Research portal · October 2026
Overview
This research investigates how a risk-aware multi-agent AI system can monitor, coordinate, and automate routine private security operations while keeping human supervisors responsible for high-risk decisions.
This portal is the working record of the study. The public introduction is on asbos.org. The main question is: How can a risk-aware multi-agent AI system automate routine private security operations using real-time operational data while preserving human oversight for high-risk decisions?
The study is at proposal stage. What follows is a design and an evaluation plan. It is not a report of system performance.
What the study is expected to contribute
These are aims. They are not claims that the work has already produced them.
- A domain-specific multi-agent architecture for private security operations
- A risk-based model for setting the level of AI autonomy
- An event-driven method for coordinating operational agents
- Realistic private-security test scenarios
- A framework for human escalation and oversight
- Experimental evidence comparing multi-agent, single-agent, and rule-based approaches
Budget
The proposal estimates $2,000 for a prototype and for testing: $900 for development and $1,100 for experiments. That figure is an estimate, not a record of money spent. The line items are on the proposal page.
Pages
- Proposal
The October 2026 problem statement, architecture, scope, ethics, and budget estimate.
- Methodology
How a rule-based system, a single agent, and ASBOS will be compared, and which measures are named.
- Experiments
The four planned scenarios and the protocol for each. No runs have been completed.
- References
Works cited in the proposal. This is not a full literature review.
- Downloads
The proposal PDF, the LaTeX source, and a BibTeX file.
Record of the work
| Work | State |
|---|---|
| Problem statement, research gap, and research questions | Specified |
| Five-agent architecture and coordination workflow | Specified |
| Autonomy levels 0–4 | Specified |
| Four operational scenarios | Specified |
| Comparative method and named evaluation measures | Specified |
| Research budget estimate | Specified |
| Event-driven prototype | Planned |
| Simulated or de-identified operational environment | Planned |
| Rule-based, single-agent, and multi-agent runs | Planned |
| Scoring rules for the evaluation measures | Not yet defined |
| Measured results | Not available |