The Right Strategies for Implementing AI Agents
Photo generated by Gemini

The AI Agents Buzzword!

We hear more and more about it, don’t we? What seemed futuristic a short while ago is already transforming the way we work. If you’re wondering how to concretely approach their implementation in your company, you’ve come to the right place! Let’s draw inspiration from shared experiences in the sources to outline a clear path.

What Exactly is an AI Agent?

Forget the basic chatbot that answers FAQs. An AI agent is a system that can perceive its environment, make decisions, and act autonomously to achieve specific goals. Thanks to machine learning and natural language processing, it understands, learns, and adapts, even when the context changes. It doesn’t just analyze data or make predictions; it can actually execute tasks autonomously.

We can distinguish two main types for enterprise software:

  • Task-oriented agents: excellent for automating repetitive or complex tasks in a specific domain, such as processing invoices or scheduling appointments.
  • Role-oriented agents: designed to assist specific employees, understanding the subtleties of their role and taking on a broader range of responsibilities to increase productivity and efficiency. For Workday, they are the ones who will define the future of work.

Why Pay Attention Now?

Because companies that take the plunge are already seeing concrete results. According to a survey, 79% of employees state that AI agents have had a positive impact on their company’s performance. Another survey among executives reveals that over half of companies (51%) have already deployed AI agents, and an additional 35% plan to do so in the next two years. By 2027, 86% of companies plan to be operational with AI agents.

Optimism is high: more than three-fifths (62%) expect a return on investment over 100% with agentic AI, with an average expected return of 171%. Adoption is even expected to be faster than that of generative AI, as the learning curve of GenAI has paved the way for a smoother transition.

AI agents can help streamline business processes, improve performance, reduce costs, and maintain service quality. They can anticipate user needs and act proactively.

Okay, But What Are the Obstacles?

It’s not always simple. Companies face several hurdles:

  • Cybersecurity fears are a major obstacle, notably the risk of sensitive data being accessible via generative AI. Paradoxically, not taking action can push employees to use unsecured personal tools, increasing the risk of data leaks.
  • A lack of AI training and knowledge among executives and employees.
  • The difficulty in identifying appropriate use cases and clearly seeing the positive effects.
  • The reluctance to fund projects. AI is a real investment with significant implementation costs.
  • Challenges related to security and privacy.
  • The complexity of integration into existing systems.
  • The blurry and evolving legal framework.
  • The need for a sufficient quantity of quality data to feed the AI.
  • Staff buy-in and the fear of the unknown.

These challenges highlight that AI doesn’t “work on its own”; it requires an understanding and apprehension of the phenomenon. Rushing in without planning is a mistake that companies having adopted GenAI seek to avoid with AI agents.

How to Go About It? Key Success Factors

For a successful deployment, the sources emphasize several crucial elements:

  • Strategic Alignment: Leaders (CEO, CFO, COO, HRD) must align on objectives and integrate AI into the company’s overall vision. Understanding the problem to solve is essential.
  • Strong Governance: Setting up a dedicated team (tech experts, compliance, legal, business units) is vital for coordination and oversight.
  • IT & Data Investments: Investments are needed in technologies and data. The HR/CIO duo must coordinate this. An AI is only effective if fed with quality data.
  • Data & AI-Based Tools: Use or develop solutions tailored to the specific needs of each profession. The company must understand where it stands in terms of capabilities, human resources, and experience.
  • Social Strategy and ROI Management: Supporting employees (reskilling, training) is indispensable. It’s also necessary to measure the efficiency and impact of AI by combining HR costs and operational gains. Defining clear expectations for return on investment is crucial.

It is important to adopt a user-centric approach, designing intuitive agents that integrate into existing workflows. Trust and transparency are key, notably by choosing reliable providers and establishing internal principles for AI usage.

