The French group Publicis Groupe is changing its strategy, directly addressing CEOs of companies who are concerned about the challenges of artificial intelligence (AI) and changes in consumer behavior. This was announced by the group’s chairman and CEO, Arthur Sadoun. “We are increasingly seeing CEOs getting involved in our work,” Sadoun told analysts during the third-quarter earnings call. He added that CEOs show great interest in AI, understanding that to maintain leadership in their companies, they need to understand this technology.
Publicis: A New Approach to Selling AI Solutions Directly to CEOs
Why Are CEOs Interested in AI?
According to Sadoun, top management’s concern about AI has driven client spending and new contracts for Publicis in the third quarter. This is confirmed by a recent victory over PepsiCo, which Publicis won without a traditional tender. At the same time, the group declined to participate in a competitive tender for Coca-Cola’s media business.
“It has to start from the top,” Sadoun emphasized, speaking about the implementation of AI strategies.
Publicis Sapient, the technology services division, played a “key role” in securing “the two biggest contracts of the year without a pitch.” It also helped the group secure three “AI-powered transformation assignments this year outside of traditional tender processes.”
Record Results and Revenue Growth
This tactic brought Publicis a “record” number of new business wins in the third quarter, according to the group’s earnings report. Sadoun noted that new clients contributed to a 2% net organic growth, equivalent to 300 million euros (335.5 million dollars) in additional revenue by 2027.
Publicis led the industry in new business wins in the first half of 2026, according to COMvergence.
This was influenced by the “new market reality” after Omnicom acquired Interpublic Group last year.
Fewer competitors mean more opportunities to compete on capabilities and business cases, rather than price. The shrinking competitive field also contributes to some clients opting out of formal reviews.
Managing Conflicts and Client Expectations
Sadoun believes this will not lead to an increase in client conflicts. “The three players [Publicis, Omnicom, and WPP] are large enough to manage conflicts,” he said, provided that agencies are transparent with brands and maintain data “firewalls” to ensure client confidentiality.
Turbulence among major competitors was not in vain. “Some have opted to outsource their IT specialists,” Sadoun noted, hinting at Omnicom’s recent decision to outsource hundreds of IT and development jobs. “Clients don’t expect us to reorganize, lay off people, or stop acquisitions. They expect us to invest on their behalf,” he added.
AI’s Impact on Revenue and Strategy
Increased spending, driven by AI, led to a 5% increase in net revenue to 3.529 billion euros ($3.95 billion) compared to the same period in 2025. Net revenue in North America grew by 4.4% and accounted for 59% of the group’s revenue.
Publicis’ media business accounted for 62% of its revenue in the quarter. Sadoun emphasized that much of the underlying revenue growth was driven by client spending to get ahead of consumer AI use for product discovery and shopping.
“Clients need to adapt how they reach and engage with [consumers],” he said. “The need to adapt is immediate. Clients are willing to allocate budgets faster because moving too slowly can mean losing market share.”
Ramon Laguarta of Pepsico recently spoke to investors, defending the company’s decision to hand its global media account to Publicis Groupe. Agencies like Goodway Group are implementing synthetic data stacks into media planning, while clients are using the technology for product design and creative strategy.
Frequently Asked Questions
How to choose a contractor for full-cycle video production?
When choosing a contractor for full-cycle short video production, it’s important to pay attention to their portfolio, experience with companies in your industry, pricing transparency, and the availability of specialists for all stages: from idea to posting. Ask about their capabilities for social media content outsourcing and how they ensure integrated promotion through short videos.
Is unique content or new content better?
The decision between repurposing existing content and creating new content depends on goals and budget. Unique content can be a more economical option for quickly obtaining content, while shooting from scratch provides maximum originality and control over the message. It’s important to assess what best suits your strategy and will allow you to effectively order both a banner, an edit, and posting.
What’s cheaper – AI or real clipping?
AI-clipping vs. Human Clipping: What to Choose?
AI-clipping can be significantly cheaper and faster for mass production of short videos from long materials. However, for content requiring a subtle understanding of context, creativity, or a specific style, real human clipping may be more effective. Compare cost and quality to understand what is best for your project and how to choose a clipping service.
How to choose a clipping service?
When choosing a video clipping service, pay attention to the following criteria: processing speed, result quality (especially if AI is used), integration capabilities with your platforms, flexibility in style customization, and availability of additional services such as adding subtitles or graphics. Clarify how they ensure mass-posting quality criteria.
Conclusion
Publicis’ shift to direct sales of AI solutions to CEOs demonstrates a deep understanding of changing market needs. In an environment where AI transformation is becoming critical for business survival, companies are looking for partners who can provide full-cycle video production turnkey and comprehensive solutions. This emphasizes the importance of choosing a reliable partner who can replace an SMM team and offer all services in a single window, ensuring budget and nerve savings. If you aim to effectively use AI for marketing and need expert support, consider contacting us to discuss your tasks.
