[00:00] (0.16s)
I have a number of group chats with my
[00:02] (2.56s)
friends who are developers and we were
[00:04] (4.96s)
all like sharing our CC usage scorecards
[00:08] (8.08s)
as like bragging rights in the chat and
[00:10] (10.96s)
I was like in one chat I was the winner
[00:13] (13.52s)
and then another chat my friend was like
[00:15] (15.44s)
oh that's cute and showed me like $4,000
[00:18] (18.08s)
in usage. It's this weird like scorecard
[00:20] (20.88s)
of like nerd cred of how much are you
[00:23] (23.04s)
using this? How much are you building? I
[00:25] (25.20s)
saw a tweet from somebody and he said
[00:28] (28.16s)
like CC usage screenshot is the next
[00:31] (31.52s)
generation of GitHub commit graph. So I
[00:36] (36.00s)
I was so I was so happy to see that.
[00:38] (38.64s)
How do you pronounce your name?
[00:40] (40.48s)
Uh Neop.
[00:43] (43.44s)
We can just jump right into it. I'm
[00:45] (45.04s)
going to run my CC usage and you can
[00:48] (48.32s)
tell me what I'm seeing here.
[00:50] (50.96s)
Oh, you installed the CC usage. All
[00:54] (54.08s)
right. Yes, I did. And I noticed maybe
[00:55] (55.92s)
you could start there. Should someone
[00:57] (57.52s)
install it or should they just run it uh
[00:59] (59.76s)
remotely?
[01:00] (60.32s)
I recommend to use a bunx. It's very
[01:04] (64.08s)
The recommended usage is bunx space CC
[01:07] (67.92s)
Yes, because because uh so I I release
[01:11] (71.20s)
CC usage quite often. So you can run BX
[01:15] (75.12s)
CC usage to face the latest one.
[01:17] (77.76s)
So here's the report I get. So tell me
[01:19] (79.92s)
what I'm seeing here. So this table is
[01:24] (84.24s)
all about the cost. It's kind of like
[01:27] (87.36s)
pay as you go plan or if you are in the
[01:30] (90.24s)
max brand or prop
[01:32] (92.88s)
how much how much cost you save.
[01:35] (95.28s)
Everyone likes talking about money or
[01:38] (98.16s)
everyone is talking about the bargain or
[01:40] (100.32s)
something like that.
[01:41] (101.44s)
For context on my end, I am on the $200
[01:44] (104.64s)
a month Cloud Max plan. Yeah.
[01:46] (106.72s)
Um I was on the $100 a month max uh
[01:49] (109.92s)
plan. And the reason I upgraded to 200
[01:52] (112.80s)
was primarily for Opus 4 usage.
[01:56] (116.56s)
So what we're seeing is probably my last
[01:59] (119.68s)
3 weeks of usage and I've gotten about
[02:03] (123.84s)
worth of tokens out of this. So um what
[02:07] (127.28s)
I noticed when I very first started
[02:08] (128.88s)
using Cloud Code and what I think
[02:10] (130.40s)
everybody noticed is how expensive it
[02:12] (132.32s)
is. And I felt like I could rack up $40,
[02:15] (135.12s)
$50 in a few hours. Um, and it was the
[02:19] (139.28s)
it created this real anxiety for me of
[02:22] (142.64s)
like, oh, is do I really want to use
[02:24] (144.56s)
this and and do I how do I optimize it?
[02:27] (147.76s)
And then when they rolled out the Cloud
[02:29] (149.36s)
Max plans, it felt sort of like
[02:31] (151.92s)
Christmas because now you could just use
[02:34] (154.00s)
Cloud Code for all the things. And then
[02:36] (156.32s)
I was shocked when I first used CC usage
[02:39] (159.68s)
to see how much I would have spent in
[02:41] (161.76s)
tokens if I had been on metered
[02:43] (163.68s)
building. I'm sure you're getting tagged
[02:45] (165.28s)
on everybody's screenshots. Like what is
[02:47] (167.44s)
the biggest what's the biggest number
[02:49] (169.20s)
you've seen down here so far?
