AI and automation expert on how leaders use AI agents to get ahead | Pascal Bornet

Key Concepts

• AI Agents: Digital entities that can understand context, plan actions, act on those plans, and learn from outcomes to improve themselves. They are an evolution from simpler automation tools.
• Agentic Artificial Intelligence: A paradigm where AI agents are central to reinventing business, work, and life, moving beyond simple task execution to proactive and intelligent assistance.
• Large Language Models (LLMs): The foundational technology (e.g., ChatGPT) that has enabled the current wave of generative AI, capable of understanding and generating human-like text, planning, and answering questions.
• SPAR Framework: A methodology for understanding and designing AI agents, encompassing Sensing, Planning, Acting, and Reflecting.
• Humics: Uniquely human abilities that AI cannot fully replicate, including genuine creativity, critical thinking, and social authenticity.
• Human-Technology Collaboration: The optimal integration of human capabilities with AI agent capabilities, where each complements the other.
• Compounding Intelligence Advantages: The continuous improvement and learning of AI agents over time, creating a growing competitive edge for early adopters.
• AI Literacy: The ability to understand AI technologies, identify their potential benefits, and effectively integrate them into work and life.

Agentic AI: Transforming Business and Work

Pascal Bornet, a leading expert in AI and intelligent automation, discusses the transformative potential of AI agents in reinventing business, work, and life. With over two decades of experience implementing automation technologies at firms like McKinsey and EY, Bornet emphasizes that successful transformations prioritize people at the center of the change.

The Evolution of Automation to AI Agents

Bornet traces the evolution of automation from early robotic process automation (RPA) and script automations, which performed autonomous actions for humans, to the current era driven by Large Language Models (LLMs). LLMs, exemplified by tools like ChatGPT, have significantly enhanced AI's capabilities, enabling agents to understand context, plan actions, and learn from their outcomes.

Key Distinction: While LLMs are adept at generating information, planning, and creative tasks (e.g., planning a weekend trip), they lack the ability to act on these plans. This is where AI agents differentiate themselves.

Information vs. Action: Bornet highlights that information without action has limited impact. AI agents bridge this gap by not only suggesting but also executing tasks, leading to tangible outcomes. This extends to numerous business use cases, from data entry and invoicing to client relationship management.

Paradigm Shift: Bornet predicts a future where "there's an agent for that" replaces "there's an app for that." Unlike current apps that require human learning, agents will learn to serve humans, orchestrating various functions through a single intelligent interface. This signifies a shift from humans adapting to technology to technology adapting to humans.

The SPAR Framework for Agentic AI

Bornet introduces the SPAR framework as a crucial tool for understanding and designing AI agents. This framework outlines four key functions:

1. Sensing: The agent perceives and understands its environment, including human input or external data.
2. Planning: Based on the sensed information and a defined goal, the agent formulates a sequence of actions.
3. Acting: The agent executes the planned actions.
4. Reflecting: The agent evaluates the outcome of its actions against the goal, learning from the experience to improve future performance.

This framework is analogous to human decision-making processes, such as cooking a meal, and provides a structured approach for leaders to assess agent capabilities and compare different technologies.

Real-World Applications and Business Impact

Bornet shares compelling examples of agentic AI in action:

• Pets at Home (UK): Achieved 99% accuracy in transcribing veterinary consultations using an ambient digital scribe, freeing up veterinarians' time for patient care. They also employ agents for fraud detection (identifying duplicate refund requests) and real-time insurance coverage checks during consultations. A store colleague assistant provides personalized guidance to staff.
• Invoice-to-Pay Agents: Automate the process of reading invoices, matching them with purchase orders, approving them, and initiating payments based on predefined conditions.
• Recruitment Agents: Source candidates, screen CVs, and schedule interviews.
• Learning Agents: Personalize employee training paths based on performance and roles.
• Marketing Campaign Agents: Design, test, and optimize multi-channel marketing campaigns in real-time.
• Retention Agents: Detect client churn risk and launch proactive retention offers.

Quantifiable Impact:

• JPMorgan: Reduced fraud by 70%.
• McKinsey: Reduced client onboarding lead time by 90%.

