LLMediator Receives Major Funding to Transform Online Mediation

14 April 2026

The LexUM Chair on Legal Information is pleased to announce that it has received funding from VADO and the Canada First Research Excellence Fund for the project Exploring the Capability of an LLM to Act as a Mediator in Complex Mediation Cases, led by Professor Karim Benyekhlef.

The Problem

Access to justice remains a global challenge. Every year, millions of citizens face legal disputes without the necessary resources to resolve them effectively. Conflicts between landlords and tenants, consumer complaints, small claims: these disputes are both very common and largely overlooked by the traditional judicial system.

Online dispute resolution (ODR) is emerging as a promising alternative to traditional legal proceedings. However, existing platforms generally lack support mechanisms tailored to unrepresented parties—leaving a significant gap that e-justice research seeks to fill.

 

Our Solution: LLMediator

LLMediator is a Large Language Model (LLMs) powered online dispute resolution (ODR) platform designed to assist or augment human mediators in resolving disputes between two parties. Developed by the Cyberjustice Laboratory at the University of Montreal, the platform uses the capabilities of most advanced LLMs, such as GPT-4o-vision, to support negotiation and mediation processes, particularly in high-volume, low-intensity legal disputes.

LLMediator provides a structured interface for three-way interactions: two disputing parties and one neutral mediator. It integrates LLMs to analyze conversation dynamics, detect emotional shifts, generate intervention suggestions. The goal of LLMediator is to make dispute resolution more accessible, efficient, and fair. We designed LLMediator to be a “co-pilot” for human mediators, empowering them with AI-driven tools.

 

Watch a short demo of the platform below:

 

Next Steps

Thanks to funding from IVADO and the Canada First Research Excellence Fund, the Cyberjustice Laboratory team will continue to improve LLMediator and will focus in particular on:

  • Building a more extensive database that will include numerous dispute cases containing not only text, but also images, audio files, and documents such as receipts;
  • Adding a tool capable of drafting a formal agreement between the parties by scanning the conversation and identifying the points on which they have agreed (amounts to be paid, deadlines, etc.);
  • Exploring a new model for a mediation agent, which will serve as an intermediary, transmitting messages and proposals from one side to the other, without the parties needing to be present in the same chat room at the same time.

This content has been updated on 14 April 2026 at 17 h 43 min.