reasoning, guarantees, and execution.