How does a global human resources and staffing company think about AI strategy — and prioritizing the myriad potential use cases?
“We’ve decided to always listen to our employees in our branches, who are performing daily tasks, and we ask them to tell us their priorities,” says Daniela Petrovic, Head of AI Transformation for Adecco in France.
Adecco is part of the Adecco Group, a $25 billion publicly-held company with 35,000 employees and operations in 62 countries.
InnoLead spoke with Petrovic about the company’s early AI deployments; training initiatives; and the five principles that guide the use of AI at Adecco.
Petrovic operates out of the company’s French division, which has 900 branches and 8,000 employees.
• • •
Can you tell me what your job entails?
Here in France, my job is split into three main missions. The first mission of my job is governance. The second part is to build an AI solution for teams, for candidates, and for the clients. And the third one is to work on the training, which is a really important mission.
The strategy we have built for the corporate teams establishes three pillars. The first is everyday AI, where we look into solutions that are used in everyday tasks, like summaries or generating images that can be used by everyone in the company.
The second one is optimizing AI. Here, we assess the maturity of the use cases that can be scaled and we evaluate the jobs and the different sections in each country.
The third pillar is around disrupting AI, and here we have a lab…where we are building AI agents. We are willing to be a leader, to be in front of this topic and to try and get adoption from our clients and our candidates and, of course, our colleagues, knowing that we need to get in front of AI Agent adoption.
Can you give me an example of one or two use cases that you are benefiting from?
One of them is called Jobposting, which was launched in 2024 and…automatically generates a complete job description based on criteria filled out on a form, while ensuring full compliance with labor laws and non-discrimination standards.
So far, the tool has produced over 35,000 job [postings], helping recruiters save an estimated 70 percent of their time in the drafting process. Prior to the introduction of these tools, recruiters were writing down the job description, [and] they needed to review it to make it okay and ready for the job boards. We have specific rules to follow before posting these job descriptions on job boards. They had to review it to see if the information and style of the description properly represented the client. It took a long time to do this, and time is really important.
Another use case…is GPT Assistant, which is a chatbot that approximately 1,000 people use every day to find out internal company information that helps them make decisions faster… Our colleagues…can gather and share Adecco’s information to help them with brainstorming sessions and problem-solving discussions that lead to better outcomes.
…To be frank, nobody has the best way to calculate the ROI.
In AI use cases such as those you’ve described — cutting the time to do the work — don’t generate numbers that show money saved or earned from the technology, how do you measure ROI?
That’s a difficult question, and I think the company is still wondering around that and still trying to have the best calculation. I am speaking to my peers regularly on this question and to be frank, nobody has the best way to calculate the ROI. When you go and you take a tool like Microsoft Copilot, you have two minutes that are spent on one task, instead of four, how do you calculate that? It is really difficult.
…Our aim in the end is to free [our colleagues] on the tasks that are more difficult or time-consuming, where they are not really happy doing those tasks. We really try to help them and to give them more time with our clients and with our candidates. That’s the goal of all of this.
You’ve applied AI to the task of creating résumés, which is another use case. Can you tell me about that?
In 2023, we launched CV Maker, which has already been used by nearly 500,000 people, and has generated close to 200,000 résumés.
The CV Maker tool has been designed for candidates who don’t know how to create résumés. The tool is part of our website, is accessible globally, and offers several languages. You just have to talk to the tool and just tell it your experience, explain what you have done, and then AI will create a résumé.
How do you prioritize your AI projects? Which projects do you select first and why?
It’s an important question, and we have lots of discussions around this topic. It is not simple, but we’ve decided to always listen to our employees in our branches who are performing daily tasks and we ask them to tell us their priorities. The priority could be about time, [or] it could be about security. …In the end, we have a last step, which is the approval of the topics and projects with our general directors. There are several steps, but it starts from our branches.
Can you tell me the companies you’ve partnered with that support your AI efforts?
We have a major and strategic partnerships with Salesforce and Microsoft, [and] we are using mainly Salesforce on our use cases connected to the daily work. It’s the daily operational tool for our recruiters. …All the tools that we are building internally are built on Microsoft environments. We use Azure for our cloud computing platform. We also use OpenAI’s tools.
How are you training your staff on generative AI tools?
More than 800 hours of training has been given to the 35,000 employees across The Adecco Group. In France, we trained our French colleagues starting in 2024 with…short training lessons on our tools, with the idea that we can go to the adoption of these tools once they’ve learned the basics, such as what is AI? How can you use it? And what are the risks? Then they can go to step two, to train for the adoption of AI. There we focus more on showing our colleagues how to understand and use the tools.
What’s the most difficult part of training your global colleagues?
The most difficult part I would say is to understand where and when, to whom and which type of population can go further in the training courses. We are now at a step where almost everyone in the company knows about the basics of AI. …Now it’s more about the adoption, and here there are a ways to go.
There are many concerns about using AI to create résumés and correspondence such as emails between candidates and employers. Concerns range from the possibility that AI tools contain algorithmic bias that reinforces discrimination, to privacy and data security risks, as well as creating correspondence that lacks the human touch. Are you seeing some of these issues occurring — and how do you handle them?
It’s a daily topic in our AI governance. …We’ve established a Responsible AI Committee that has five “Responsible AI Principles,” which reflect the commitment of the Adecco Group to responsible AI. The five principles The Adecco Group adheres to are the ethical, human-centric, transparent, safe, and the lawful use of AI.
… When we begin a use case, we do an assessment at the beginning with these five pillars. …We even do this when a project is live.
I’m sure you’ve seen the introduction of new features on OpenAI’s videos generation app Sora 2, which is causing some concerns. Employers who view candidate’s social media footprint before choosing a candidate for employment might view a Sora 2 generated video that carries deepfake images and misinformation about a candidate. What are your thoughts about this development?
As I said before, we have principles, and I think these kinds of tools and AI applications in general must be ethical, human-centered, and safe. I think employers have to be ready to adhere to all the principles to know if these videos are relevant or not.
I think you need to have agility and be open to learning almost every day….
What lessons have you learned during the nearly two years that you’ve been the Head of AI Transformation at Adecco?
Learn how to learn. It’s important that I say that to my teams, because in the AI environment, you never know everything, things are moving, teams are also evolving in their knowledge… not everyone is up-to-date at the same time. I think you need to have agility and be open to learning almost every day….
What are your future plans for AI?
My future plans are to follow the three pillars: use AI everyday, optimize AI, and disrupt the staffing industry with new AI tools. Disrupt AI is thinking about process and pure innovation. You think about the current process and ask yourself, do I need to change that process? If the answer is yes, then we can think about new AI solutions to modify and improve it.
(Featured image by Poppinsdepoppins / Wikimedia Commons / Creative Commons Attribution-Sharealike 4.0 International)
You must be logged in to post a comment.