AI and productivity: where do French companies really stand?
The landscape has changed in three years. The opening of ChatGPT to the public in November 2022 has, according to the Digital Society Lab, reactivated debates on the potentially massive effects of AI on employment and productivity. In terms of personal uses, the Digital Barometer 2024 indicates that 33% of French people have already used an AI tool in 2024, compared to 20% in 2023.
On the business side, INSEE documents artificial intelligence in companies in its most recent publications, while the Directorate General for Enterprise reminds us that France has 590 start-ups dedicated to AI, including 16 unicorns: the ecosystem is mature, the technical building blocks available, and the subject is no longer forward-looking.
On the ground, the gains can be measured. According to France Num, using artificial intelligence saves 2 hours per week for craftsmen who have equipped themselves with it. At the level of an SME, this represents a significant margin lever.
The 7 levers for optimizing AI in business
1. Automation of repetitive tasks
The first source of productivity: freeing teams from low value-added tasks. Data entry, report generation, email sorting, first level of customer support, writing product sheets… Generative AI tools (conversational assistants, co-pilots integrated into office suites) and AI-augmented RPA platforms support these flows.
The goal is not to replace but to reallocate the time saved to higher-value activities. This logic is in line with our work on operational efficiency and the standardization of processes via the core model.
2. Predictive analysis and activity management
Predictive AI leverages historical data to anticipate. France Num details how predictive AI can use your data to better manage activity : sales forecasting, anticipation of stock-outs, early detection of weak customer signals.
In concrete terms, the solutions range from native ERP predictive modules (SAP, Oracle, Sage) to specialized platforms (Dataiku, Alteryx) and augmented BI tools (Power BI Copilot, Tableau AI). The subject directly links performance indicators and management control.
3. Supply Chain Optimization
Refined demand forecasting, route optimization, dynamic inventory allocation between warehouses, detection of supplier anomalies: the supply chain concentrates use cases with high ROI. Machine learning algorithms reduce overstock, secure lead times, and improve service levels.
4. Personalization of the customer experience
Product recommendations, behavioral segmentation, appetite scoring, generation of personalized emails at scale: AI makes it possible to industrialize the personalization that marketing teams have until now practiced by hand. On the tool side: Salesforce Einstein, HubSpot AI, Adobe Sensei for large accounts; lighter platforms for SMBs.
5. Human Resources Transformation
Sorting applications, semantic CV/job matching, generation of job descriptions, adaptive training paths, analysis of the social climate via language processing: the HR function is becoming one of the most active fields of applied AI. This dynamic is in line with the broader subject of digital HR and the GPEC/GEPP diagnosis.
6. Industrial Predictive Maintenance
IoT sensors, learning models and monitoring platforms make it possible to anticipate failures before they occur. The result: fewer unplanned downtimes, longer equipment life, and optimization of spare parts. The subject is central for industrial departments and completes our analysis of industrial audits.
7. Energy optimization and sobriety
Real-time consumption management, adjustment of operating ranges, forecasting of renewable production: AI is directly involved in the objectives of decarbonization and energy cost control. Smart tertiary buildings and energy-intensive industrial sites are the first to be affected.
Risks to anticipate to secure deployment
Data security and compliance
An AI project is based on large volumes of data. The ANSSI has published security recommendations for a generative AI system : access governance, partitioning, control of outgoing flows to third-party models. GDPR compliance remains a prerequisite, especially for HR and customer data.
Algorithmic biases and data quality
A model reproduces the biases of its training data. In HR, credit, evaluation, these biases can generate discriminatory decisions. The solution: quality of datasets, fairness tests, human supervision of sensitive decisions.
Technological dependence and sovereignty
Relying on a limited number of foreign suppliers exposes you to continuity, price and compliance risks. The State is pushing a logic of sovereignty with a call for expressions of interest on generative AI solutions for the public sector, which is gradually irrigating the supply available to private companies.
Impact on employment and social dialogue
The subject has become a central issue in social dialogue. The dossier published in February 2025 by the Digital Society Lab explicitly places the impacts of AI on work and employment as a “new challenge for social dialogue”. The work of the LaborIA laboratory, created by the Ministry of Labour and Inria, argues for an “enabling AI”, i.e. one that strengthens skills rather than marginalising them.
For management, this means anticipating change management, overcoming resistance to change and structuring HR risk management.
Implementation costs and ROI measurement
The question of return on investment is central. data.gouv.fr provides an Enterprise AI ROI Barometer covering 200 deployments 2024-2025, which documents the cost, implementation time and actual ROI observed in the field. This type of standard, based on operational audits and not on declarations, finally makes it possible to get out of the “hype” and to arbitrate with full knowledge of the facts.
Ethics and acceptability
Employee monitoring, processing of personal data, transparency of automated decisions: the ethical dimension conditions the acceptability of the project internally. This point is in line with our analyses on the integration of ethics in business.
How to Succeed in an AI Project: Wayden Method
Beyond the tools, success lies in the method. Four pillars structure the projects we manage.
- Identify priority use cases by business value and technical feasibility, rather than by fad. A targeted MVP is better than an ambitious plan that is not delivered.
- Securing data upstream: without quality and governance, no model will keep its promises.
- Onboard teams through training, social dialogue and clear roles. LaborIA’s recommendations on “enabling AI” provide a useful framework.
- Continuously measure : business indicators, ROI, user satisfaction. Project-based management and the agile method offer the right frameworks.
For organizations that do not yet have the skills in-house, the use of a specialized interim manager allows the process to be quickly structured, from the initial audit to the industrialized deployment, without committing to a permanent position before consolidating the model.
Are you launching an AI project in your organization?
Wayden mobilizes senior interim managers experienced in data and AI transformations, in general management, CIO, finance, HR or industry. An exchange to frame your needs and identify the right profile.
FAQs: AI and Business Productivity
What percentage of French SMEs are already using generative AI?
According to Bpifrance Le Lab, 31% of VSEs and SMEs use generative AI, a proportion that has doubled in one year. However, French companies need support to move from individual use to structured deployment.
What time savings can we expect from AI on a daily basis?
The orders of magnitude vary according to the profession. France Num observes, for example, that AI saves 2 hours per week for craftsmen who have equipped themselves with it. For tertiary functions, the gains are concentrated on writing, document analysis and meeting preparation.
Is there public support for AI projects?
Yes. The “Dare to use AI” plan aims to spread AI in all companies, and Bpifrance finances and supports SMEs in this transition to AI. A catalogue of sovereign solution providers is also available.
How to measure the real ROI of an AI project?
The ROI barometer for AI in business published on data.gouv.fr, which covers 200 deployments from 2024 to 2025, documents costs, implementation times and observed returns. It is a useful reference for framing a profitability analysis, in addition to specific business indicators.
Will AI destroy jobs?
The debate is posed. The Digital Society Lab reminds us that the central issue is now that of social dialogue. The logic of “enabling AI” defended by LaborIA consists of using AI to strengthen skills rather than to purely automate positions.
Articles récents
Smart Factory: the future of industry is digital and smart
Feedback from an agile coach in the context of a digital transformation
Artificial intelligence in industry: what are the challenges?
The CFO: a key role in the company’s digital transformation
What is omnichannel business transformation?
ERP core model: how to successfully deploy it?