PROJECTS / RESEARCH IN PRACTICE

Questions, made concrete.

My published work, ongoing research, and the questions I want to pursue next. Each project connects to a different part of my research agenda.

RELICCONCEPT / RESEARCH SYSTEM
Agent OrganizationCapabilities become organizational when they are encoded into persistent structure, responsibility, and governance. RESPONSIBILITY / GOVERNANCEPLANEXECUTEREVIEWPERSISTENT ORGANIZATIONAL STATEMEMORY / REUSABLE CAPABILITYGATEMEMBERS CHANGE. ORGANIZATIONAL STATE PERSISTS.
Governed protocols · capability beyond members
Agent organizationsCurrent research

RELIC

Governed protocol formation, runtime binding, and capability that persists across work and member turnover.

Organizational capabilityExecutable protocolsGovernance
Explore project
ProtocolBenchCONCEPT / RESEARCH SYSTEM
ONE TASK · COMPLEMENTARY WORKSHARED DELIVERABLEContributions → integration → shared outcomeINTERFACES · ROUTING · MESSAGE EXCHANGEProtocolRouterA2AACPANPAgoraRequests ↔ responses · protocol-dependent paths
Protocol interfaces · routing · message exchange
Multi-agent communicationICML 2026

ProtocolBench

Comparing A2A, ACP, ANP, and Agora across success, latency, overhead, and failures—with learned selection through ProtocolRouter.

Protocol evaluationProtocolRouterFailure robustness
Explore project Paper Code
ModelingAgentCONCEPT / MODELING WORKFLOW
FROM QUESTION TO MODELING REPORTMODELINGAGENTTool use · retained context · iterative refinement01ProposeModeling ideas02Find dataGround the model03ModelFormulate & compute04WriteExplain the resultsCritique → refine → continue
One modeling workflow · tools · iterative refinement
Agentic problem solvingFindings of EMNLP 2025

ModelingAgent

A tool-using modeling workflow that develops ideas, gathers data, computes models, and refines a report through critique. ModelingBench supplies the tasks; ModelingJudge evaluates the results.

Mathematical modelingTool useOpen-ended tasks
Explore project Paper Code
MultiAgentBenchCONCEPT / RESEARCH SYSTEM
ONE TASK · COMPLEMENTARY WORKSHARED DELIVERABLEContributions → integration → shared outcomeINTERFACES · ROUTING · MESSAGE EXCHANGEProtocolRouterA2AACPANPAgoraRequests ↔ responses · protocol-dependent paths
Complementary contributions · a shared outcome
Collaboration & competitionACL 2025

MultiAgentBench

MARBLE’s interactive environments evaluate task outcomes and coordination milestones across different communication structures and planning strategies.

Collaboration & competitionMilestone evaluationMARBLE
Explore project Paper Code
Human–Agent OrganizationCONCEPT / RESEARCH SYSTEM
Hybrid Human–Agent OrganizationHumans interface with an organization through judgment, authority, and translation layers. RESPONSIBILITY / GOVERNANCEPLANEXECUTEREVIEWPERSISTENT ORGANIZATIONAL STATEMEMORY / REUSABLE CAPABILITYGATEHUMAN WORK UNITHUMANLIAISONTRANSLATIONDELEGATED AGENTDELIVERABLE → COMMON REVIEW
Human office · liaison · common governance
RELIC · HCI extensionExploratory research

Human–Agent Organization

RELIC’s P1–P3 interface designs: direct participation, transparent liaison, and liaison-first interaction within shared organizational governance.

P1 / P2 / P3Human officeShared governance
Explore project
Agent SocietyCONCEPT / RESEARCH SYSTEM
Agent SocietyBeyond a single organization lies a larger ecology of interacting organizations, roles, norms, and institutions. RESPONSIBILITY / GOVERNANCEPLANEXECUTEREVIEWPERSISTENT ORGANIZATIONAL STATEMEMORY / REUSABLE CAPABILITYGATESHARED INFRASTRUCTUREIIIIIIRESPONSIBILITY / GOVERNANCEPLANEXECUTEREVIEWPERSISTENT ORGANIZATIONAL STATEMEMORY / REUSABLE CAPABILITYGATECOMMONSABNORMS / RELATIONS / INSTITUTIONSRESEARCH ORGANIZATIONGOVERNANCE ORGANIZATION
Organizations · shared infrastructure · institutions
Long-term research agendaFuture direction

Agent Society

A long-term agenda on inter-organization coordination, shared norms, and human participation—not a completed society-scale system.

InstitutionsInter-organization relationsShared norms
Explore project
OPEN CONVERSATION / Projects

Continue the conversation.

Questions and perspectives on this page are welcome.

Prefer a private conversation?

Loading comments…

Leave a public comment

This conversation belongs to Projects. Comments appear only after review. For contact details or personal matters, use the private message form.

Public · reviewed before appearing