Automate The Mundane, Elevate The Human: Inside Publicis Sapient's AI Strategy

| News | April 29, 2026 | 503 views | 33:16

TL;DR

Teresa Barrera explains why companies should avoid 'random acts of AI' and instead transform workflows before deploying technology. She details how Publicis Sapient deconstructed work into 900 tasks to let AI handle 80% of tactical work while experts train agents, enabling staff to focus on high-value activities.

🎯 Strategic AI Deployment 3 insights

Ban random acts of AI

Barrera warns against deploying AI for technology's sake without intentional transformation, as adding AI to broken workflows only accelerates existing inefficiencies.

Transform business, not just digital

Using McDonald's delivery app as an example, she emphasizes that digital transformation requires redesigning underlying business processes like kitchen workflows to match digital interfaces.

People-first transformation

Successful AI implementation starts with transforming how people work rather than starting with the technology itself.

⚙️ The 80/20 Human-AI Model 4 insights

Decompose jobs into tasks

Barrera's team identified approximately 900 discrete tasks to determine which require full AI autonomy, human oversight, or purely human execution.

The 80/20 value shift

Roughly 80% of tasks can be handled by AI or human-AI collaboration, allowing employees to refocus on the 20% of high-value work only humans can perform.

Employees as agent builders

Rather than IT building tools, subject matter experts create and train AI assistants for their specific tasks, ensuring quality and contextual knowledge transfer.

From assistants to agents

Individual task-specific assistants connect as nodes in an agentic platform to form autonomous content, brand, and lead-nurturing agents.

🏗️ Organizational Evolution 3 insights

Shift to people plus product

Publicis Sapient evolved from pure consulting to pairing proprietary AI platforms with human expertise using an Iron Man analogy where employees provide context and platforms provide scale.

Codify institutional knowledge

Every client solution feeds back into the platforms, creating compound knowledge that evolves the systems while freeing people to provide contextual expertise.

Innovation incubator model

A dedicated function spends 20% of time exploring future opportunities, separating future-focused experimentation from daily operations to prevent distraction.

Bottom Line

Deconstruct your workflows into specific tasks before implementing AI, then have your best experts train AI assistants for the 80% of tactical work while redirecting human effort toward the 20% of high-value activities that drive growth.

More from Forbes

View all
A Nobel Laureate's Honest Review of AI In Biology
30:25
Forbes Forbes

A Nobel Laureate's Honest Review of AI In Biology

Nobel laureate Thomas Südhof discusses how AI is revolutionizing biology while warning that poor data quality and industry 'AI washing' hinder progress, emphasizing that true innovation requires both rigorous data standards and protected freedoms for scientific inquiry.

about 21 hours ago · 8 points
President Trump Holds Far-Reaching Press Briefing At The Conclusion Of The G7 Summit
1:08:50
Forbes Forbes

President Trump Holds Far-Reaching Press Briefing At The Conclusion Of The G7 Summit

President Trump announced a historic agreement with Iran to prevent nuclear weapons development and reopen the Strait of Hormuz. He credited his administration's military strategy—including the Soleimani assassination and bombing of enrichment facilities—for enabling the diplomatic breakthrough that caused stock markets to surge and oil prices to drop.

1 day ago · 9 points
Why This Innovator Is Not Afraid Of AI Replacing Creativity
39:25
Forbes Forbes

Why This Innovator Is Not Afraid Of AI Replacing Creativity

Former Disney Imagineering leader Bran Ferren argues that artificial intelligence will eliminate mundane "toil" but cannot replace human creativity, emphasizing that true innovation thrives at the intersection of art and engineering rather than through rigid specialization.

2 days ago · 10 points