A Step-by-Step Methodology

The sources suggest a structured approach:

  1. Step 1: Raise Awareness and Acculturate to AI. All employees must be made aware and educated so they understand how AI can transform their daily lives and accept it as an ally. All companies plan to train their employees, often via internal seminars or external courses.
  2. Step 2: Map and Analyze Needs. Identify current processes and systems, pain points (inefficiencies, errors) where AI can have the most impact. Clearly define the objectives.
  3. Step 3: Develop and Integrate AI Solutions. Develop or integrate suitable solutions to automate repetitive tasks and optimize value-added processes. Start with simple and repetitive tasks. It is important to decide early on whether AI will be internalized or outsourced. Organizations must master simpler use cases (like summarization) before attempting complex multi-agent systems.
  4. Step 4: Continuously Train, Evaluate, and Adjust. Provide continuous training. Gather user feedback and adjust strategies. Measure the impact of AI initiatives with key performance indicators. Continuous evaluation and experimentation are essential. Having the right observability tools is crucial at every step.

Potential Gains in Action: Some Concrete Examples

AI agents aren’t just theoretical; they are already transforming operations across various sectors:

  • Lead Generation: Waiver Group used its bot Waiverlyn to capture, qualify prospects, and book consultations. Result: 25% increase in consultations and visitor engagement multiplied by 9. The bot paid for itself in 3 weeks.
  • Customer Service: Ruby Labs handles over 4 million chats per month. Their AI bot resolves 98% of chats without human intervention. It also saved $30,000 per month by offering targeted discounts before cancellation. Botmind helps e-merchants automate frequent responses. La Bécanerie automates over 90,000 conversations a year. Maison Lascours reduced its contact rate by about 30%. Sena automated 50% of its support. Artsper saved 5 hours per agent per month. Cobbaï integrated an AI into an ERP to qualify customer requests and generate automatic responses; 45% of tickets no longer require human intervention, and manual processing time was cut in half.
  • Competitive Intelligence: A bot at Botpress scans competitor sites, detects changes in price, features, SEO, etc., summarizes, and sends reports. It provides a competitive advantage.
  • Content Discovery: Pinterest uses an agent that analyzes data, adapts to interactions, and personalizes feeds. This contributed to reaching 553 million monthly active users, an 11% increase in one year.
  • Trend Forecasting: Zara’s AI agent analyzes social platforms and purchasing data to spot emerging trends. This helped increase sales by 7% between 2023 and 2024.
  • Travel Recommendation: American Express uses an AI assistant to help its counselors create hyper-personalized suggestions. Over 85% of counselors say AI saves them time and improves the quality of their recommendations.
  • HR Support: Harry Botter, an AI agent at Botpress, answers HR questions on Slack, checks leave balances, finds documents, and helps with onboarding. It provides instant answers, saving HR time. HR AI agents can improve the employee experience.
  • Sales Assistance: Coach AI at JPMorgan helps advisors by extracting research, anticipating questions, and suggesting recommendations.
  • Route Optimization: UPS’s ORION plans routes in real-time. Gains: 100 million miles saved/year, $300 million reduction in annual costs, reduction in carbon emissions by approximately 100,000 metric tons.
  • Diagnostic Imaging: Aidoc’s agent detects and prioritizes urgent cases (like pulmonary embolism) in CT scans. It spotted 14 severe cases missed otherwise, allowing for faster decisions and better care.
  • Recruitment Diversification: PwC used AI for skills matching, allowing them to diversify the retained talent, going from 3% “atypical” profiles to about 15%.
  • General Optimization: AI agents can also streamline HR processes, optimize supply chain management, and improve Financial Planning & Analysis (FP&A).

More than half (52%) of companies expect agentic AI to automate or accelerate 26 to 50% of their workloads.

In Conclusion

Agentic AI is much more than simple automation or an interface like a chatbot. It is a technology capable of deeply transforming how companies operate by autonomizing tasks, personalizing interactions, and offering valuable insights. The approach must be strategic, focused on understanding real needs, supporting teams, and concretely measuring benefits. Deployment requires time and a progressive approach. The question is no longer whether you will use AI agents, but when.

Companies of all sizes and in many sectors can benefit, whether in healthcare, finance, retail, manufacturing, marketing, sales, HR, or even agriculture. For the Quebec market, specialized consultants can help navigate local specificities, including regulatory compliance.

Ready to embark on this revolution?

The Right Strategies for Implementing AI Agents
Older post

2025 Status Report on Cloud Threats and Strategies

As the year is well underway, let's take stock of the specific threats to the Cloud and recommended strategies.

Newer post

The responsibility of security in the Cloud

When security is everybody's business, it becomes nobody's responsibility. Why total decentralization of IT security is a bad idea.

The Right Strategies for Implementing AI Agents