[02:51] (171.04s)
I already saw like 8.7K
[02:56] (176.48s)
per month.
[02:58] (178.80s)
And he he's using CL in parallel every
[03:02] (182.72s)
day and he only uses Opus. What are are
[03:05] (185.84s)
some of the hidden features or or just
[03:07] (187.52s)
the features that people might not
[03:08] (188.96s)
necessarily know about?
[03:10] (190.72s)
Okay, so okay, you can hear the B XC
[03:13] (193.76s)
usage does help.
[03:20] (200.00s)
Got it.
[03:21] (201.20s)
Yeah, you know there are a bunch of
[03:23] (203.28s)
features.
[03:24] (204.24s)
Awesome, right? Yeah. Yeah.
[03:25] (205.84s)
Walk me through some of these.
[03:27] (207.76s)
So when you hit CC usage, so the default
[03:31] (211.84s)
command is daily, right? Right. So,
[03:35] (215.28s)
let's go through the monthly.
[03:38] (218.08s)
One. Okay. Yeah. So, you can you can
[03:40] (220.24s)
type it monthly.
[03:42] (222.88s)
And so, I'm going to do CC usage
[03:44] (224.88s)
monthly.
[03:46] (226.32s)
Like that. Got it.
[03:49] (229.92s)
So, you can see how much tokens you used
[03:52] (232.80s)
in the last month and this month, right?
[03:55] (235.52s)
Okay. Yeah. Yeah. So, you you you
[03:58] (238.40s)
used a lot.
[03:59] (239.60s)
It's all relative, I think. Where are
[04:01] (241.36s)
those logs that you're reading? I'm
[04:03] (243.68s)
guessing this is in my like claw
[04:05] (245.36s)
directory in my home folder perhaps.
[04:07] (247.60s)
So you can go to project directory
[04:10] (250.72s)
projects. Yeah.
[04:12] (252.00s)
Yep. Yep.
[04:13] (253.44s)
So there are bunch of directories and
[04:16] (256.96s)
you can choose your uh look at that
[04:20] (260.72s)
favorite project and you can see some
[04:23] (263.20s)
JSON L.
[04:25] (265.04s)
Yep. Yep. Yep. Oh, look at that. Oh, so
[04:28] (268.88s)
cool. So it has the JSON L and that has
[04:32] (272.64s)
everything that it does. And then this
[04:34] (274.64s)
is also I I suppose when I run claw-res,
[04:39] (279.68s)
this is what it's reloading, right?
[04:41] (281.84s)
Yeah. Yeah. Yeah, definitely.
[04:43] (283.68s)
Oh yeah. You can see here cache creation
[04:45] (285.36s)
input tokens
[04:46] (286.88s)
24,000 input tokens for Oh, super super
[04:50] (290.64s)
cool. Okay. Request ID and then you're
[04:52] (292.64s)
going to when you get the response back,
[04:54] (294.00s)
you're going to see the uh the tokens as
[04:55] (295.76s)
well. That's super interesting.
[04:57] (297.36s)
Yeah. Let's try the session one.
[04:59] (299.68s)
Uh-huh.
[05:02] (302.64s)
Oh, interesting.
[05:05] (305.36s)
Yeah. So, so, so you can see how many
[05:09] (309.36s)
you how many tokens you used in the
[05:12] (312.48s)
specific projects.
[05:14] (314.80s)
So, the main project I've been working
[05:16] (316.48s)
on uh is this thing called YouTube track
[05:19] (319.12s)
tracker. But I this last week I was
[05:21] (321.92s)
experimenting a lot with work trees and
[05:24] (324.56s)
trying to run cloud code in parallel.
[05:27] (327.44s)
And so a lot of what we're seeing here
[05:29] (329.92s)
are um branches for specific issues and
[05:34] (334.16s)
the the work trees where I was trying to
[05:36] (336.40s)
to have cloud code work in parallel.
