Co-founder, Agent2.AI · SF Bay Area
Linghao Yang (James)
I build practical AI products and the ops behind them — and publish research on multi-agent systems along the way.
01About
I love the loop of shipping, talking to customers, and iterating quickly — and I thrive in teams that stay close to real user problems.
I'm the co-founder of Agent2.AI, where we built a general multi-agent system that helps users orchestrate the right tools and workflows to automate real work.
Before going full-time on my startup, I was an Associate at Morgan Stanley in the FRTB PMO and Risk Quant org — coordinating 20+ global teams and keeping a major regulatory program on deadline.
I also invest in AI: I've invested in xAI and Anthropic, and I'm an LP in a small VC fund.
Outside of work I dance, play tennis, lift, and do magic tricks. In my freshman year at Penn State I founded ACEs Dance Crew, the school's largest dance club, and I've competed in professional breaking battles.
If you're in SF / the Bay Area, I'm always happy to connect — AI, VC, stocks, breakdancing, or Pokémon, over boba :)
02Selected work
Things I've built & shipped
Featured · Agent2.AI
General Multi Agent
A Super Agent that turns goals into finished work — competitor research, slide decks, web pages, cross-tool workflows — by splitting the task and routing each part to the best executor.
Watch the demo →
Interactive Product Demo ↗
An AI-guided product playground that turns static SaaS demos into live, hands-on experiences.
Resume Tailor AI ↗
Tailors your resume to any job description in minutes — built after watching qualified people get filtered out.
Routinli ↗
An AI supplement coach that turns wearable health data into clear, actionable daily routines.
ReplAI ↗
An AI keyboard that suggests context-aware replies, so dating conversations keep flowing.
03Research
Papers
My research explores how AI agents can collaborate more effectively in complex, real-world environments. Across my work, I study multi-agent coordination, trust-aware information sharing, role consistency, adaptive routing, long-term memory, and structured knowledge reasoning, with the goal of making LLM-based agent systems more reliable, efficient, and practical. Full list on Google Scholar.
- i MIN-Trust: A Minimum Necessary Information Trust Orchestration Framework for Multi-Agent Collaboration
- ii Detecting and Repairing Role Drift in Multi-Agent Collaboration with Lightweight Protocols
- iii Budgeted Multi-Agent Routing: Adaptive Role Assignment and Communication Compression for Efficient LLM-Agent Collaboration
- iv Cognitive Modeling for Long-Horizon Agent Learning via Integrated Long-Term Memory and Reasoning
- v A Multi-Agent Large Language Model Framework for Marketing Decision-Making with Auditable Attribution Analysis
- vi Self-Supervised Representation Learning and Structured Knowledge Mining for Heterogeneous Multi-Source Data
04Background
Experience
-
2024 —Agent2.AI
Co-founder
-
2022 – 24Morgan Stanley
Associate, FRTB PMO & Risk Quant
-
2017 – 20Penn State
Teaching Assistant
-
2019Accenture
Summer Analyst
Education
-
2024University of Pennsylvania
M.S. CIT — left to build Agent2.AI
-
2020 – 22University of Chicago
M.S. Applied Data Science
-
2016 – 20Penn State
B.S. Finance
05Connect
In the Bay? Let's grab boba.
Happy to talk AI, startups, investing — or breaking and Pokémon.