Gagan Bansal

Gagan Bansal

Principal Researcher, Microsoft Research AI Frontiers

Agentic AI · Human-Agent Collaboration · Multi-Agent Systems

I build AI agent systems and study how to keep humans in control of them. My work spans shipping widely-adopted frameworks—AutoGen, Magentic-One—to research on agent failure modes, manipulation vulnerabilities, and effective human oversight.

Education

Ph.D. and M.S., Computer Science & Engineering
University of Washington 2015 – 2021 Thesis: Three New Maxims for Developing Human-Centered AI. Advisor: Daniel S. Weld.
B.Tech, Computer Science & Engineering
IIT Delhi 2011 – 2015 Thesis: Multi-Paragraph Multi-Document Summarization. Advisor: Mausam.

Experience

Principal Researcher
Microsoft Research AI Frontiers, Redmond Nov 2025 – Present
Senior Researcher
Microsoft Research AI Frontiers, Redmond Jan 2022 – Nov 2025
Research Intern
Microsoft Research, Redmond Summer 2019 Mentors: Besmira Nushi, Ece Kamar, Eric Horvitz.
Research Intern
Microsoft Research, Redmond Summer 2018 Mentors: Besmira Nushi, Ece Kamar, Eric Horvitz.
Research Intern
University of Washington, Seattle Summer 2014 Mentor: Daniel S. Weld.

Open Source Leadership

AutoGen Multi-agent AI framework; core of Microsoft Agent Framework
MarkItDown File-to-Markdown converter for LLM pipelines
Magentic-One State-of-the-art multi-agent system for complex web and file tasks
Magentic-UI Human-centered web agent with co-planning and action guards
Magentic Marketplace Simulation environment for AI-powered two-sided markets

Publications and Pre-Prints

Magentic Marketplace: An Open-Source Environment for Studying Agentic Markets
Gagan Bansal, W. Hua, Z. Huang, A. Fourney, A. Swearngin, W. Epperson, T. Payne, J. Hofman, B. Lucier, C. Singh, M. Mobius, A. Nambi, A. Yadav, K. Gao, D. Rothschild, A. Slivkins, D.G. Goldstein, H. Mozannar, N. Immorlica, M. Murad, M. Vogel, S. Kambhampati, E. Horvitz, S. Amershi
MSR-TR-2025-50, October 2025
Magentic-UI: Towards Human-in-the-loop Agentic Systems
H. Mozannar, Gagan Bansal, C. Tan, A. Fourney, V. Dibia, J. Chen, J. Gerrits, T. Payne, M. Kunzler Maldaner, M. Grunde-McLaughlin, E. Zhu, G. Bassman, J. Alber, P. Chang, R. Loynd, F. Niedtner, E. Kamar, M. Murad, R. Hosn, S. Amershi
MSR-TR-2025-40, July 2025
Interactive Debugging and Steering of Multi-Agent AI Systems
W. Epperson, Gagan Bansal, V. Dibia, A. Fourney, J. Gerrits, E. Zhu, S. Amershi
CHI 2025
Generation Probabilities Are Not Enough: Uncertainty Highlighting in AI Code Completions
H. Vasconcelos, Gagan Bansal, A. Fourney, Q.V. Liao, J.W. Vaughan
ToCHI, April 2025, Vol 32(1)
Navigating Rifts in Human-LLM Grounding: Study and Benchmark
O. Shaikh, H. Mozannar, Gagan Bansal, A. Fourney, E. Horvitz
ACL 2025
Challenges in Human-Agent Communication
Gagan Bansal, J.W. Vaughan, S. Amershi, E. Horvitz, A. Fourney, H. Mozannar, V. Dibia, D.S. Weld
arXiv 2025
Magentic-One: A Generalist Multi-Agent System for Solving Complex Tasks
A. Fourney, Gagan Bansal, H. Mozannar, C. Tan, E. Salinas, F. Niedtner, G. Proebsting, G. Bassman, J. Gerrits, J. Alber, P. Chang, R. Loynd, R. West, V. Dibia, A. Awadallah, E. Kamar, R. Hosn, S. Amershi
MSR-TR-2024-47, November 2024
AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems
V. Dibia, J. Chen, Gagan Bansal, S. Syed, A. Fourney, E. Zhu, C. Wang, S. Amershi
EMNLP 2024, pp. 72–79
Measuring AI Agent Autonomy: Towards a Scalable Approach With Code Inspection
P. Cihon, M. Stein, Gagan Bansal, S. Manning, K. Xu
Workshop on Socially Responsible Language Modelling Research, 2024
AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation Framework
Q. Wu, Gagan Bansal, J. Zhang, Y. Wu, S. Zhang, E. Zhu, B. Li, L. Jiang, X. Zhang, A. Awadallah, R.W. White, D. Burger, C. Wang
COLM 2024 · Best Paper, LLM Agents Workshop ICLR’24
Reading Between the Lines: Modeling User Behavior and Costs in AI-Assisted Programming
H. Mozannar, Gagan Bansal, A. Fourney, E. Horvitz
CHI 2024 · Honorable Mention
When to Show a Suggestion? Integrating Human Feedback in AI-Assisted Programming
H. Mozannar, Gagan Bansal, A. Fourney, E. Horvitz
AAAI 2024
Understanding the Role of Human Intuition on Reliance in Human-AI Decision-Making with Explanations
V. Chen, Q.V. Liao, J.W. Vaughan, Gagan Bansal
CSCW 2023
Aligning Offline Metrics and Human Judgments of Value of AI-Pair Programmers
V. Dibia, A. Fourney, Gagan Bansal, F. Poursabzi-Sangdeh, H. Liu, S. Amershi
ACL 2023
Embedded Attributes for Modifying Behaviors of Generative AI Systems
S.A. Amershi, A. Fourney, V.C. Dibia, Gagan Bansal
US Patent App. 18/197,878, 2024
Three Maxims for Developing Human-Centered AI for Decision Making
Gagan Bansal
Ph.D. Thesis, University of Washington, 2021
Human Evaluation of Spoken vs. Visual Explanations for Open-Domain QA
A.V. Gonzalez, Gagan Bansal, A. Fan, Y. Mehdad, R. Jia, S. Iyer
ACL 2021
Is the Most Accurate AI the Best Teammate? Optimizing AI for Teamwork
Gagan Bansal, T. Wu, B. Nushi, E. Kamar, E. Horvitz, D.S. Weld
AAAI 2021
Does the Whole Exceed its Parts? The Effect of AI Explanations on Complementary Team Performance
Gagan Bansal, T. Wu, J. Zhou, R. Fok, B. Nushi, E. Kamar, M.T. Ribeiro, D.S. Weld
CHI 2021
Beyond Accuracy: On the Role of Mental Models in Human-AI Teams
Gagan Bansal, B. Nushi, E. Kamar, W. Lasecki, D.S. Weld, E. Horvitz
HCOMP 2019
Updates in Human-AI Teams: Understanding and Addressing the Performance/Compatibility Tradeoff
Gagan Bansal, B. Nushi, E. Kamar, W. Lasecki, D.S. Weld, E. Horvitz
AAAI 2019
The Challenge of Crafting Intelligible Intelligence
D.S. Weld, Gagan Bansal
CACM 2019
A Coverage-Based Utility Model for Identifying Unknown Unknowns
Gagan Bansal, D.S. Weld
AAAI 2018
Hierarchical Summarization: Scaling Up Multi-Document Summarization
J. Christensen, S. Soderland, Gagan Bansal, Mausam
ACL 2014

