Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

5 steps to reengineer with AI agents

Let AI transform your daily work.

[Photo: Getty Images]

Profile photo of Teresa Barreira

Chief Marketing Officer · Publicis Sapient

By the time I was 10, I was running my parents’ grocery store, opening and closing the shop, making quick decisions, and solving problems in real time. Every day demanded judgment, intuition, creativity, and a human touch.

I didn’t know it then, but that experience shaped how I think about work, leadership, and responsibility. Those lessons matter because AI is changing what “the work” is. Leaders have to be clear about what stays human-owned, what moves to systems, and where judgment still carries weight.

This is the most profound transformation of my career. AI uncertainty often causes hesitation. But the longer you wait to adapt, the further behind you fall.

The mistake most leaders make

Many organizations still layer AI onto existing ways of working. That approach automates work without changing it.

AI has the potential to reshape how knowledge work happens and how decisions are informed across the enterprise. If organizations treat AI as a bolt-on, they miss the opportunity to rethink roles, workflows, and priorities. Capability without context creates bad outcomes. What’s required is a redesign, because the gain happens when AI starts collaborating.

Ask a different question

My team faced a familiar challenge: We needed to grow faster than the market, but couldn’t scale our budget and headcount at the required pace.

We took what I call an “agent-first” approach to re-engineer our marketing organization. Instead of asking, “How can our teams use AI?” we asked, “What can AI do on its own?” We identified what AI could do to determine how people spend their time.

This forced us to reinvent our marketing operating model. We hosted 12 agent-design workshops and mapped more than 700 tasks across brand, social, revenue marketing, content, strategy, operations, and PR—from idea review and campaign briefs to social planning, email drafting, and brand approvals.

In those workshops, we broke work into discrete tasks and debated where AI could lead or accelerate and where human judgment had to stay in the loop.

The impact has been measurable:

  • Campaign timelines were compressed from 20 days to 3-5 days, with the process reduced from more than 250 reviews, approvals, and transfers to just 11 coordinated steps.
  • We delivered seven times as many lifecycle programs, moving more prospects into sales-ready conversations and generating 20% more qualified opportunities.
  • We ran 20 times as many creative tests at the same media spend, testing more messages and creative versions without increasing spend, which helped prospects move through the funnel 15% to 25% faster.
  • We saw a 10% to 20% conversion lift from hyper-personalized, intent-based messaging, where messages reflected people’s demonstrated interests.
  • We had three to four times more campaigns live simultaneously without adding headcount, giving us capacity for work that previously wasn’t on the roadmap.

AI accelerated nearly 90% of our marketing tasks. The endgame was growth and capacity for work we couldn’t previously resource.

5 steps to an AI agent-first reorganization

If you want to make this approach real in your organization, here’s the sequence we used.

Step 1: Experiment


Give your team time to learn different AI tools. We also built our own agentic campaign management system, which helped the team learn by doing—building, testing, and refining AI in daily work—faster than any classroom-style training program could.

Step 2: Map tasks


Document tasks across functions. Classify what is AI-led versus human-led. For example, AI might lead a campaign brief draft or social campaign research and planning, while humans still own the brand narrative and the final strategic call.

Step 3: Build AI assistants

Have team members build task-specific assistants. Then evaluate them relentlessly to close gaps. In practice, that meant testing output for quality, context, and brand voice, retraining them when they missed the mark, and giving them more responsibility only as trust was earned.

Step 4: Redesign workflows

Embed assistants into end-to-end processes. Our campaign launch went from 50-plus handoffs to 11 orchestrated steps, with human checkpoints only where judgment mattered. For example, deciding whether a draft was actually any good, on-brand, and right for the moment, before moving forward.

Step 5: Redefine roles

Write new job descriptions based on the new operating model. In our case, brand marketers moved from campaign managers to cultural and brand visionaries, while social marketers evolved from schedulers to active voices in the conversation.

We also protected learning time, offering AI Thursdays for exploration, Failing Fridays for failure-related insights, and a Monthly Pause for Learning.

Leadership and work moved upstream

As AI took on execution, coordination, and analysis, human work moved upstream. Brand marketers shaped direction and meaning, social teams shaped bigger conversations, and revenue marketers made faster strategic decisions that drive growth.

We treated AI agents like new hires. Agents were trained, tested, and given increasing responsibility as trust was earned. Early attempts at full autonomy failed.

In one early system, AI could generate a long-form piece, a short-form version, a social post, and an ad in minutes. But the output was too generic, lacked context, and often had to be rewritten. So, we started small, iterated, and delegated more as confidence grew.

We found context is non-negotiable. AI can execute, synthesize, and analyze at scale, but can’t replace cultural fluency, strategic judgment, ethical accountability, or the ability to make a consequential call under pressure. Those capabilities belong to people.

The real transformation

The most important shift was human. When execution moves to systems, people set the vision, interpret context, shape meaning, build trust, and make decisions that matter. That’s where leadership lives.

Builders, explorers, and learners adapt fastest because they’re willing to experiment, iterate, and let go of old ways. Most leaders ask how they can use AI to do more. The better question is: What would you do if the constraints you’ve operated under for years were suddenly lifted?

For us, the answer was growth we couldn’t previously achieve. That’s the real transformation: It starts not with technology but with rethinking the work.

Teresa Barreira is the chief marketing and communications officer at Publicis Sapient.