IN THIS ISSUE

OpenAI chief product officer Kevin Weil claimed 2025 to be the year of the AI agent at the World Economic Forum in Davos in late January.

• But, first - What is an AI Agent?

• How Do AI Agents Help Businesses?

• The AI Agent Battles

TOP PICKS

But, first - What is an AI Agent?

An AI agent is a system designed to reason through complex problems, create actionable plans, and execute these plans using a suite of tools.

It can think > plan > act > reflect. An AI agent can process the given data and context, strategize, execute that strategy, and reflect and refine its responses.

AI Agents use LLMs to power their work.

These LLMs can be publically available models or open-source models like Meta’s LLaMA 3., and Google’s BERT. They can be small LLMs that work efficiently on local hardware. Additionally, these can be trained and tuned to work in any environment.

What happens when you ask an AI agent to work on a task?

1. The user or system triggers a request.

2. LLM-based agent uses an internal reasoning loop (think > plan > act > reflect) to interpret the request and plan the next steps.

3. The agent makes calls to GenAI APIs or enterprise data/services (via GraphQL, OpenAPI, etc.) to gather more information.

4. he agent processes the gathered data and formulates a response or an action.

5. It executes that action or generates a final output to the user or system.

6. Logging and monitoring systems record metrics (accuracy, compliance, fairness, relevance) so that developers and administrators can refine the agent’s behavior.

Why This Matters

AI Agents will, soon, become mainstream. Just like human colleagues.

We will be working with them closely for specific tasks.

For now, they are in an experimental stage and businesses/researchers are still figuring out how to work and manage AI agents. (We will cover the challenges to governing AI agents compared to human employees in our next edition of the newsletter).

Credits: Armand Ruiz, VP Product - AI Platform, IBM

How Do AI Agents Help Businesses?

AI agents are capable of thinking, reasoning, and learning, and they even know when to seek help.

These agents can carry out business processes, ranging from basic prompt-and-response interactions to more sophisticated, fully autonomous agents that handle an entire workflow from start to finish.

You can train agents to perform new tasks, such as sending emails, creating support tickets, or updating records. They can also be set up to react to events or triggers, like receiving an email from a customer. If an agent encounters something unclear, it can always check in with you for guidance. Because these agents are integrated with your company’s data and knowledge, their actions are aligned with your business context and specific needs.

Creating agents can be surprisingly simple. Some platforms allow you to build them using natural language, making it accessible even if you don’t have an engineering background.

When would you use an AI agent?

You’ll turn to an agent when you need to apply AI to a business process, especially one that spans multiple employees and functions.

For example, an agent can manage an entire fulfillment process—everything from taking the order and processing it to making intelligent substitutions for out-of-stock items, all the way to shipping the product to the customer.

Agents unlock the ability for organizations to scale like never before.

For instance, if you’re managing a sales team, agents can handle time-consuming internal tasks, freeing up your team to focus on building strong customer relationships. And they open up new possibilities. Let’s say a customer places a large order—the agent could instantly fulfill the order and alert the sales team, suggesting it’s a prime opportunity to offer additional products. In just seconds, this small action could turn into a much more valuable deal.

Why This Matters

Picture a future where organizations operate with entire networks of agents, working seamlessly behind the scenes to refine existing processes and create new ones.

In sales, these agents will research leads, prioritize opportunities, and guide customer outreach with tailored emails and responses. In supply chain management, they’ll reduce disruptions by autonomously tracking supplier performance, spotting delays, and recommending real-time adjustments. Across your business, they’ll help cut costs, enhance decision-making, and drive continuous improvement.

Credits: Jared Spataro, CMO, AI at Work @ Microsoft

The AI Agent Battles

The real power of agentic workflows lies in their potential for optimization. Each step of the workflow can be fine-tuned to improve efficiency, speed, and cost-effectiveness. For example, a large model might be used to handle the planning and overall orchestration of the task, while smaller, specialized models could be deployed for individual sub-tasks. This division of labor within the workflow can lead to significant improvements in how tasks are performed.

Businesses can also create specialized tools tailored to the needs of their workflow. By integrating these tools, companies can further enhance the performance of their agents, making them highly adaptable to specific business processes.

Another key advantage of agentic workflows is their ability to run autonomously, often eliminating the need for human intervention. This increases scalability and allows businesses to offload repetitive tasks to AI, freeing up human resources for more strategic and high-value activities.

The Role of Context and User Interfaces

With the emergence of platforms like OpenAI’s Operator, two new dimensions of agent development have come into focus: context and user interfaces.

1. Context: The context an agent operates within is crucial for its success. To effectively carry out tasks, an agent must understand not just the immediate input but also the broader context in which it is working. This includes the organization's data, its workflows, and the specific needs of the user. An agent with access to relevant context is far more capable of delivering accurate and meaningful results.

2. User interfaces: As AI agents become more integrated into business processes, the ability to monitor and oversee these agents becomes essential. A user interface that allows individuals to interact with and manage agents can provide valuable oversight and control, ensuring that the agents stay aligned with organizational goals and deliver results in a consistent manner.

The Battle for Dominance

The competition in the AI agent market is heating up, with companies vying to create the most capable and versatile agents.

OpenAI’s Operator, for example, represents a significant leap forward in terms of agent capabilities. It combines the power of large language models with specialized tools, enabling a higher degree of flexibility and autonomy compared to previous models. However, Meta and other companies are also developing their own AI agents, each with unique strengths.

The key differentiator among these agents is the balance between specialization and generalization. 

Some agents are designed to handle very specific tasks, excelling in particular domains, while others, like Operator, aim to be more versatile and capable of managing a broader range of tasks. This ongoing battle for dominance will likely lead to the development of increasingly sophisticated agents capable of transforming business operations across industries.

Why This Matters

The AI agent market is one to watch, with major players like OpenAI and Meta pushing the boundaries of what’s possible with autonomous systems.

As AI agents become more advanced in context awareness, specialized tools, and user interfaces, they will be able to tackle increasingly complex workflows, reducing the need for human intervention and optimizing business processes at scale.

The future will likely see entire organizations built around these agent-driven workflows, with AI taking on a central role in everything from customer service to supply chain management. The possibilities for innovation are vast, and businesses that can use the power of AI agents will gain a significant competitive edge.

Read Here

FOR YOUR READING PLEASURE

Sneak Peek Into the Next Edition

1. AI Agents are Not Human Employees. Then, How Do You Govern Them?

2. Fully Autonomous Agents Should Not Be Developed.

3. 5 Trends Shaping The Future Of Leadership In The Age Of Agentic AI