AI & HRIS customer support: “Augmented RUN is not about replacing consultants: it's about giving them more time to analyse."

AI is everywhere in the HR world. What changes have you observed on the client support side?

Cédric Angin : "If we go back a few decades, support meant a library of technical manuals you hoped the vendor would update once a year. Then came call centres, the internet, ticketing tools, and the massive digitisation of information. Every era brought its own transformation.

What I observe today is that we are once again at an inflection point, but with one key difference: speed. The adoption cycle has shortened considerably. And with AI, we no longer have the luxury of waiting. Those who do not engage now will find themselves in difficulty within a few years."

Why is the traditional RUN model "reaching its limits" today?

"There are, in my view, three tensions that every AMS model, that is, every ticket management solution, must face today, and which no longer hold as well as they once did.

The first is volume versus quality. Tickets are multiplying, SLAs are tightening, and the pressure on deadlines often comes at the expense of analytical depth.

The second tension is between standardisation and customisation. Everyone says they want to stay in standard, and everyone ends up customising, for a country, a process, sometimes a single department. And that is entirely legitimate. HR configuration is your operational DNA. The problem is that the consultant has to learn this on the job, and is not available around the clock.

And the third is the shift from reactive to proactive. Today, RUN responds. You open a ticket, it gets handled. But nobody anticipates the upcoming release, the anomaly taking shape, the configuration drift that has been building."

How does agentic AI address these three tensions?

"The fundamental idea is not to add tools, but to add colleagues. What we call AI agents are autonomous entities that reason, read your systems in real time, remember context, and decide when to act. This is not a chatbot. It is not a script. It is something qualitatively different.

And the key, within the RUN framework, is that the Arago consultant stays in control. The agent handles the repetitive, the mechanical, the high-volume. The consultant, in turn, regains time for what they genuinely do best: analysis, judgement, and client relationships."

Concretely, what does that look like?

"Let us take tickets. Today, when a ticket comes in, the consultant manually searches for notes, opens internal documentation, re-reads the context. Sometimes they consult an expert to confirm a lead. All of this can take anywhere from two hours to a full day before a first human analysis is even produced.

With an agent, the ticket arrives pre-analysed. The agent has cross-referenced your description with the documentation, examined your instance, proposed leads with a confidence level, and prepared a response ready for validation. In a matter of seconds. The consultant arrives at a decision, not a search.

That is a fundamental shift: the consultant is no longer searching. They are deciding."

And what about SAP releases, which are often a pain point?

"That is an excellent example. Twice a year, SAP publishes a generic release note, applicable to all clients. It falls to us to identify what is relevant to each client's specific configuration. Today, that work takes several weeks, sometimes under very tight deadlines ahead of the production upgrade.

An agent connected to your instance can cross-reference the release with the actual configuration, isolate the changes that concern you from among hundreds of items in the note, assign a criticality level to each, and produce a personalised impact report in a matter of minutes. It is no longer a generic analysis that the client has to make sense of themselves, it is a targeted, ready-to-share analysis."

Two other important topics in the RUN context are workbooks and steering committees. Where do things stand with AI?

"Indeed, these are two perfect examples of tasks with a high human workload but low human added value.

Workbooks first. For audit reasons in particular, clients request them from us. But they are not always up to date, sometimes lost, often out of sync with your actual instance. Recreating them takes weeks of work. A recent example: days spent on a recruitment workbook, to what end? So that an auditor could stamp it. In terms of creative value, we are really scraping the bottom.

An agent can automatically detect configuration changes, regenerate the workbook up to date, and identify conflicts. In a matter of tens of seconds. The consultant no longer does that. And they can finally use that time to actually analyse the workbook. That is where the value is created.

For steering committees, the same logic applies. Today, preparing a monthly COPIL means extracting from Redmine, formatting in Excel, building a PowerPoint, often a full day's work for the consultant. And the trends arrive too late to genuinely influence decisions.

An agent can generate the dashboard, flag anomalies, and identify three priority recommendations. The consultant arrives at the steering committee to analyse, not to format slides."

One question keeps coming up: will AI replace consultants?

"No. And I want to be very clear on this, because it is a concern that is understandably felt.

There are two types of activity in support. Those where it is the outcome that matters: generating a document, extracting data, formatting a report. And those where the process itself has value: analysis, decision-making, client relationships, judgement. The former are ideal for an agent. The latter remain deeply human.

What we are aiming for is for the consultant to free up time from the first category in order to concentrate on the second. And I would add something that is said less often: a consultant who finds meaning in their work produces superior quality. Nobody genuinely enjoys spending a full day formatting a PowerPoint. Free up that time, and you free up energy and engagement.

So yes, the same contract, the same team. But a quality of service that evolves."

Where does Arago currently stand with this approach?

"Together with the consultant teams, we have reviewed all the tasks that could be handed over to agents: everything punctual, recurring, mechanical. We have identified around fifteen concrete use cases, some already achievable today, others still to be developed.

What I tell the teams is that we will not get perfect results immediately. We need to educate the agents, refine them, and draw lessons from the first deployments. But the early feedback makes me very optimistic.

And what gives me even more confidence is that the technology is moving fast. We are not talking about a five-year horizon. We are talking about months. The Augmented RUN is arriving now."