On July 13, 14, and 16, we had the pleasure of training over 80 faculty members and researchers from the University of Granada in one of the most promising fields of artificial intelligence: agentic and multi-agent systems. Over three intensive online sessions, we created a space to bridge academic knowledge with the real-world experience of those already building AI solutions in professional environments.

The Challenge: Bringing Agentic AI to the Classroom

Generative artificial intelligence has democratized access to capabilities that seemed unattainable just two years ago. But there's a significant gap between using ChatGPT for specific tasks and designing autonomous agent architectures capable of solving complex problems in a coordinated manner.

Faculty members at the University of Granada have already taken the first step: understanding what AI is and how to integrate it into their educational and research work. Now, the natural next leap is understanding how to build systems that don't just respond, but act, reason, and collaborate.

What is Agentic AI?

An AI agent isn't a chatbot waiting for instructions. It's a system capable of:

• Planning actions to achieve complex goals
• Using tools (APIs, databases, code) autonomously
• Reasoning about multi-step problems
• Collaborating with other agents in multi-agent architectures
• Adapting to changing contexts

In simple terms: we move from "question → answer" to "goal → planning → execution → validation".

The Program: Three Sessions, One Practical Vision

Throughout the three program sessions, we covered everything from fundamentals to practical implementation of agentic systems:

Session 1: Agentic AI Fundamentals

• Evolution: from language models to autonomous agents
• Agent anatomy: memory, planning, tools
• Design patterns: ReAct, Chain-of-Thought, Tool Use
• Real-world use cases in research and teaching

Session 2: Multi-Agent Architectures

• Coordination between specialized agents
• Collaboration patterns: hierarchical, autonomous, hybrid
• Framework landscape: LangChain, LangGraph, CrewAI, AutoGen
• Designing conversational and agentic flows
• Integrating external tools (search, code, APIs)
• Memory and context management

Session 3: Agentic AI in Production

• Agent harnesses and frameworks
• Observability and monitoring
• Security in Agentic AI
• Applications in academic environments: research assistants, adaptive tutoring
• Ethics and governance in autonomous systems
• What's coming and where to advance

Open Resources for the Community

All training materials are available as open source in our GitHub repository: montevive/agentic-ai-course

The repository includes:

• Interactive notebooks with executable examples
• Reference code to implement your own agents
• Use cases documented step-by-step
• References to papers, frameworks, and additional resources

This material isn't just for UGR faculty: any professional or researcher interested in agentic AI can leverage these resources to learn and experiment.

The Value of Connecting University and Business

This type of collaboration is fundamental to Granada's ecosystem. The AI Granada Foundation is doing exceptional work as a bridge between the cutting-edge research generated at the University of Granada and the real needs of the productive sector.

We especially want to thank Alberto Fernández Hilario, who served as the link between participating faculty and the Foundation's Training department, coordinating, facilitating, and making it possible for everything to come together.

Why Does This Matter?

Granada has all the ingredients to become a reference hub in applied AI:

1. Top-tier academic talent: UGR is a leader in AI, computing, and data science research
2. Growing business ecosystem: More and more startups and companies are betting on technological innovation
3. Knowledge transfer: Initiatives like this turn research into applied capacity

When we connect these three elements, we don't just generate training: we create collaboration, shared learning, and new opportunities for the entire ecosystem.

From AI Consumers to AI Builders

One of the key messages we conveyed during the course is this: we can't remain AI consumers; we must become AI builders.

The difference is profound:

• Consumer: Uses closed tools, depends on third parties, limited by what others build
• Builder: Designs custom solutions, controls privacy and data, adapts AI to specific needs

Agentic systems are one of the frontiers where this transition is most relevant. This isn't future technology: companies and research teams are already using agents in production for tasks like code analysis, customer support, massive data analysis, or complex workflow automation.

Lessons Learned

After three intensive sessions with such a diverse audience (from computer engineers to humanities experts), some reflections:

1. The Demand is Real

The level of participation and engagement during the sessions confirms that the academic community is ready not just to use AI, but to build with AI.

2. The Gap Isn't Technical, It's Conceptual

The main obstacle isn't learning to program (though it helps), but changing the mental model: moving from "question-answer" to "agent architecture".

3. Use Cases Matter

Abstract examples are fine, but what really connects with participants are concrete applications to their daily work: generating training datasets, analyzing papers at scale, creating adaptive tutors, automating code review...

4. Open Source is Key

The availability of frameworks like LangChain, AutoGen, or CrewAI has democratized agent development. You no longer need a Google research team: anyone can build sophisticated systems with open tools.

Next Steps

This training is just the beginning. We hope participants will:

1. Experiment with the repository materials
2. Integrate agentic AI concepts into their research or teaching projects
3. Share their experiences and use cases with the community
4. Collaborate with companies and startups to apply this knowledge

From Montevive, we'll continue contributing to the local ecosystem with training, open source code, and projects that demonstrate that pragmatic, secure, and applicable AI is possible.

Join Granada's AI Ecosystem

If you're a faculty member, researcher, or professional interested in learning more about agentic AI:

• Explore the repository: github.com/montevive/agentic-ai-course
• Follow us on LinkedIn: To stay updated on new training and events
• Contact AI Granada Foundation: If you want to organize similar training in your organization

Are you building AI agents at your university or company? We'd love to hear about your experience and explore potential collaborations.