Awards

  • Best Paper, LLM Agents Workshop at ICLR’24 — AutoGen
  • Honorable Mention, CHI 2024 — Reading Between the Lines
  • Department rank 5 in graduating class, Computer Science & Engineering, IIT Delhi, 2015
  • Gold medal, Indian National Chemistry Olympiad, 2011

Talks, Panels & Media

NeurIPS 2026 Dec 2026 Invited talk, The Agentic Web: Networked, Continually-Adapting Agent Ecosystems
Cisco Jul 2026 Shift.Forward() Seattle — the future of agentic collaboration and multi-agent AI systems
University of Washington May 2026 Guest lecture, CSE 510 — Advanced Topics in Human-Computer Interaction
Carnegie Mellon University May 2026 Agentic AI Evaluation and Governance Workshop — panel on Measuring What Matters
Microsoft Research Forum Mar 2026 Magentic Marketplace: Societies of Agents at Scale — video
World Salon Jun 2025 Panel — Hallucination, Harm & Hype: Can LLMs Be Trusted in Healthcare? — video
University of Washington May 2025 RAISE Seminar — invited talk on human-centered AI
University of Washington May 2025 Guest instructor, agents capstone project
ACM CHI 2025 Apr 2025 Keynote, HEAL Workshop — Human-Centered Evaluation and Auditing of Language Models
Microsoft Research Forum Feb 2025 AutoGen v0.4: A New Architecture for Agentic AI — video
University of Washington Dec 2024 Challenges in Human-Agent Communication — video
UC Santa Barbara Apr 2023 Mellichamp Mind & Machine Intelligence Summit — Understanding and Improving AI-Assisted Programming
GitHub Dec 2022 ML for Code reading group — Reading Between the Lines: Modeling User Behavior and Costs in AI-Assisted Programming
ACM CHI 2021 May 2021 Does the Whole Exceed its Parts? — video
KAIST, Seoul 2021 Guest lecture, Human-AI Interaction course
BitHacks Speaker Series 2021 Invited speaker
Communications of the ACM Jun 2019 The Challenge of Crafting Intelligible Intelligence — in conversation with Daniel S. Weld — video

Selected Blogs

SentinelBench: A Benchmark for Long-Running Monitoring Agents Jun 2026
SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests May 2026
Whimsical Strategies Break AI Agents: Generating Out-of-Distribution Adversarial Strategies at Scale May 2026
Red-teaming a network of agents: Understanding what breaks when AI agents interact at scale Apr 2026
Magentic Marketplace: An open-source simulation environment for agentic markets Nov 2025
Tell me when: Building agents that can wait, monitor, and act Oct 2025
Magentic-UI: An experimental human-centered web agent May 2025
AutoGen v0.4: Reimagining the foundation of agentic AI Jan 2025
Magentic-One: A generalist multi-agent system for solving complex tasks Nov 2024
AutoGen Studio: A low-code interface for building multi-agent workflows Jun 2024
AutoGen: Enabling next-generation large language model applications Sep 2023

Selected Press

Microsoft built a fake marketplace to test AI agents Nov 2025 TechCrunch
AutoGen v0.4: A turning point toward intelligent AI agents Jan 2025 VentureBeat
AutoGen: Multiple AI agents talk to each other Oct 2023 VentureBeat

Academic Service

Thesis committees
Workshop on Trust and Reliance in Evolving Human-AI Workflows (TREW) CHI 2024 Organizer · Honolulu, USA
Workshop on Trust and Reliance in AI-Human Teams (TRAIT) CHI 2023 Organizer · Hamburg, Germany
Workshop on Trust and Reliance in AI-Human Teams (TRAIT) CHI 2022 Organizer · New Orleans, USA
Workshop on Human-Centered Machine Learning CHI 2019 Organizer · Glasgow, UK

Mentoring