AI & Decision Design

A full-day seminar and hands-on workshop to analyze recent AI experiences in decision-making situations, then co-design an AI agent to support a budget decision in a professional context.

Practical information

Date: 30 January 2026 - 9:00-17:00

Venue: Duplex (7th floor), Campus SNCF Étoiles, 2 Place aux Étoiles, 93633 La Plaine Saint-Denis

Registration: Free - limited seats (reservation required)

Date

Friday January 30, 2026
Table of contents

About

This seminar explores how generative AI is used within decision-making situations, and how AI agents can be designed to fit organisational contexts (information constraints, responsibilities, evaluation criteria).

The day combines introductory talks with a hands-on workshop: participants first analyse recent AI-assisted decision experiences and derive transferable properties of "ideal" AI support. In the afternoon, groups co-design an AI agent to support a budget decision in a professional setting, by formalising its context, resources, capabilities, workflow, constraints, and expected outputs before implementation. 

Program

9:00 – 9:40 : Welcome & workshop introduction

Coffee, arrival, and start of the day. Framing, practical overview, and feedback from the previous seminar (formats, strengths/limits).

9:40 – 10:35 : Introductory talks

Decision-making influence factors; AI uses and CSR at SNCF; GenAI rollout at BPCE; AI as a medium (research).

10:45 – 11:15 : Describe an AI experience (individual)

Each participant documents a recent decision-making experience involving AI, using the "AI Experience" worksheet.

11:15 – 11:45 : Share and surface properties (collective)

Short sharing of experiences, then formulation of desired properties for AI support (what it should do and guarantee).

13:00 – 13:30 : Co-design workshop & AI agent example

Workshop briefing: co-design an AI agent for a budget decision in a professional context. Presentation of an example agent (BPCE) and group formation.

13:30 – 14:30 : AI agent design brief

Framing the context and need; mapping an AI agent ecosystem; defining the properties and characteristics of the selected agent.

14:30 – 15:00 : Documentary corpus

Collecting and structuring sources (open sources), converting content into Markdown, and preparing the knowledge base.

15:00 – 16:00 : System prompt

Writing the system prompt, configuring the agent, creating an avatar, testing and iterating.

16:00 – 16:15 : Agent testing

Assessing the agent's outputs against the brief produced by the group.

16:15 – 16:45 : Group presentations & discussion

Cross-group sharing (short pitch + discussion): tensions, contributions, limitations.

16:45 – 17:00 : Wrap-up

Closing and possible next steps.

17:00 – 19:00 : Farewell drinks

Team

  • Giulia Marcocchia

    Giulia Marcocchia is associate professor in Management Sciences. She holds the "Junior Professor Chair" in Design and Management

  • Annie Gentes

    Annie Gentes is professor of Information and communication sciences and design. She is the director of research of CY School of Design, CY University.

  • Anthony Ferretti

    Independent interface designer - Co-founder of Praticable & Collectif Bam

  • David Serrault

    David Serrault is a doctoral candidate and a designer. As the Director of Digital Design at the BPCE Group, he is concurrently conducting a research on the impact of artificial intelligence on the customer experience of retail banking services.

Workshop materials

  • AI Experience analysis (A3, single-sided)

    One sheet per participant to document a decision-making situation involving AI and derive transferable properties for an "ideal" AI support. 

  • AI agent design brief (one kit per group)

    One cover sheet + a set of A3 worksheets to structure: Context, Ecosystem, Purpose, Resources, Capabilities, Process, Properties, and Outputs (good vs. bad examples), to make design decisions explicit before implementation.

Outputs

    Individual : A completed "AI Experience" sheet (experience description + ideal AI properties).

    Per group :

    • A mapped AI agent ecosystem and a selected agent.
    • A complete agent design brief (resources, capabilities, workflow, constraints, expected outputs).
    • A structured documentary corpus (including Markdown conversion) and a system prompt.
    • A configured agent, tested against the group's brief and presented in plenary.

    Research Chair

    This seminar is part of the Decision Design Research Chair, which aims to invent tools and instruments that support more creative and fair decision-making processes. 

    • Decision design

      We want to invent the tools and instruments that will support creative and fair decision-making processes.