As a leader in enterprise cloud computing and customer relationship management solutions, Salesforce France supports thousands of French companies in their digital transformation.
By joining Positive AI, the company reaffirms its commitment to ethical, responsible, and sustainable artificial intelligence.
Interview with Julie Ravillon, Sustainability Director Salesforce France
Why did you choose to join the Positive AI Do Tank, and what do you hope to bring or develop there?
By joining Positive AI, Salesforce aims to play a leading role in the implementation of ethical, responsible artificial intelligence that complies with the European AI Act. This commitment is a continuation of our efforts to promote practical AI that is useful to businesses and employees and aligned with real-world corporate use cases.
We want to actively contribute to this collective momentum by participating in a practice community dedicated to sustainable AI, while also strengthening the training of our teams on the challenges of responsible AI.
Salesforce employs several strategies to develop efficient AI while keeping environmental impact in check, based on a three-pillar framework: Smart Demand, Efficiency, and Clean Supply.
What levers does Salesforce use to develop high-performing AI while managing its environmental impact?
Salesforce employs several strategies to develop efficient AI while keeping environmental impact in check, based on a three-pillar framework: Smart Demand, Efficiency, and Clean Supply.
A key example of this approach is our Agentforce solution. Unlike DIY AI solutions, which can result in significant energy consumption due to extensive training of massive generic models, Agentforce is “sustainable by design”:
-
Pre-trained models: Agentforce uses pre-trained models, avoiding the costly and energy-intensive training of new ones.
-
Agentic architecture: Its architecture allows for the use of smaller, use-case-specific models.
-
Data optimization: Personalized data, metadata, and the unified Salesforce platform help optimize performance while reducing energy consumption.
By integrating sustainability at the heart of its AI strategy, Salesforce helps clients innovate responsibly and sustainably, minimizing AI’s environmental footprint.
Additionally, collaboration among stakeholders in this space is essential. That’s why we engage in various initiatives, such as the AI Energy Score in partnership with Hugging Face, Cohere, and Carnegie Mellon University which provides a transparent framework for measuring and comparing the energy consumption of AI inference.
Salesforce is the first major company to disclose the energy consumption of its proprietary models.
How do you support your clients in adopting more responsible and sustainable AI?
Salesforce is deeply committed to ethical and responsible AI, guided by five clear principles: accuracy, safety, honesty, empowerment, and sustainability.
We focus on developing AI tools—especially AI agents—that offer full control, enabling teams to collaborate successfully and responsibly with AI. The company implements rigorous safeguards, extensive testing (including red teaming and trust tests), and full transparency to ensure AI systems can be trusted.
Regarding sustainability, Salesforce has launched Agentforce, a built-in solution of autonomous AI agents on its platform. Agentforce is “sustainable by design,” using pre-trained models to avoid energy-intensive training. Its architecture supports the use of smaller, tailored models for each use case, optimizing performance while lowering energy usage.
Through Agentforce, we help clients innovate in a responsible and sustainable way by reducing the environmental impact of AI.
