Leadership
How AI changed my job: it gave me back time for people
People often ask how, as a CEO, I find the time to talk to every candidate, every client, every member of my team. The truth is I don't find that time. AI gave it back to me. Here is what I automate, what I refuse to automate, and why the real change is a human one.
“It's rare to talk to the CEO directly. How do you find the time?” I hear it at the end of roughly every other first call. And every time, the same answer comes to mind. I don't find the time. I get it back. And what gave it back to me is AI.
That may sound surprising coming from someone who sells AI for a living. We tend to picture AI as a way to produce more, faster, to replace people. In my day to day, it did the opposite. It returned me to my real job: people.
Before, I spent my days doing plumbing
Two years ago, a typical day looked like this. Sorting dozens of emails. Chasing a client who hadn't replied. Reformatting a CV for the third time. Copying call notes into the CRM. Digging for the right contact at a prospect. Answering the same message for the fortieth time.
All that plumbing was necessary. But it asked for none of what actually makes up my job. No judgment, no listening, no relationship. And worse, it came before everything else. By the time I finally got to a call with a candidate or a check-in with a teammate, it was often late in the day, my head elsewhere, half fried.
What I automated, with no guilt
I spent a year handing these tasks over to systems. Not gadgets. Serious, tested workflows with guardrails. Concretely, here is what runs today without me:
- Inbound triage. Spontaneous applications, client needs landing by email, first-level qualification. A machine reads, sorts, prepares a clean record.
- Follow-ups. Multichannel tracking, reminders, that “I'll get back to you” we all forget on a busy day.
- Production paperwork. Candidate records, skills dossiers, meeting notes filed in the right place in the CRM.
- Cold outreach. Finding the right person to talk to, preparing a personalized first contact.
None of these tasks builds a relationship. None of them needs my human eye. Handing them to machines takes nothing away from me. It gives me hours back.
What I refuse to automate
The reverse is just as clear in my mind. There is a line AI does not cross with us.
The first call with a candidate, that's me. The coffee with a hesitant client, that's me. The 1-1 with a teammate going through a hard time, that's me. Trust is not generated by a prompt. It is built by looking at someone, truly listening, and keeping your word.
The day a candidate senses they're talking to a bot in their interview, we've lost. The day a client feels like a ticket in a queue, we've lost too. Those moments, I protect them. They are what make Abbeal.
“But don't you lose quality?”
That's the question that comes right after. The fear is that by automating you let errors slip through, send a bad message, rub someone the wrong way. It's a fair worry.
The answer fits in one word: verification. Every system we put in place has a human at the end of the chain for the sensitive moments. The machine prepares, proposes, roughs it out. It never blindly sends a message that commits the brand or a relationship. A draft, yes. The decision to hit send when it matters, that's us.
It's exactly what we recommend to our clients. AI in production is not 100% autonomy. It's 80% engineering and guardrails, and 20% model. Human judgment stays in the loop wherever the risk is real.
The real change is not technological. It's human.
The result: I spend far more time one on one than I used to. With candidates, with clients, with the team. That's where the decisions that matter get made. That's where you feel a profile will click, a project is going sideways, a talent needs a sign.
No AI would have handed me that by scanning a database. It took a conversation. It took time. And that time, before, I simply didn't have.
So when someone tells me, a bit surprised, “it's cool to talk to the CEO directly,” I smile. It's not a luxury I allow myself. It's the core of my work. I just cleared away the rest.
We industrialize AI for our clients. We did it on ourselves first.
At Abbeal, our job is to move AI from PoC to production, for demanding clients. Banking, energy, retail, fintech. We know what a reliable, monitored, compliant system that holds under load actually means.
The best proof we believe in it is that we plugged it into our own daily work before selling it. Not to replace our people. To give them time back, and reinvest it where only a human makes the difference.
If I had to sum up what AI changed in my job, it would be this. It didn't make me more productive in the usual sense. It made me more available. And in a job that is, at its core, about relationships, that's the best gift it could give me.
// Read next
Business
Output-based vs Time & Material: why we killed T&M at Abbeal.
78% of Abbeal portfolio runs on Output-based pricing in 2026. Gross margin +18 pts, NPS +24, engagement length ×1.7. How we operate and 3 success conditions.
11 min
Talent
How to build a senior engineering team across Asia, Europe and North America
The playbook for assembling a senior engineering team that operates across three continents — Asia, Europe and North America. The Abbeal three-hub model: Paris · Montréal · Tokyo.
7 min
IA
How I automated a tech consulting CEO's day with Claude (and what you can learn from it).
30 workflows orchestrated on Notion + BoondManager + Google Workspace + LinkedIn + Apollo + Calendly + Tactiq, no new SaaS. 4 pillars: multichannel anti-duplicate sales, 48h recruitment, inbound SEO/LinkedIn/AI citations, founder productivity. Zero lost leads in 6 months, 15 min/day vs 3-4h before.
7 min
IA
7 patterns for AI agents in production (no demo theater).
Real-world patterns from RAG, agents and MLOps deployments. Senior teams shipping AI from POC to prod across Paris, Montréal, Tokyo.
9 min
GreenOps
GreenOps: seven levers that cut 30% of your cloud bill.
Without sacrificing performance. Concrete cases: -30% on the bill, same SLOs.
6 min
