All projects
POCAI ResearchAgentic AI

ACM — Actions Consequential Model

See what an action changes before taking it

Personal research project · builds on MLM

Research proof of concept.

Built on the Meaning Language Model, ACM predicts what an action does: its consequences for the actor, for those it affects and for other actors’ goals — across a blast radius from the place it happens out to society and the earth.

Problem & business context

AI agents are starting to take actions. Before acting, a system should be able to trace who and what an action will affect — including its own planned actions.

My role & contribution

Designed and built the research prototype.

What I built or designed

  • Uses MLM’s Universal Phrase DNA (who does what, to whom, where, when, how and why)
  • Builds a graph of the full action in context and predicts consequences circle by circle
  • Blast-radius layers: site, neighbours, organization, city, country, society and earth
  • Stores every node as a vector to compare with past actions