[05:38] (338.96s)
Interesting thing is that the cloud code
[05:41] (341.36s)
uses cash prompt so much, right? Yes. So
[05:45] (345.36s)
there are lots of cash tokens. What
[05:48] (348.00s)
would be an example of the types of
[05:49] (349.28s)
tokens that you're you suspect they're
[05:51] (351.04s)
caching?
[05:52] (352.08s)
I don't I I don't know everybody know
[05:55] (355.60s)
about cash token prompt. Yeah, cash
[05:58] (358.80s)
talker is a feature of code when you use
[06:02] (362.24s)
clo codes. So there are long
[06:04] (364.80s)
conversations and uh and the previous
[06:08] (368.08s)
conversation should be like duplicate of
[06:11] (371.52s)
the last request, right? So they they
[06:15] (375.28s)
see that okay we received the uh
[06:17] (377.92s)
conversation before so we see this
[06:21] (381.60s)
conversation as a cached one cash
[06:24] (384.32s)
prompting is really tricky when content
[06:27] (387.44s)
changed a little bit it's not cached one
[06:30] (390.24s)
anymore I think uh they optimize how to
[06:33] (393.92s)
use the cache prompting
[06:36] (396.56s)
in the middle of June and also before
[06:39] (399.20s)
the sonet 4 and opus 4 the uh length of
[06:42] (402.96s)
The save time is like five minutes.
[06:46] (406.40s)
After the phone, they save the cash for
[06:50] (410.32s)
an hour. So this is another way for them
[06:53] (413.92s)
to like optimize the cash use.
[06:56] (416.16s)
It was just astounding to me that like
[06:58] (418.16s)
input here for instance, let's just call
[07:00] (420.08s)
it 50,000. Output tokens, let's just say
[07:02] (422.72s)
it was 10 times that amount. Let's just
[07:04] (424.56s)
say it was 500,000.
[07:06] (426.56s)
And but then the the cash tokens
[07:12] (432.24s)
uh I'm sorry the total yeah the total
[07:14] (434.96s)
cash read here or total tokens let's say
[07:17] (437.28s)
total tokens here are 500 million. So so
[07:22] (442.16s)
you're you're talking like the the
[07:24] (444.64s)
output tokens are 1 1,000th of the total
[07:29] (449.92s)
tokens. And so they're that basically
[07:32] (452.72s)
like the cash is representing if I'm
[07:34] (454.80s)
reading this right the cash is
[07:36] (456.00s)
representing like 99.9%
[07:39] (459.60s)
the tokens that like that's pretty
[07:41] (461.36s)
incredible.
[07:42] (462.88s)
I think I think they reuse the code base
[07:46] (466.24s)
as a cash token. I think
[07:48] (468.96s)
Oh, interesting.
[07:49] (469.84s)
I think at the end of May, so the
[07:52] (472.48s)
algorithm is different. They send the
[07:55] (475.52s)
they send the prompt first and they then
[07:58] (478.64s)
send they send the code base I think but
[08:01] (481.52s)
they change the they optimize the
[08:03] (483.44s)
algorithm first they send all code base
[08:06] (486.88s)
and then add add the prompt at the end.
[08:10] (490.80s)
So, so it increase the cost cache
[08:14] (494.40s)
efficiency, right?
[08:15] (495.60s)
That's interesting too about the code
[08:17] (497.04s)
base because I think a lot of the
[08:21] (501.84s)
Were you using cursor before you were
[08:24] (504.16s)
using cloud code so much?
[08:25] (505.76s)
No, no,
[08:26] (506.72s)
no. You were never on cursor.
[08:28] (508.64s)
I'm a big actually.
