吳恩達 Agentic AI 第一講 Welcome!
- URL: https://www.youtube.com/watch?v=XD9DbNrrggw
- Channel: 扣子說AI
- Fetched: 2026-08-12T16:31:08+00:00
- Language: English (auto-generated)
Transcript
[00:01] Welcome to this course on agentic AI.
[00:03] When I coined the term agentic to
[00:05] describe what I saw as a important and
[00:08] rapidly growing trend in how people were
[00:09] building own base applications, what I
[00:12] did not realize was that a bunch of
[00:13] marketers would get hold of this term
[00:16] and use it as a sticker and put this on
[00:18] almost everything in sight. And that has
[00:20] caused hype on agentic AI to skyrocket.
[00:24] The good news though is that ignoring
[00:26] the hype, the number of truly valuable
[00:29] and useful applications built using
[00:31] agentic AI has also grown very rapidly,
[00:33] even if not quite as rapidly as the
[00:36] hype. And in this course, what I’d like
[00:38] to do is show you best practices for
[00:40] building agentic AI applications. And
[00:43] this will open up a lot of new
[00:44] opportunities to you in terms of what
[00:47] you can now build. Today, agentic
[00:48] workflows are being used to build
[00:50] applications like customer support
[00:52] agents or to do deep research to help
[00:55] write deeply insightful research reports
[00:58] or to process tricky legal documents or
[01:00] to look at patient input and render or
[01:03] to suggest possible medical diagnoses.
[01:06] On many of my teams, a lot of the
[01:08] projects we build just would be
[01:10] impossible without agentic workflows.
[01:12] And so, knowing how to build
[01:14] applications with them is one of the
[01:16] most important and valuable skills in AI
[01:18] today. It turns out that one of the
[01:21] biggest differences I’ve seen between
[01:23] people that really know how to build
[01:24] agentic workflows compared to people
[01:26] that are less effective at it is the
[01:28] ability to drive a disciplined
[01:29] development process, specifically one
[01:32] focused on evals and error analysis. And
[01:34] in this course, I’ll tell you what that
[01:36] means and show you what allows you to be
[01:38] really good at building these agentic
[01:40] workflows.
[01:42] Being able to do this is one of the most
[01:44] important skills in AI today and will
[01:46] open up a lot more opportunities, be it
[01:48] job opportunities or opportunities to
[01:50] just build amazing software yourself.
[01:53] With that, let’s go on to the next video
[01:55] to dive more into what are agentic
[01:58] workflows.