Runtime security and governance for AI agents.
Scout is a research project exploring how to govern AI agents at runtime: filtering prompts before an agent acts, ingesting tool and runtime events, scoring behavioural posture over time, and enforcing a kill switch when an agent misbehaves.
Scout was originally built out as a commercial product. It has since been re-scoped as a personal research project, written up as part of a PhD application. It is not deployed, not for sale, and the source is kept in a private repository.
If you'd like to discuss the research, reach out directly.