[08:31] (511.36s)
Yeah. Yeah. Amazing. Well, I I think
[08:34] (514.48s)
that a lot of folks were wondering when
[08:36] (516.56s)
cloud code came out and and to today why
[08:39] (519.60s)
it feels like you're getting such better
[08:42] (522.48s)
code generation from cloud code than you
[08:46] (526.08s)
do from cursor when even if you're using
[08:48] (528.80s)
the cloud 4 models in cursor and I
[08:52] (532.32s)
suspect that this huge amount of context
[08:56] (536.24s)
and the caching
[08:58] (538.16s)
has a lot to do with that, right? Like I
[09:00] (540.32s)
just suspect that when you're working
[09:01] (541.60s)
with claude code, enthropic is passing
[09:04] (544.24s)
in a whole lot more context on every
[09:07] (547.28s)
inference than what cursor is doing
[09:10] (550.16s)
given the fact that you're paying cursor
[09:11] (551.60s)
$20 a month and given the fact that you
[09:14] (554.88s)
know for me at least I've I've spent
[09:17] (557.44s)
$1,200 worth of tokens on uh uh on cloud
[09:20] (560.80s)
code over the last few weeks. So
[09:23] (563.36s)
and and another and also think this is
[09:25] (565.68s)
my guess but I think I think they have
[09:28] (568.64s)
their own document
[09:31] (571.28s)
uh code base in their server. So before
[09:35] (575.12s)
before we cannot generate fives. Yeah.
[09:39] (579.20s)
So but we cannot deal with it but so
[09:42] (582.56s)
right now we can deal with it and we can
[09:45] (585.04s)
see there are lots of cash tokens. So I
[09:47] (587.76s)
I think so this is my guess but they
[09:50] (590.88s)
they they sent the documentation to the
[09:53] (593.92s)
cloud code.
[09:54] (594.88s)
It'd be cool to get answers from them
[09:57] (597.04s)
someday on that. But uh it's just really
[09:58] (598.96s)
fascinating how this view over the logs
[10:02] (602.40s)
gives you a little more insight into
[10:04] (604.80s)
some of the magic that's happening that
[10:07] (607.28s)
makes cloud code feel so magical. Uh
[10:10] (610.00s)
another command blocks. Blocks. Got it.
[10:13] (613.68s)
Blocks.
[10:17] (617.12s)
So yeah, this is a really cool feature.
[10:19] (619.76s)
So one of our contributors
[10:22] (622.80s)
worked on this. When you start using
[10:25] (625.44s)
cloud code, so there's like there's a
[10:28] (628.32s)
time range in five hours. And when you
[10:32] (632.32s)
so when you consume a lot of tokens in
[10:36] (636.64s)
the five hours, you cannot use clo
[10:39] (639.52s)
anymore. This view can show you how the
[10:43] (643.76s)
five block five hours blocks works.
[10:46] (646.88s)
I like that.
[10:49] (649.44s)
that's really cool.
[10:50] (650.72s)
Yeah. Yeah. So you can you can see Yeah.
[10:52] (652.56s)
Tell me your thoughts on the table. I
[10:55] (655.20s)
guess I have not since bumping up to the
[10:57] (657.52s)
$200 a month plan. I have not hit that a
[11:00] (660.72s)
lot. But I do have a couple friends who
[11:03] (663.04s)
were trying to see how much they how
[11:05] (665.20s)
much usage they could get out of either
[11:07] (667.44s)
the $100 or the $20 a month plans. And
[11:10] (670.48s)
this seems like this view would be very
[11:12] (672.16s)
very useful to them.
[11:13] (673.76s)
Yeah. Oh, you use $76
[11:16] (676.88s)
per block, right?
[11:20] (680.64s)
Yeah. 75. Yeah. Yeah. Yeah.
[11:25] (685.20s)
That's so Yeah. So 7. So right here. So
[11:28] (688.56s)
on July 1st,
[11:31] (691.28s)
I spent $75
[11:33] (693.76s)
in worth of tokens in five hours. So
[11:36] (696.88s)
that's really And then yeah, if we go
[11:39] (699.44s)
and the next block, I spent another $33.