In today’s business environment, many leaders face the same headache: how to optimize workflows and free up team time for strategic tasks. Instead of just using AI for one-off requests, imagine having a whole staff of AI employees who perform routine operations, following your standards and schedule. This article will explain how to turn AI from a simple tool into a full-fledged member of your team, capable of taking on some of the work and increasing business efficiency.
Implementing AI: From One-Off Requests to a Systemic Approach
Many companies claim to actively use AI. However, as Callan Faulkner, founder of The Uncommon Business, notes, this often boils down to one-off requests. For example, a sales manager might spend an hour creating a commercial proposal using Claude or ChatGPT, but the next time a request comes in, the process starts from scratch. This is inefficient and does not allow for the full potential of the technology to be utilized.
Lack of reusable instructions.
Failure to create a knowledge base (successful proposals, objection handling methods).
AI does not become a repeatable system that supports the process.
Creating an AI employee is a completely different skill. Callan defines it as “a trained, reusable AI system that performs a specific business function as well as or better than a human.”
Benefits of a Systemic Approach to AI
Systemic AI implementation brings significant benefits. Callan Faulkner’s company, The Uncommon Business, projects $40 million in revenue with a staff of about 50 people. A few years ago, such a result would have required a much larger team. The goal is not to reduce staff, but to unlock human potential.
“Callan has never fired a person because of AI. Instead, AI allows employees to focus on strategic and creative work that truly inspires them.”
However, a gap is already noticeable between employees trained in AI and those who are not. A marketer without AI skills can spend weeks creating landing pages and images, while a trained specialist does 10 times more in a day. The problem is not replacing humans with AI, but the need for humans who can effectively manage AI.
Three Fundamental Concepts for Creating AI Employees
Before you start creating an AI employee, you need to master two key concepts and one important asset.
Concept 1: Finding the shortest path to an “A+” result
Many professionals, by habit, strive to do everything themselves, proving their worth. Today, this becomes a burden. Callan’s philosophy: the fast track to an outstanding result is a new competitive advantage. This does not mean “cutting corners,” but rather optimizing the process.
For example, Callan’s team used to spend days developing Instagram carousels. Now, after training Claude Design on examples of their best work, they create a polished carousel in eight minutes, and ten carousels in half an hour. The quality remained the same, but time and energy costs were reduced.
Concept 2: Training equals quality of result
Without proper training, AI produces generic and impersonal content. To get a unique and high-quality result, you need to document standards:
Publication goal.
Brand tone and voice.
Examples of successful content.
If these details are not documented and passed on to the AI, it will have nothing to base itself on. Callan gives an example from her podcast: every step of the process must be described in detail so that AI employees can work at the proper level.
Asset: Your “business brain”
A “business brain” is a centralized repository of all important company documentation: prices, processes, standards, brand voice, ideal customer profiles, past successes, etc. If the documentation is insufficient for a new employee, it will be insufficient for AI. AI employees are only as effective as the data they have access to. The time of chaotic files on Google Drive is over.
How to build an AI employee from scratch
The process begins with an audit of current tasks. Callan recommends identifying all daily, weekly, or monthly tasks that generate revenue but do not bring satisfaction or are valued at less than $50 per hour. This could be:
Content design and editing.
Creating presentations.
Generating commercial proposals.
Writing email newsletters.
For example, Callan’s copywriter, Nick, spent most of his time writing sales pages. He documented his approach, analyzed the most successful pages, and began creating AI employees to produce content. Now Nick manages a team of AI copywriters and QA specialists, focusing on strategy and creative direction.
Professional tip: Even tasks in which you feel like an expert are worth analyzing. Callan found that after training AI in processes in which she considered herself the best, the AI’s results sometimes surpassed her own. Transitioning from creator to editor is a more efficient use of time.
AI Interview Method
Once you’ve defined the task, open a new chat in Claude and describe the situation in detail:
Who you are.
What the company does.
What specific task the AI should perform.
What the ideal outcome looks like.
Then, ask Claude to conduct an interview by asking five high-impact questions to extract your exact process. Callan prefers voice input as it provides richer context. Tools like Wispr Flow allow you to speak rather than type, which facilitates a more complete and nuanced description.
“Board of Directors” Technique
For complex decisions, Callan uses Claude to assemble a team of 3-5 experts in the relevant field to discuss the best approach. For example, when creating executive compensation packages, she asked Claude to draw on frameworks from Mark Cuban and Sara Blakely. This resulted in recommendations based on proven business thinking, rather than generic search results.
Training and Testing AI Employees for Outstanding Results
Once you’ve achieved an excellent result in a chat with Claude, the next step is to save it as a “skill.” Within Callan’s framework, a skill is a saved, reusable set of instructions that is available at any time and connected to Claude. It’s like a “prompt on steroids”: one command triggers a complex, trained process. “A skill is an AI employee,” says Callan.
Managing Skill and Memory Changes
After creating a skill, Callan asks Claude: “What do I need to do to turn this into a reusable skill?” The answers usually point to creating a Claude Project — a workspace where knowledge files, instructions, and skills are stored. For example, for a sales proposal skill, the project might include past successful and unsuccessful proposals, all pricing information, and 15 key objections with detailed descriptions of how to handle them.
