The Year of Agentic AI

Ali Arsanjani
3 min readJan 2, 2025

--

(Part 2 of 6)

2025 will be considered to be among other trends, the “Year of AI Agents”, where our collective experiences in the industry has melded generative AI more and more into the fabric of business systems and we transition from monolithic AI to Distributed AI integrating generative and predictive AI into practical, real-world applications.

This blog is part of a series (2 of 6) in which I explore (not delve 😉) into a set of salient trends that I speculate will emerge or are emerging as top ones.

Part one was the expansion of the spectrum of RAG.

AI Agents: The Next Frontier

In the Anatomy of Agentic AI we saw how agents are intelligent systems designed to autonomously perform tasks or assist users across a wide variety of domains. Unlike traditional AI models that focus on single tasks, AI agents are multi-modal, multi-agent systems capable of interacting with data, users, and each other in more sophisticated ways.

In 2025, we can expect the rapid increase in the following Agentic trends.

1. Autonomy: AI agents will handle tasks with minimal human intervention, from scheduling and research to decision-making in complex scenarios.

2. Personalization: These agents will adapt to individual users’ preferences and contexts, offering tailored solutions and enhancing user experiences.

3. Collaboration: Multi-agent systems will enable diverse agents to work together to tackle complex problems, akin to how humans collaborate.

Key Drivers of AI Agents’ Growth

1. Advances in LMs: Large Language Models like Gemini have become foundational for enabling natural language understanding and generation. Small language models will continue to thrive and get to the edge.

2. Integration with Knowledge Graphs and Ontologies: Agents are now able to ground responses in factual, structured knowledge, improving reliability.

3. Agentic Ecosystems: Companies are investing in creating frameworks for multi-agent orchestration, such as conversational agents, research assistants, and recommendation systems.

4. Vertical-Specific Applications: AI agents are being specialized for industries like healthcare, finance, and education.

Examples of AI Agents in Action

• Research Assistants: Agents that autonomously gather and summarize information from multiple sources.

• Customer Support: Conversational agents resolving queries across various platforms in real time.

• Financial Advising: AI systems monitoring market trends and optimizing portfolio management.

• Education: Personalized tutors offering lessons, assessments, and real-time feedback.

Challenges and Considerations

1. Ethics and Privacy: The deployment of autonomous agents raises concerns about data misuse and decision-making transparency.

2. Reliability: Ensuring agents are factually grounded and avoid biases is critical.

3. Interoperability: Multi-agent systems must effectively communicate and collaborate in a unified manner.

Conclusion

The “Year of AI Agents” signals an evolutionary paradigm shift where AI evolves from a tool into an active participant in our lives, reshaping workflows, business models, and personal interactions.

At the same time, it highlights a significant leap forward in both technology and adoption, by promising greater productivity, accessibility, and innovation in integrating traditional business processes and workflows with AI-driven solutions. Instead of the now more commonplace but previously magical response, the llm at the heat or really the brain of the agents will get work done .

But to do so we need to allow trust and empower them — the question is — will they be able to earn our trust?

--

--

Ali Arsanjani
Ali Arsanjani

Written by Ali Arsanjani

Director Google, AI | EX: WW Tech Leader, Chief Principal AI/ML Solution Architect, AWS | IBM Distinguished Engineer and CTO Analytics & ML

No responses yet