February 17, 2026

MI9: An Integrated Runtime Governance Framework for Agentic AI

paper

Wang, Singhal, Kelkar, Tuo · August 2025

  • Proposes FSM-based conformance engines, graduated containment (monitoring to tool restriction to isolation), and per-agent telemetry — all patterns that map directly to OTP supervision trees, process isolation, and :observer
  • The paper’s graduated containment escalation (restrict, then isolate, then kill) is exactly what an OTP supervisor already does with its restart strategies: one_for_one, rest_for_one, one_for_all
  • Validates the BEAM thesis from the other direction: AI safety researchers are independently deriving the runtime governance model that Erlang shipped in production thirty years ago
February 13, 2026

Making Reliable Distributed Systems in the Presence of Software Errors

paper

Joe Armstrong, PhD Thesis · December 2003

  • Joe Armstrong’s PhD thesis defining the principles behind Erlang/OTP: concurrency-oriented programming, process isolation, and error recovery through supervision
  • Argues that reliable systems must be built from unreliable components by isolating failures and encoding recovery strategies declaratively
  • Introduces the six rules of Erlang: isolation, everything is a process, error detection across processes, fault detection is built-in, processes can be upgraded, and failures are handled by other processes