'Architecting the Future The Enterprise Revolution through GenAI' | RAISE Summit 2024 | Paris

Panel Context & Objective

• The session, moderated by Charles-Édouard Boué of Adagia Capital, shifts focus from generative AI "test and learn" phases to enterprise-scale deployment and value realization.
• Five panelists (Florence Verzelin, Emily, David Martineau, Gilles Babinet, Victoria Veller) discussed GenAI's impact on growth, productivity, sustainability, and public policy.

Enterprise Transformation & Use Cases

• Gucci: Leveraged GenAI to triple e-commerce growth by optimizing data strategy around customer insights.
• Accor: Achieved a 37% increase in conversion rates for its 20-million-client loyalty program using GenAI to manage massive data volumes.
• Life Sciences: Virtual twins enable virtual screening of molecules against viruses, accelerating R&D timelines and reducing treatment development costs.
• Dassault Systèmes: GenAI code generation accelerates developer productivity on routine tasks, maintaining competitive speed in software updates.
• Bouygues Construction: Utilized GenAI with virtual twins to optimize building energy efficiency by 40–60%.
• Eleven Labs: Demonstrated a shift from cost reduction to scope expansion, enabling hobbyists to produce audiobooks and enterprises to dub content into multiple new markets instantly.

Strategic Frameworks & Roadmaps

• David Martineau's "Three-Category Matrix":
  • Category 1: Embedded AI tools (e.g., Salesforce, Photoshop) where competitive advantage is neutralized as all competitors adopt them.
  • Category 2: "Market standards" where adoption is essential for survival but offers no unique differentiation.
  • Category 3: Custom-trained, specific models that offer genuine competitive advantage but require high cost and investment.
• Six-to-Seven Pillar Methodology (Consulting Firm Approach):
  • Strategic alignment on growth vs. productivity goals.
  • Use-case identification to ensure ROI before scaling.
  • Business model development including "make vs. buy" ecosystem decisions.
  • Adaptation of delivery models to accommodate rapid model obsolescence (new models every 2–3 weeks).
  • Workforce planning for reskilling and upskilling.
  • Ecosystem building with tech partners.
• SME Accessibility: David Martineau highlighted that multi-pillar frameworks are too complex for small/medium enterprises; the next challenge is democratizing accessible AI tools for companies with ~100 employees.

Workforce Dynamics & Employment Trends

• Productivity Claims: A CAC (Car Company) CEO reported a 35% productivity increase in programming with GenAI, though the theoretical benchmark for replacement is 4x.
• The "Toyota Reset" Analogy: Gilles Babinet referenced the NUMMI plant transformation where firing and rehiring employees resulted in a 95–98% retention rate and world-class performance, suggesting human transformation is possible but requires a reset.
• Employability vs. Employment: Panelists agreed that fear of job loss stems from a lack of skills; the shift is from "losing jobs" to "losing employability" if staff cannot use AI tools effectively.
• Junior Talent Risk: Reducing entry-level coding tasks via AI threatens the training pipeline for future senior engineers; panelists emphasized preserving internships to maintain skill progression.
• Human-in-the-Loop: Victoria Veller and Gilles Babinet stressed that "machine + human" is the current reality, with human supervision essential to prevent hallucinations and ensure quality, particularly in coding and audio synthesis.

Public Policy & Ecosystem Development

• Environmental Application: Gilles Babinet identified GenAI as a critical tool for the 2050 decarbonization agenda, noting that $3 trillion in annual spending is unattainable without AI-driven efficiency in complex systems (energy grids, supply chains).
• Regulatory Sandboxes: Babinet called for government-created "playing fields" to allow experimentation with new use cases (e.g., autonomous mobility) without immediate regulatory blocking.
• Investment Philosophy: A paradox noted where society demands environmental and AI-driven safety improvements (e.g., fewer accidents) but resists paying for the transition.

Technical Implementation & Data Strategy

• Data Prerequisites: Florence Verzelin emphasized that GenAI efficacy relies entirely on high-quality data; companies must identify specific data needs for each use case before implementation.
• Data Cleaning Myth: Emily noted that data does not need to be 100% clean or fully secured to generate value; "fact-based" results with <100% accuracy are often sufficient for immediate business impact.
• Vendor Ecosystems: Emily advised against reliance on a single LLM, advocating for hybrid modes and open ecosystems to orchestrate multiple models and storage solutions.
• Prompt Engineering: Upskilling focuses on the ability to prompt effectively rather than coding GenAI internals.

Challenges & Risks

• Hallucinations:
  • General: Users must fact-check outputs from tools like ChatGPT to avoid blind acceptance of errors.
  • Audio: Eleven Labs encounters hallucinations where AI generates unintended screaming, imitations, or artifacts, requiring human curation.
  • Compliance: In regulated areas like CSRD reporting, algorithms must be tuned to prefer "no answer" over generating a fake number to avoid compliance nightmares.
• Skill Gaps: 81% of NASDAQ CEOs report a lack of digital skills within their organizations.
• Autonomy Limits: The industry is currently stuck at the 95% threshold; systems are not yet reliable enough for full transactional autonomy in critical fields.

Key Takeaways & Forward-Looking Statements

• Gilles Babinet: GenAI may become the primary force for decarbonization and ecosystem restoration in the near future.
• David Martineau: Acculturation at the C-suite (SCOM) level is the critical first step for transformation, especially for mid-sized firms.
• Florence Verzelin: Time for experimentation is now to avoid falling behind; GenAI is a "superpower" for innovation, productivity, and sustainability.
• Victoria Veller: Organizations should view GenAI as an output expander rather than just a cost-cutter and consider partnering with vendors to co-develop solutions.
• Emily: Trust is foundational; companies must define the positive purpose of AI and maintain critical thinking through trusted ecosystems.