Artificial intelligence (AI) has become part of our everyday lives. It helps us prepare our shopping lists, find answers to simple and complex questions, and even write messages in a matter of seconds. For many of us, using AI has become second nature.
And large companies have understood this: this technology is a powerful lever for productivity and innovation. But between spectacular announcements and operational reality, one question arises: how can we transform the excitement surrounding AI into robust, relevant, and sustainable uses?
This is precisely the challenge facing large corporations. To shed light on this issue, we spoke with Benjamin L'hyver, Chief Data Science Officer, and Guillaume Jaeger, Director of Consulting & Transformation at Eulidia, a consulting and technical expertise firm specializing in data and AI since 2008. They share their observations from the field on the real needs of organizations and best practices to adopt.
A booming market, but in need of streamlining
AI is everywhere today, with businesses multiplying their experiments, which is an excellent thing. But real value comes when we succeed in transforming these initiatives into robust and sustainable solutions.
However, behind this enthusiasm lie new challenges: how can we guarantee the relevance of projects, ensure their compliance, and prevent each department from working in a disjointed manner? For large groups, the challenge now is to structure this energy, align it with business priorities, and incorporate it into a comprehensive support approach.
Today, management teams are looking less for “creativity” at any cost and more for robustness: reliable, fast, compliant solutions that can be integrated into existing processes. The year 2025 should therefore mark a turning point, with three major priorities:
Acculturate teams to make projects more sustainable and maximize their adoption.
Harmonizing practices to better control compliance.
Streamlining usage to reduce costs and unify methods.
Some examples of AI use cases in large corporations
While media coverage of generative AI has broadened the scope of possibilities, the use cases currently being deployed in large corporations are primarily pragmatic and focused on operational value. These include:
Search engines capable of responding in natural language.
Business assistants for technical and back-office functions.
Automation of responses to requests for proposals (RFPs).
Narrative reporting.
On the technology side, certain standards are becoming established: these include Snowflake, Dataiku, Databricks, but also classic LLMs (from providers such as OpenAI, Mistral AI, Google, etc.), and no-code solutions such as N8N.
A new playing field, but the fundamentals remain the same
For Guillaume Jaeger, an AI system, like any data project, is based on three complementary layers. The first is the business layer, with end users expressing their needs on one side and management defining priorities and the roadmap on the other. The second is the technical layer, made up of data scientists, data engineers, data architects, and analysts, who develop solutions and build the technological foundations. Between the two lies an essential layer: that of Product Owners and Product Managers, who act as conductors, linking business and technology, translating needs into concrete use cases, and ensuring that each project generates real business value.
Without clear coordination between business, PO/PM, and technical teams, projects risk remaining fragmented, experimental, or disconnected from business value. But when the system works as a whole, AI becomes a robust, relevant product that is adopted by users.
And contrary to some preconceived notions, AI has not changed the skills required. “What makes a good Product Owner hasn't changed with AI,” explains Guillaume Jaeger. What makes the difference remains:
Clear communication, both verbal and written.
The ability to simplify and explain technical concepts.
An understanding of business issues.
Anticipating the long-term use of the application.
Effective project management.
However, AI highlights a new dimension: the importance of UX/UI. AI tools are aimed at audiences who are not necessarily accustomed to using technical solutions, which is their strength. However, the interface becomes a critical criterion for adoption.
Choosing the right use case for a solid AI project
Beyond technology, the success of an AI project depends above all on a solid methodology. At Eulidia, this involves first defining and then prioritizing use cases using an evaluation grid of 54 criteria, divided into five pillars and then synthesized into two axes: interest and feasibility, in order to focus efforts on the most relevant projects.
Adoption is another pillar that relies on a sponsor on the end-user side, the product owner, who is involved from the outset. They play a key role in gathering feedback and continuously developing the solution.
In terms of ROI, Eulidia advocates a “humble and pragmatic” approach, explains Guillaume Jaeger. Measurable gains such as the number of users or time saved are tracked, but it remains difficult to quantify everything, especially when the tool is intended for cross-functional use. Nevertheless, here are some successes that have already been observed:
Several weeks saved on filling out AO forms.
Autonomy for management and business analysts
Access to complex knowledge bases for end users
Increased speed and efficiency of complex industrial interventions
Reduction in unnecessary requests for assistance by promoting “self-care”
Acceleration of onboarding for new arrivals via dedicated assistants
Reduction in the impact of key personnel leaving
Finally, there are some points to watch out for. While maintenance costs are planned for from the design stage, it is the usage costs, linked to the volume of requests, that remain the most difficult to control. Added to this are the issues of cybersecurity and confidentiality, which are of paramount importance and must be integrated from the outset to ensure the security of the project.
Towards a new balance between IT departments and business lines
AI is not only disrupting processes, it is also reshaping project governance. As Benjamin L'hyver points out: “With GenAI, it is no longer just IT that drives projects, but also the business lines that come looking for expertise.”
For large corporations, this marks a turning point. The days when data initiatives were confined to the IT department are over: now, business departments are directly involved in defining and driving projects. This evolution requires new ways of collaborating, where each stakeholder, IT, business, and cross-functional departments plays its full role.
To support this movement, LittleBig Connection gives large corporations quick access to the expertise they need by creating a direct and transparent link with freelancers and specialized consulting firms such as Eulidia. This provides an additional lever for transforming AI into concrete, value-creating projects.
