Gunda Karnīte

Associate, GenAI, IT Advisory, PwC Latvia

AI agents are a hot topic in the world of technology now. They represent a new breed of existing AI solutions that aim to automate processes and improve efficiency by providing each employee with a personal assistant. Over the past year, AI tools such as chatbots have become an integral part of everyday life. In a survey conducted by PwC in August 2024, 73% of respondents stated that they already use chatbots such as Microsoft's Copilot and ChatGPT daily or plan to use them. At the moment, however, the focus is increasingly on AI agents that combine a range of the skills of chatbots with broader automation and personalisation capabilities.

AI agents are AI systems that are capable of making autonomous decisions and taking actions to achieve specific goals with limited human intervention or without direct human intervention. They offer more specialised and efficient AI solutions than the mainstream chatbots currently used.

Experts believe that AI agents will be a key factor in achieving a positive return on investment (ROI) from AI initiatives, as agent activity is a measurable variable. (AI Agents will be the key to awakening ROI from AI)

This is an essential step because until now the day-to-day AI infrastructure of many companies has consisted of chatbots based on the idea that activities are activated by chat and that they are generic, without industry-specific expertise. AI agents change these conditions as they act under the influence of defined events and are trained by a specific information set, giving them a more exact knowledge base. Such changes allow agents to constantly act in each of their domains without human intervention. For example, if a company receives a large number of customer enquiries, AI agents can handle 80% of the enquiries, leaving only the most complex cases to employees. This allows companies to reduce the manual workload, and employees can focus on unresolved tasks.

Google, Microsoft, Salesforce and other companies already offer the option of integrating AI agents into their tools. Microsoft, for example, offers its users the option of creating Copilot agents as part of the Microsoft 365 Copilot licence. Creating an AI agent is a no-code or no-code and low-code or minimal-code approach. A simple no-code agent can be created in the Copilot app and extended with minimal code on the Copilot Studio platform. Copilot Studio allows you to customise the functionality of your agents and align them with the specific requirements of your business. This flexibility is a key element in enabling organisations to respond to changing market conditions and requirements.

A common example of the use of AI agents is an email application assistant. Such an agent is activated as soon as a new email arrives in the inbox and reacts quickly to new messages. The agent automatically checks the content of the email, analyses its meaning and prioritises the necessary steps to respond effectively or forward the letter to the appropriate team or system. In this way, an agent helps to reduce the time spent managing emails and improve the efficiency of communication processes in your organisation.

To implement AI agents in your day-to-day business, applying similar principles when implementing any AI system is important, i.e., start with a small task that provides immediate feedback and sets key performance indicators (KPIs). When implementing AI agents, it is important to focus on the organization's existing AI infrastructure so that the agents can navigate and integrate successfully.

The possibilities of AI agents are enormous; they will become an important tool for everyday B2B in 2025. In the future, specialised AI agents, rather than general AI solutions will play the main role. (The rise of AI Agents: UNLOCKING their full potential)