The final layer is connectors. If a workflow involves external tools (e.g., saving a proposal to a CRM or sending it via Salesforce), Claude connectors allow the skill to interact directly with these platforms. A fully built AI employee works as a skill, based on a knowledge base, with activated connectors, and the entire team is trained to use it.
Testing and Refinement
After creating a skill, the real work begins. Callan advises: test each skill to exhaustion. Most people are content with mediocre results from AI. Callan recommends persistently demanding better. If Claude produces something mediocre, she explicitly states that it can do better and asks it to try again. Often, it is after such a “push” that the desired result is achieved.
Her skill as a voice copywriter, one of the most valuable AI employees, required fifteen hours of work. This investment is justified when compared to human training. It took three months of weekly work for her copywriter to start writing in her style. Many, however, give up on AI training after 45 minutes, believing it cannot grasp their style.
The testing process follows a specific pattern:
Launch the skill.
Check the result.
Manually correct parts that do not meet the standard.
Return corrections to Claude with a specific request: “I made the following changes to your output to meet my standards. Please update your skill to incorporate these changes and use them in future outputs.”
A useful trick: during the conversation, ask Claude: “What have you learned from our interactions?” and ask it to update the skill so that these lessons are reinforced. An important technical detail: skills and memory in Claude Projects can become out of sync. Claude can update its memory based on the conversation but not update the underlying skill. Callan recommends explicitly telling Claude to update the skill itself, and then manually clicking the save button.
Projects also have their own invisible project-level skills that can conflict with workspace-level skills. When creating a skill that should work across all projects, specify “workspace skill” or “core skill” so that it is available everywhere, not just in one project.
Pro Tip: For teams building skills at scale, Callan’s company copies each skill into a Notion database, tracking who created it, when, the version, and its purpose. They even have a skill that checks their skill database for duplicates. Each department manages its skill repository, and skill management is a standard part of every quarterly review.
Scheduling AI Employees for Autonomous Execution
Once a skill has been thoroughly tested and refined to a high standard, it can be scheduled for autonomous work without human intervention.
Current limitation: the computer running Claude must be turned on for scheduled tasks to execute. Callan uses a dedicated office computer logged into the corporate Claude account. Any employee wishing to schedule a skill logs into this account on that machine.
Callan uses Claude’s scheduled task feature via the Claude Cowork desktop application. There are options to select the day, time, and frequency of skill execution. There is no limit to the number of active scheduled tasks.
One of Callan’s scheduled skills is the “Instagram Researcher.” Every Monday morning at 6:00 AM, it visits competitor Instagram accounts, identifies viral content in the AI education sphere, analyzes the approaches and “hooks” used, extracts content transcripts from the previous week, and adds content ideas directly to the Notion database. Her social team logs in on Monday morning and manually approves the ideas they want to implement. What used to be a routine task for the social team on Monday morning is now performed autonomously.
Another scheduled skill runs hourly, extracting meeting transcripts from Granola and archiving them into a Notion database that powers her “second brain.”
Conclusion: Invest in AI Employees for Business Scaling
Creating AI employees is not just automation; it’s an investment in the strategic scaling of your business. Shifting from routine operations to managing efficient AI systems allows your team to focus on creativity and innovation, addressing “customer pain points” related to lack of time and resources. If you are looking for full-cycle short video production or comprehensive promotion through short videos, consider how AI employees can take on some of the work. This allows you to order everything in one place, effectively replacing part of the team with a service. Are you ready to rethink your business processes and build a team where AI and people work in harmony? Start with a task audit and document your standards today. Learn more about how to choose a contractor for social media content outsourcing, and what mass posting quality criteria are truly important to avoid wasting your budget.
Frequently Asked Questions
What is an AI employee and how does it differ from regular AI use?
An AI employee is a trained, reusable AI system that performs a specific business function as well as or better than a human. Unlike one-time AI requests, an AI employee is part of a systematic approach that automates repetitive tasks, following defined standards and schedules.
How do I start creating my first AI employee?
Start with a self-audit: identify tasks that are time-consuming, revenue-generating, but unfulfilling or undervalued. Then document each step of performing that task, its purpose, quality standards, and examples of successful outcomes. Use the AI interview method in Claude to describe the process in detail.
Do I need to be a technical specialist to create AI employees?
For 90% of business operations, creating AI employees does not require technical skills. The key is communication (the ability to clearly articulate needs) and creativity (resourcefulness to bring AI to an outstanding result).
How to ensure high-quality results from an AI employee?
The key to quality is thorough training and testing. Document all standards, tone of voice, examples of ideal results. After creating a skill in Claude, test it exhaustively, demanding improvements from the AI until the result meets your “A+” standards. Don’t forget to regularly update the AI’s skills and memory.
Can AI employee work be automated on a schedule?
Yes, once a skill is debugged, it can be scheduled for autonomous execution. Use scheduling features in tools like Claude Cowork to set the day, time, and frequency for the skill to run. This allows for the automation of routine tasks such as data collection, competitor analysis, or content generation.
