Playbook: 3 steps to integrate AI into your CMS

As part of our reflections on the best way for a structure to get to grips with the subject of AI, we've gathered the words of experts on the subject, in order to offer you a relevant approach. We have come up with 3 essential steps:

• Digital maturity audit,
• Proof of Concept,
• Scale-up.

Digital maturity audit

This first step may not be intuitive for everyone, but it's essential. To integrate AI, as with any digital transformation, you first need to make sure that all the indicators are green.

Digital transformation expert Frédéric Cavazza insists on the risk of "digital debt" for structures seeking to integrate new solutions without taking this time for introspection. By the term "digital debt", Frédéric Cavazza refers to theaccumulation of digital practices without regularly questioning them.practiceswithout regularly questioning them, and gradually losing sight of the skills needed to operate or use them.

So, before you decide to integrate AI into your business, whatever the target area, you need to :

1. Clearly define the objectives to be met by AI, and the role it can play in your current strategy,
2. Ensure that the tools already in place are mastered by your teams,
3. Check that AI does not overlap with other tools.

Proof of Concept

A PoC, for Proof of Concept, is a way of validating an idea without incurring excessive costs. In the context of the introduction of new tools, a PoC can, for example, take the form of a small-scale implementation (a trade, an office, a business unit, etc.).

The success of a PoC depends largely on the people who make up the pilot group. Involving end-users and process owners from the outset is essential. Their feedback is essential to assess the tool's suitability for real needs, and to encourage its future adoption.

But this is not the only criterion to take into account. The experience of PoC stakeholders is also key, as is the presence of "champions".

Selecting the right team for your PoC

On the first point, it would seem that productivity gains through AI are proportional to employee experience. In fact, Gartner claims that they are greater when the tools are taken in hand by complex professions.

Put another way: in the hands of lower value-added professions, AI tools quickly improve productivity, because they're easier to get to grips with. But this ease of use often goes hand in hand with lower potential benefits.

The 2nd point is that you need to include a few people in this group who see the potential of the solution and are capable of changing their practices to integrate these new tools.

Tool education

Once the scope of the PoC has been defined, in addition to a period of skill-building, the implementation of your tools sometimes requires a period of acculturation. Indeed, when it comes to implementing core tools - a CMS, for example - the change that takes place can be as much a matter of digital transformation as a cultural evolution in its own right. It's a step that promotes both team buy-in and efficiency.

If you're not convinced, here's a figure that should change your view of this training and acculturation stage: in a Gartner study, 72% of IT managers claim that employees have difficulty integrating artificial intelligence into their day-to-day tasks.

All we've done up until now is digitize what already exists, but digital transformation means transforming an organization from within. And that includes artificial intelligence.

Scaling up

Going to scale... Or not: what the experts we consulted, particularly the Gartner experts, clearly show is that a successful PoC does not guarantee the viability of going to scale.

Organizations must not lose sight of the fact that AI still has many limitations:

• Reliability - Despite significant improvements since ChatGPT's inception, content is still not 100% reliable.
• Security - The conditions under which data is processed by tools based on artificial intelligence - whether generative or not - are sometimes controversial. This is why OpenAI Azure is favored by large companies over the classic version of the solution.
• Cost - according to Gartner, software vendors charge up to 50% more when they integrate AI functionality.

However, on this last point, scaling up can sometimes entail costs that are not proportional. This is not only because the solution itself becomes extremely expensive, but also because the costs of change management at BU and company level, for example, are not the same. With the generalization of a solution, some organizations lose a large part of the ROI evaluated during the PoC.