Multi-Agent Systems | Yiyang "Diana" Wang

🌱 New preprint — CultivAgents

Fri, 22 May 2026 00:00:00 +0000

We released , a relationship-centered multi-agent system that grounds personalized gardening support in users’ skills, local ecologies, and cultural contexts. Read the .

CultivAgents

Fri, 22 May 2026 00:00:00 +0000

Gardening supports well-being, cultural continuity, and food autonomy, yet most digital tools give generic advice that ignores a gardener’s skills, local ecology, and cultural context. CultivAgents is a relationship-centered multi-agent system, grounded in an ethics of care, that coordinates three specialized agents — an Experience Agent attuned to skill level, an Environmental Agent grounded in local and seasonal conditions, and an Ethnobotanical Agent that connects plants to cultural knowledge.

My role. I led this project, designing the system and a three-phase mixed-methods evaluation with domain experts, HCI researchers, and community gardeners. Participants reported increased confidence, motivation, and trust in acting on AI advice, and valued the complementary perspectives of the agents — while also surfacing limits in cultural specificity and agent coordination that motivate future work.

CultivAgents: Cultivating Relationship-Centered Multi-Agent Systems for Personalized Gardening

Fri, 22 May 2026 00:00:00 +0000

Abstract

Gardening supports well-being, cultural continuity, and food autonomy, yet digital tools often give generic advice that overlooks gardeners’ skills, local ecologies, and cultural contexts. We introduce CultivAgents, a relationship-centered multi-agent system grounded in ethics of care that coordinates an Experience Agent, an Environmental Agent, and an Ethnobotanical Agent to deliver personalized, socio-culturally grounded support. A three-phase mixed-methods study with experts, HCI researchers, and community gardeners found that CultivAgents helped gardeners translate interest into situated action and increased their confidence, motivation, and trust in acting on AI advice.

🎉 CompanionCast accepted at the ACM CHI 2026 Workshop

Sun, 08 Mar 2026 00:00:00 +0000

— a framework for orchestrating multiple specialized AI agents as social collaborators in live shared experiences — was accepted at the ACM CHI 2026 Workshop on Human-Agent Collaboration. The work builds on my Ph.D. research internship at Dolby Laboratories.

📝 New preprint — MASCOT

Tue, 20 Jan 2026 00:00:00 +0000

We released , a multi-agent framework for multi-perspective socio-collaborative companions that improves persona consistency and reduces redundant, sycophantic dialogue. Read the .

MASCOT

Tue, 20 Jan 2026 00:00:00 +0000

Multi-agent systems are increasingly used as companions for emotional and cognitive support, but they often drift into generic assistant behavior (persona collapse) or pile on redundant, agreeable dialogue (social sycophancy). MASCOT is a framework for multi-perspective socio-collaborative companions that addresses both problems with a bi-level optimization strategy: a persona-aware behavioral alignment pipeline that gives each agent a distinct identity, and a collaborative dialogue optimization step that pushes the group toward complementary, productive discourse.

My role. I led this project — formulating the framework, building the system, and designing the evaluation. Across in-domain and out-of-domain settings, MASCOT improves persona consistency by up to +14.1 and social contribution by up to +10.6, verified through human evaluation, multiple LLM judges, and automatic metrics.

MASCOT: Towards Multi-Agent Socio-Collaborative Companion Systems

Tue, 20 Jan 2026 00:00:00 +0000

Abstract

Multi-agent systems are emerging as promising socio-collaborative companions for emotional and cognitive support. However, existing systems frequently suffer from persona collapse, where agents revert to generic assistant behaviors, and social sycophancy, where agents produce redundant, non-constructive dialogue. We propose MASCOT, a multi-agent framework for multi-perspective socio-collaborative companions, with a bi-level optimization strategy that harmonizes individual identities and collective discourse. MASCOT improves persona consistency by up to +14.1 and social contribution by up to +10.6 across in-domain and out-of-domain settings.

CompanionCast

Thu, 11 Dec 2025 00:00:00 +0000

Shared experiences are central to social connection, yet media consumption is increasingly solitary. CompanionCast is a general framework for orchestrating multiple specialized AI agents as social collaborators within a live shared context. It combines multimodal event detection, rolling context caching for better grounding, and spatial audio to enhance co-presence, with an LLM-as-a-judge step that refines the agents’ dialogue.

My role. I led this work, developed during a Ph.D. research internship at Dolby Laboratories. We validated CompanionCast in the domain of sports viewing; pilot studies with soccer fans showed significant improvements in perceived social presence and emotional sharing compared to viewing alone. The work was accepted at the ACM CHI 2026 Workshop on Human-Agent Collaboration.

CompanionCast: Toward Social Collaboration with Multi-Agent Systems in Shared Experiences

Thu, 11 Dec 2025 00:00:00 +0000

Abstract

Shared experiences are fundamental to social connection, yet media consumption is increasingly solitary. We present CompanionCast, a general framework for orchestrating multiple specialized AI agents as social collaborators within a live shared context, integrating multimodal event detection, rolling context caching, and spatial audio to enhance co-presence. Validated through sports viewing, pilot studies with soccer fans show that CompanionCast significantly improves perceived social presence and emotional sharing compared to solitary viewing. This work was conducted in part during a Ph.D. research internship at Dolby Laboratories and accepted at the ACM CHI 2026 Workshop on Human-Agent Collaboration.

🔬 Started a Ph.D. research internship at Dolby

Mon, 19 May 2025 00:00:00 +0000

I began a Ph.D. research internship at Dolby Laboratories, designing a multi-agent conversational AI framework with spatial audio to reimagine sports streaming through AI companions for social co-viewing.