LEXFLOW 2027
Workshop on Agentic Workflows for
Human-AI Legal Reasoning
Co-located with the 32nd International ACM Conference on Intelligent User Interfaces (ACM IUI 2027)
Helsinki, Finland · February 8, 2027
01 / About
About the Workshop
LEXFLOW 2027 explores how intelligent user interfaces can support people in understanding, supervising and steering agentic legal reasoning workflows.
As AI systems become increasingly embedded in legal research, drafting, compliance, fact verification and other forms of legal decision support, users no longer interact only with isolated model outputs. Legal work may unfold across multiple steps, tools, evidence sources, intermediate decisions and moments of human intervention - increasingly involving multiple specialised AI agents coordinating with one another.
Legal professionals remain responsible for evaluating evidence, interpreting uncertainty, identifying errors and deciding whether an AI-supported conclusion can be trusted or acted upon. Yet many current legal AI interfaces remain output-centred, offering limited visibility into task decomposition, consulted sources, intermediate choices, uncertainty, agent contributions or opportunities for meaningful human input.
LEXFLOW asks how legal AI can move beyond producing answers towards supporting process-level supervision. We are interested in interaction mechanisms that make plans, evidence, provenance, uncertainty, intermediate reasoning, agent coordination and intervention points visible and actionable.
The workshop brings together researchers and practitioners from HCI, IUI, AI and Law, legal AI, Human-Centered AI and related communities. Through paper and demo presentations, co-design and collaborative synthesis, we will work towards reusable design patterns, evaluation questions, workflow representations and a shared research agenda for legal AI that is more inspectable, steerable, verifiable, contestable and accountable.
The workshop will include
- Invited Talk: A keynote-style talk on human-centered legal AI and emerging directions for agentic legal reasoning.
- Paper & Demo Presentations: Lightning presentations showcasing research, systems, and early-stage ideas from the community.
- Workflow Gallery: Participants will share and compare Workflow Cards describing legal tasks, AI agents, evidence sources, intervention points, and failure modes.
- Co-design Sprint: Small groups will collaboratively design interface concepts for more inspectable, steerable, and accountable legal AI workflows.
- Live Critique: Groups will present their concepts and receive structured feedback from organizers and participants.
- Collaborative Synthesis: The workshop will conclude by identifying reusable design patterns, open research questions, and opportunities for future collaboration.
02 / Call for papers
Call for Papers
We invite submissions that explore the design, development, evaluation, and use of intelligent interfaces for agentic legal reasoning workflows. We welcome work from HCI, IUI, AI and Law, LegalTech, human-centered AI, and related communities, including empirical studies, system and interface designs, conceptual frameworks, position papers, and demonstrations.
Submissions should address topics relevant to human oversight and interaction with AI-supported legal reasoning, such as process visibility, evidence provenance, verification, uncertainty communication, intervention and steering, multi-agent coordination, accountability, and responsible legal AI.
Submission Types
Short Papers — 5–9 pages
We welcome position papers, early-stage research, conceptual contributions, design explorations, and work-in-progress that can stimulate discussion and collaboration at the workshop.
Research Papers — 10+ pages
We welcome mature research contributions, including empirical studies, system development and evaluation, theoretical or methodological work, and substantial investigations of human-AI interaction in legal reasoning workflows.
Demo Papers — 5–9 pages
We welcome demonstrations of interactive systems, prototypes, tools, or other technical artifacts related to intelligent legal interfaces and agentic legal workflows.
Page counts refer to the standard CEURART manuscript format. References may be included in addition to the main paper length where permitted by the final proceedings guidelines.
Formatting and Submission
Submissions should be prepared using the CEURART template and submitted in PDF format through EasyChair.
All submissions should present original work that is not simultaneously under review elsewhere. Each submission will be reviewed by at least two members of the workshop Program Committee based on relevance, originality, clarity, and potential to contribute to workshop discussion.
Accepted authors will be expected to attend the workshop and present their work. Authors may also be invited to prepare a lightweight Workflow Card describing the legal task, human and AI roles, evidence sources, reasoning workflow, intervention points, uncertainty, and potential failure modes represented in their work.
Accepted papers will be published in the workshop proceedings in accordance with the IUI 2027 workshop publication process and CEUR-WS requirements.
Submit via EasyChair ↗04 / Topics
List of Topics
Topics of interest include, but are not limited to:
- Agentic legal reasoning workflows
- Human-AI legal reasoning and collaboration
- Intelligent user interfaces for legal AI
- Process-level supervision and human oversight
- Inspectable and transparent reasoning processes
- Steering and intervention in AI-assisted legal workflows
- Evidence provenance and source verification
- Uncertainty communication in legal AI
- Contestability and correction of AI-supported reasoning
- Accountability and responsibility in human-AI legal work
- Multi-agent coordination in legal reasoning systems
- Workflow decomposition, planning, and intermediate decisions
- Interaction design for legal AI systems
- Evaluation methods for agentic and intelligent legal interfaces
- Reusable design patterns for human-centered legal AI
05 / Organisers
Organisers
02
Yuyao Zhang ↗
PhD Student · HKUST
Researching Human-AI Interaction, agentic AI, and intelligent interfaces for legal workflows.
03
Yidan Huang ↗
PhD Student · HKUST
Working on human-centered legal AI, legal reasoning, and GenAI-supported legal workflows.
04
Chuxue Cao ↗
PhD Student · HKUST
Researching LLM safety and alignment, formal reasoning, and trustworthy AI systems, including safety evaluation from a legal perspective.
05
Yujin Zhou ↗
PhD Student · HKUST
Researching LLM-based agents and multi-agent systems for complex legal reasoning and dispute resolution.
06
Guoying Lu ↗
PhD Student · HKUST
Researching legal AI, reliable LLMs, and the societal and governance implications of generative AI.
01
Sirui Han ↗
Assistant Professor · HKUST
Working on AI, LegalTech, law and finance, and human-centered AI systems.
06 / Contact
For sponsorship, collaborations, media, or questions, reach out to us:yzhang075@connect.ust.hk