[00:00] (0.48s)
All right, so let's just do a quick
[00:01] (1.76s)
review. I'm doing an agent right now for
[00:03] (3.84s)
product research and you can see we're
[00:05] (5.60s)
running into some errors. So let's do a
[00:07] (7.12s)
practical exercise on this, right? So if
[00:09] (9.68s)
you're doing any coding related work,
[00:11] (11.28s)
you can go ahead and if you're in
[00:13] (13.04s)
cursor, cursor is an integrated
[00:15] (15.60s)
development environment that you can do
[00:17] (17.20s)
for coding and you don't need to
[00:18] (18.48s)
actually know code. You just need to be
[00:19] (19.84s)
able to tell it what to do. So one of
[00:21] (21.76s)
the benefits of it is you can go through
[00:23] (23.44s)
and you you can take the error in this
[00:25] (25.28s)
scenario that I'm looking to build and
[00:27] (27.20s)
you can click add to chat. It's going to
[00:29] (29.28s)
add the lines where your error is. And
[00:32] (32.16s)
then what we can do is we can select the
[00:33] (33.84s)
model that we want to iterate with. So
[00:36] (36.00s)
for this one, I'm going to choose 03
[00:37] (37.60s)
because it's usually really good at
[00:39] (39.28s)
complex problems. So I'm going to use
[00:41] (41.52s)
that for right now. Now this is going to
[00:43] (43.44s)
change depending on when you're looking
[00:44] (44.64s)
at this recording. Now what we're going
[00:46] (46.48s)
to do is we already have the error
[00:47] (47.92s)
message there. Now I could type this,
[00:50] (50.08s)
but what I've noticed is there's a
[00:51] (51.60s)
really cool program called Whisper Flow.
[00:53] (53.68s)
Essentially, what it's going to do is
[00:54] (54.88s)
allow me to just talk and it'll it'll
[00:57] (57.04s)
take it into transcripts. So, this is
[00:59] (59.12s)
something that I've started to use a lot
[01:00] (60.72s)
more because I can talk a lot quicker
[01:02] (62.40s)
than I can type. Okay. What I'd like you
[01:05] (65.12s)
to do is review the error message that
[01:06] (66.96s)
you've just received and try to debug
[01:09] (69.20s)
it. Use reverse chain of thought
[01:11] (71.12s)
reasoning to be able to think through
[01:12] (72.72s)
the context of the problem. Please use
[01:15] (75.04s)
the full code base and any relevant
[01:16] (76.88s)
files that you find that are necessary
[01:19] (79.28s)
to help resolve this issue. If you're
[01:21] (81.84s)
unsure, use any associated tools that
[01:24] (84.32s)
are at your disposal, whether that's the
[01:26] (86.32s)
fetch MCP model context protocol server
[01:29] (89.20s)
or any others that you've been already
[01:30] (90.72s)
pre-programmed to leverage. Once you
[01:32] (92.80s)
have a plan of action and have thought
[01:34] (94.72s)
through this, then go ahead and proceed
[01:36] (96.32s)
with implementation.
[01:39] (99.20s)
Now, what's going to happen is I'll go
[01:40] (100.56s)
ahead and kick this off. You can see
[01:41] (101.92s)
it's transcribed what I just said from
[01:43] (103.76s)
audio into text and it's going to do its
[01:46] (106.00s)
thing. So, just wanted to give a quick
[01:48] (108.00s)
little overview of this is the process
[01:49] (109.68s)
steps that you can use to be able to
[01:51] (111.60s)
start doing what is known as vibe
[01:53] (113.20s)
coding, but essentially being able to
[01:55] (115.12s)
use AI technologies to be able to
[01:57] (117.12s)
achieve any type of task that you're
[01:58] (118.56s)
working