Critical Success Factors for AI Transformation

While many leaders plan to adopt AI and agents (over 80% in a Microsoft Work Trend Index), successful implementation requires more than just ambition. Bornet identifies common pitfalls:

• Improper Design: Agents not performing actions correctly or lacking adequate security and guardrails.
• Limited Scope: Focusing on single use cases rather than integrating across departments for scalability and economies of scale.

Key to Success: Bornet reiterates that the most successful companies place people at the center of their transformation. This involves:

• Informing: Transparently communicating the implications of AI for roles and the company.
• Educating: Equipping employees with the knowledge to understand and utilize AI technologies.

Frontier Firms and Competitive Advantage

Bornet describes "frontier firms" as those that are AI-first and strategically integrate AI. These firms exhibit four key characteristics:

1. Thinking Beyond Automation: Reimagining business processes rather than simply automating existing ones.
2. Investing in Human-Technology Collaboration: Redesigning work to optimize the synergy between human and agent capabilities.
3. Building Agent Ecosystems: Creating interconnected networks of specialized agents that collaborate and learn from each other, avoiding siloed solutions.
4. Building Compounding Intelligence Advantages: Leveraging the continuous learning and improvement of AI agents to create a growing and insurmountable competitive moat. Early adopters gain an advantage not only through data generation but also by accustoming their workforce to hybrid digital teams.

The Power of "Humics"

Bornet introduces the concept of "humics" – uniquely human abilities that AI cannot fully replicate. These are crucial for avoiding redundancy and creating complementarity:

1. Genuine Creativity: Generating truly original ideas rooted in lived experience, emotions, and personality, beyond recombining existing information.
2. Critical Thinking: Exercising nuanced judgment, questioning assumptions, navigating ethical complexities, and asking the right questions.
3. Social Authenticity: Building deep, trust-based relationships through genuine empathy, communication, and shared consciousness.

These abilities are innate and can be continuously developed, offering a distinct human edge.

Managing Hybrid Human-Agent Teams

The integration of humans and agents necessitates a shift in management styles:

• Clear Role Definition: Humans focus on value-driving activities requiring emotional intelligence, while agents handle routine tasks and process coordination.
• Orchestration: Managers become orchestrators, designing human-agent workflows, shifting from task assignment to goal setting and boundary definition.
• Building Trust: Managers must foster trust between human team members and agents through clarity, transparency, and gradual collaboration. This involves starting with small, low-risk tasks and progressively expanding the agent's responsibilities as trust is built.

Metrics for Success in a Hybrid Environment

Measuring success requires a comprehensive review of metrics for both humans and agents:

• Human Metrics: Incentivize experimentation, learning, and the development of uniquely human skills (judgment, creativity).
• Agent Metrics: Monitor decision quality, collect user feedback, and set up alerts for performance.
• Hybrid Team Metrics: Combine qualitative (customer/employee experience, satisfaction) and quantitative (cost, time, efficiency) metrics to provide a holistic view of performance.

Navigating Rapid Technological Change

Leaders must manage the accelerating pace of technological change by fostering three future competencies:

1. Being Human-Ready: Developing uniquely human abilities ("humics").
2. Being Change-Ready: Adapting to the increasing magnitude and frequency of change.
3. Being AI-Ready: Identifying, testing, and using AI technologies that offer meaningful benefits.

This requires dedicating time (e.g., 20% of working time) for employees to explore new AI developments, experiment, and learn. The focus should be on developing an AI literacy mindset rather than traditional, quickly outdated classroom training. Incentivizing experimentation and learning, even from failures, is crucial.

Blueprint for Starting the AI Journey

For CEOs looking to embark on this transformation, Bornet offers a blueprint:

1. Clear Vision from the Top: C-level executives must have a clear vision for how AI will transform the company, addressing specific business issues rather than adopting technology for its own sake.
2. Invest in People and Talents:
   • Establish a "center of excellence" or AI talent group, potentially starting with one knowledgeable individual.
   • This group identifies use cases, builds business cases, and develops the necessary technical teams.
   • Invest in informing, educating, incentivizing, and empowering employees with access to the right tools.
3. Start Small, See Big: Begin with a pilot project that demonstrates business impact and gain visibility to create momentum and encourage broader adoption.

Bornet concludes by emphasizing that successful AI transformation is an ongoing journey, not a one-off event, and that prioritizing people is the most critical factor for sustained success.