Agentic AI MOOC (UC Berkeley, Fall 2025)
by UC Berkeley RDI
Berkeley's graduate-level Agentic AI course — LLM reasoning, planning, and multi-agent systems from the field's leaders.
Overview
The Fall 2025 Agentic AI MOOC (UC Berkeley RDI, CS294/194-280 lineage) is the newest installment of Dawn Song's LLM-agents course series, taught with co-instructor Xinyun Chen and a roster of guest lecturers from OpenAI, Google DeepMind, NVIDIA, Microsoft, Stanford, and Sony AI (including Noam Brown, Oriol Vinyals, and Yann Dubois). Over a semester of Monday lectures — all recorded and posted to YouTube — it covers the foundations of LLMs, reasoning and planning, agentic frameworks and infrastructure, agent evaluation, multi-agent systems, AI safety and security, and representative applications spanning code generation, robotics, web automation, and scientific discovery. It builds on the Fall 2024 LLM Agents and Spring 2025 Advanced LLM Agents MOOCs and is free and open to a 32K+ global learner community.
At a Glance
- Topic
- Agentic
- Level
- Advanced
- Format
- Course
- Cost
- Free
- Duration
- ~12 weekly 2-hour lectures, self-paced video
- Provider
- UC Berkeley RDI
- Hands-on
- No
- Certificate
- None
What You’ll Learn
- ✓How modern LLM agents are designed: model, tools, and orchestration
- ✓Reasoning, planning, and inference-time techniques that make agents reliable
- ✓Agentic frameworks and infrastructure for building real systems
- ✓How to evaluate agents and reason about multi-agent coordination
- ✓Applications of agents in code generation, robotics, web automation, and science
- ✓Current thinking on AI agent safety and security from the field's researchers
Highlights
- •Taught by Prof. Dawn Song with guest lectures from OpenAI, Google DeepMind, NVIDIA, and Microsoft
- •Newest (Fall 2025) edition of Berkeley's acclaimed LLM-agents MOOC series
- •All lectures recorded and freely available on YouTube
- •Research-grade coverage of reasoning, multi-agent systems, and agent safety
Who It’s For
Best For
- ✓Experienced AI engineers who want a rigorous, current grounding in agent design
- ✓Researchers and grad students working on LLM agents
- ✓Builders who want to learn agent evaluation and multi-agent patterns from experts
Prerequisites
- •Solid machine learning / deep learning background
- •Familiarity with LLMs and Python
- •Comfort with research-level material
FAQ
What is Agentic AI MOOC (UC Berkeley, Fall 2025)?
A free, open UC Berkeley MOOC on Agentic AI taught by Prof. Dawn Song (with Xinyun Chen), covering LLM foundations, reasoning, planning, agent frameworks, and real-world agent applications. Aimed at engineers and researchers who want a rigorous, up-to-date grounding in how autonomous LLM agents are built and evaluated.
Is Agentic AI MOOC (UC Berkeley, Fall 2025) free?
Agentic AI MOOC (UC Berkeley, Fall 2025) is free to access.
What level is Agentic AI MOOC (UC Berkeley, Fall 2025) for?
Agentic AI MOOC (UC Berkeley, Fall 2025) is aimed at a advanced audience. Recommended background: Solid machine learning / deep learning background, Familiarity with LLMs and Python, Comfort with research-level material.
How long does Agentic AI MOOC (UC Berkeley, Fall 2025) take?
Expect roughly ~12 weekly 2-hour lectures, self-paced video. Most learners work through it at their own pace.
What will I learn from Agentic AI MOOC (UC Berkeley, Fall 2025)?
You'll learn: How modern LLM agents are designed: model, tools, and orchestration; Reasoning, planning, and inference-time techniques that make agents reliable; Agentic frameworks and infrastructure for building real systems; How to evaluate agents and reason about multi-agent coordination; Applications of agents in code generation, robotics, web automation, and science; Current thinking on AI agent safety and security from the field's researchers.