Publicis Says AI Will Cut Tasks, Not Jobs. What It Means for Smaller Agencies

Publicis Groupe held its Q3 2026 earnings call on October 13, and the headline wasn't the 5% revenue growth to roughly 3.5 billion euros. It was CEO Arthur Sadoun putting a hard number on what AI is supposed to do inside the agency: cut manual and repetitive tasks by an estimated 25%, while the company's workforce keeps growing toward 114,000 people. Coverage from Storyboard18 and Digiday both flagged the same thing: this is one of the most specific public statements a major holding company has made about where AI is actually supposed to land inside agency operations, not just in pitch decks.

What Sadoun Actually Said

The claim has two parts, and they are doing different jobs. First, Publicis says more than 90% of its roughly 114,000 employees have already been "equipped and trained to use AI productivity tools," a push the company has expanded through a Microsoft partnership. Second, Sadoun set a goal of reducing manual, repetitive tasks by about 25% through AI agents, explicitly to free people for higher-value work rather than to shrink the org chart.

To back up the job-security framing, Sadoun pointed to headcount history: Publicis has grown from around 65,000 employees in 2017 to roughly 115,000 today. The implication is that AI-driven efficiency and workforce growth have coexisted before, so they can again. What the company has not done is commit to that math going forward. There is no public headcount target tied to the 25% task-reduction goal, and Sadoun has not said whether the initiative changes hiring pace, redundancies, or staffing mix by function or region.

AI-driven productivity does not automatically mean fewer jobs in a growing business — but a 25% target without a stated headcount commitment is a number clients and employees alike should read carefully.

Why a Holding-Company Number Matters at Smaller Agencies

Most agencies are not managing 114,000 people or a global AI rollout. But Publicis setting a public, specific efficiency target changes the conversation every agency is having with clients and staff, for three reasons.

• Clients now have a benchmark. When the industry's largest holding company says AI should remove a quarter of manual work, a client who reads that headline will reasonably ask why their smaller agency partner isn't quoting a similar number. Be ready with your own honest answer, even if it's "we're not there yet and here's why."
• "Tasks, not jobs" is a framing clients will start repeating. It's a useful, defensible distinction — AI removing repetitive work like reporting, first-draft copy, or campaign setup is different from AI replacing strategic judgment. Agencies that can explain where that line sits in their own workflow will sound more credible than ones that simply promise "we use AI."
• Vague AI claims are getting harder to sell. Sadoun's number is specific and tied to a stated mechanism, deploying agents against defined repetitive tasks. Clients evaluating agency partners, including smaller ones, are increasingly going to expect that same specificity rather than a general AI pitch.

What To Do With This

You don't need a 25% target to take something useful from this. Walk through your own account workflows this month and identify which tasks are genuinely repetitive and low-judgment, things like status reporting, first-pass social copy, meeting notes, or campaign setup checklists, versus tasks that require real strategic judgment about a client's brand or market. Naming that distinction clearly, even informally, gives you language for client conversations that's more credible than a blanket AI claim, and it's the same distinction Publicis is now using in public.

If you want help figuring out where AI tooling actually fits your team's workflow versus where it would just add noise, book a free consultation and we'll walk through it with you.