Techsplainers by IBM

This episode of Techsplainers explores the concept of tool calling in artificial intelligence, explaining how it enables AI models to interact with external tools, APIs, and systems beyond their native capabilities. We walk through how tool calling works, from recognizing when external assistance is needed to selecting appropriate tools and processing responses. The episode highlights the powerful combination of tool calling with retrieval augmented generation (RAG) and examines real-world applications in information retrieval, code execution, process automation, IoT device control, and personalized recommendations. By bridging the gap between AI reasoning and action, tool calling is transforming passive AI assistants into proactive digital agents capable of completing complex, multi-step tasks through dynamic access to external resources.

Find more information at https://www.ibm.com/think/podcasts/techsplainers.

Narrated by Selma Pacheco Jimenez

This episode of Techsplainers explores the crucial components of AI agent memory and agentic reasoning. We delve into how AI agents store and recall information through different memory types—including short-term, long-term, episodic, semantic, and procedural memory—and how frameworks like LangChain and LangGraph implement these capabilities. The episode also examines various reasoning paradigms that power AI decision-making, from simple conditional logic to sophisticated approaches like ReAct, ReWOO, and multiagent reasoning. By understanding these complementary components, listeners gain insight into how modern AI systems transform from passive models into intelligent agents that can maintain context across interactions, learn from past experiences, and make autonomous decisions to achieve complex goals.

Find more information at https://www.ibm.com/think/podcasts/techsplainers.

Narrated by Selma Pacheco Jimenez

This episode of Techsplainers explores two fundamental capabilities of AI agents: perception and planning. We examine how agents perceive their environment through visual, auditory, textual, environmental, and predictive means, breaking down the four-step perception process from sensory input collection to decision-making. The discussion then shifts to how agents use this perceived information to plan their actions, covering goal definition, state representation, action sequencing, and optimization techniques like heuristic search and reinforcement learning. We also explore how different planning frameworks operate and how planning becomes more complex in multi-agent systems where coordination is essential. By understanding these interconnected components, listeners gain insight into what makes AI agents truly intelligent and capable of operating autonomously in complex environments.

Find more information at https://www.ibm.com/think/podcasts/techsplainers.

Narrated by Selma Pacheco Jimenez

This episode of Techsplainers explores the essential components that make AI agents function, breaking down the "brain" of these intelligent systems. We examine how perception enables agents to understand their environment through various inputs, while planning allows them to map out complex task sequences. The discussion covers memory systems that provide both short-term context and long-term learning, reasoning modules that power decision-making, and action capabilities that execute tasks through tool calling. We also investigate how communication facilitates interaction with humans and other agents and how learning capabilities enable continuous improvement over time. By understanding these interconnected components, listeners gain insight into how AI agents operate across various industries and applications.

Find more information at https://www.ibm.com/think/podcasts/techsplainers.

Narrated by Selma Pacheco Jimenez

This episode of Techsplainers explores reinforcement learning, a machine learning approach where AI agents learn to make decisions through trial and error by interacting with their environment. Unlike supervised learning's labeled data or unsupervised learning's pattern discovery, reinforcement learning teaches through reward signals—similar to how we might train a pet with treats. The episode breaks down the core components of this approach, including the Markov decision process framework, the critical exploration-exploitation tradeoff, and key elements like policy, reward signals, and value functions. We also examine major reinforcement learning methods, such as dynamic programming, Monte Carlo techniques, and temporal difference learning. The discussion covers real-world applications in robotics and natural language processing, highlighting both impressive successes like AlphaGo and ongoing challenges in creating effective learning environments with meaningful reward systems.

Find more information at https://www.ibm.com/think/podcasts/techsplainers.

Narrated by Anna Gutowska