AI didn't replace our support team. It gave us time back.
Most of what fixed our backlog was experience and colleagues helping each other. AI arrived late, and it made both go further.
- Type
- Thought
- Topic
- AI, customer support
- Written
- Aug 2026
- Origin
- My time at Ivalua, 2024 – 2026
Starting small
I joined Ivalua, a procurement software company, in December 2024 as a support engineer. The first month was onboarding. Then I got my first client, and it was a calm one. They did not ask for much.
So I used the extra time. Little by little, I took on part of the team's backlog, especially the complex issues others had put aside. And whenever a colleague had an upgrade project, moving a client from one version of the software to the next, I chose to help.
The upgrade that went wrong
Eight months in, I was helping a colleague with an upgrade when he went on holiday. The client was in a tense moment: issues were blocking the move to production, and nobody was sure yet how to get there.
I stepped in as backup. In two weeks, we fixed most of the issues. The client went live, and I kept the account afterwards to improve their satisfaction. There was no escalation after that.
Issues blocking the client's move to production when I took over, and the minor ones still open when I left Ivalua.
That made three clients, the maximum for one person in our team. Later, a fourth client needed a dedicated technical account manager. I volunteered, since I was already spending more than half my time helping on other projects. Again, no escalation, and a good relationship. By the end I had four clients: two in France and two in Germany.
Then AI arrived
AI came into our work only four months before I left. We used it to analyse issues, and we built a few tools that ran a first analysis before anyone looked at a ticket.
It made our analysis better and faster. And that gave us the one thing a support team never has enough of: time to talk with clients.
My team at Ivalua, by the time I left in August 2026
The drop in major incidents came from both sides. The tools helped us sort and prioritise. And after so many upgrades, I could often see when a new incident looked like an old one, so we could fix the cause and not just the ticket.
AI made our experience go further. It did not replace it.
The point of this piece
What I take from it
Most talk about AI in support is about replacing people. That is not what I saw. Most of our results came before AI, from experience and from colleagues helping each other.
AI was most useful on the slow, repetitive part of the job. It freed experienced people to spend their time where it counts: with the client.
- Most of our results came from experience and colleagues helping each other.
- AI made the analysis faster, and gave us time back for clients.
- Use AI on the slow, repetitive part of the job, not to replace the people who know the clients.
