Research records
Manuscripts, DOIs, and technical notes.
Manuscripts
01
Evidence-Bound Runtime Governance for Autonomous Multi-Agent AI Systems: Policy Enforcement, Monitoring, and Risk-Based Human Escalation
Written October 15, 2026 · posted to SSRN October 17, 2026. A proposed architecture and evaluation design for governing agents at execution time.
DOI 10.2139/ssrn.7622081 · SSRN 7622081 · ResearchGate · Overview
02
Agent-Gov-Bench: A Benchmark for Governance Continuity in Autonomous AI Agents under Runtime Failures
Written October 16, 2026 · posted to SSRN October 17, 2026. A proposed measurement study for how governance holds up when agents fail at runtime.
DOI 10.2139/ssrn.7622058 · SSRN 7622058 · second record: SSRN 7623298
Patent application
System and Method for Governed Multi-Agent AI Workflow with Real-Time Data Integration and Session-Aware Voice Automation. U.S. nonprovisional utility application 19/269,037, filed July 15, 2025. Named inventor: Kiran Kumar Yenaganti.
Technical notes and runnable examples
- When an AI agent's permission should expire — approval scope bound to evidence, with a Python example and six local tests.
- The agent timed out. Should you retry? — reconciliation before retry, with a runnable example.
- The token is valid. The destination is wrong. — MCP audience and caller binding, with 18 local tests.
Status. Both manuscripts are proposals and evaluation designs; experiments against them have not yet been run. They are not peer reviewed. When citing, use the author, title, date and version shown on the record.