[11:43] (703.52s)
That's so interesting.
[11:46] (706.24s)
I love that. What a great view. And then
[11:48] (708.24s)
we can see actually this one here that I
[11:49] (709.68s)
was working on right before we hopped on
[11:52] (712.00s)
the call. So it's saying that block is
[11:53] (713.84s)
still usage still still open right now.
[11:56] (716.56s)
Yeah. So I think I think you're in
[11:59] (719.28s)
active block session active block. So
[12:02] (722.88s)
you can you can use the last command.
[12:06] (726.24s)
This is really magical. So so you
[12:09] (729.52s)
last one is not individual sub command.
[12:12] (732.64s)
So you can hit BX.
[12:16] (736.56s)
CC usage uh blocks
[12:21] (741.36s)
blocks dash dash live
[12:25] (745.68s)
Yeah. There we go.
[12:31] (751.28s)
oh, nice.
[12:35] (755.60s)
it's awesome.
[12:37] (757.04s)
So, this is a live view and yeah,
[12:39] (759.52s)
actually it's a buggy. So, I'm I'm
[12:41] (761.52s)
fixing a lot actually, but you can see
[12:45] (765.04s)
how many session you have or and also
[12:49] (769.36s)
you can see the burn rate of tokens per
[12:53] (773.20s)
blocks. So this is a really cool
[12:55] (775.20s)
feature, right?
[12:56] (776.80s)
I love that.
[12:58] (778.72s)
And yeah, you can see the projection of
[13:01] (781.52s)
the current blocks, right? For example.
[13:04] (784.64s)
So I think I think it estimate that you
[13:08] (788.00s)
you going to use $44
[13:11] (791.04s)
for this five hour session. Five hour
[13:14] (794.24s)
block. Yeah. Yeah. Yeah.
[13:15] (795.76s)
This would be so fun to have this up on
[13:17] (797.52s)
a separate monitor. It's like you have
[13:20] (800.08s)
this going on one monitor while you run
[13:21] (801.68s)
your cloud code sessions in the other
[13:23] (803.12s)
one to see how much more work you can do
[13:25] (805.04s)
over the next few hours.
[13:26] (806.48s)
Yeah. So my friend spread the terminal
[13:29] (809.20s)
screen using T-Max and so always show
[13:33] (813.12s)
the live monitor.
[13:35] (815.20s)
Are you running cloud in parallel at
[13:37] (817.36s)
Yeah, I sometimes use like use cloud
[13:40] (820.40s)
code in parallel using g work. I
[13:42] (822.88s)
recently find the tool which provides
[13:45] (825.76s)
great integration for uh git work squad
[13:50] (830.48s)
you know that
[13:51] (831.28s)
cloud squad okay I do not know that
[13:53] (833.92s)
cloud squad integrates cloud code and
[13:57] (837.92s)
t-max and g get work so it's really
[14:02] (842.08s)
useful tool I
[14:03] (843.52s)
this is the one you're talking about
[14:05] (845.36s)
yeah yeah yeah this one
[14:07] (847.12s)
oh super interesting
[14:09] (849.44s)
so so when so when we use we don't we
[14:11] (851.84s)
don't tie I get the work work tree
[14:14] (854.32s)
command anymore. Yeah.
[14:15] (855.92s)
Right. Good. Okay. Yes. Cuz I I've been
[14:19] (859.36s)
like I I wrote like a shell script to
[14:21] (861.28s)
like try to spin up the work tree and
[14:23] (863.04s)
then tear it down afterwards and then
[14:24] (864.72s)
I'm still like running into I feel like
[14:27] (867.44s)
merge conflicts. All like this sort of
[14:29] (869.84s)
UI makes a whole lot more sense.
[14:32] (872.08s)
For that.
[14:33] (873.12s)
All right. I like that a lot. Okay.
[14:35] (875.20s)
All right. I'm going to check that out
[14:36] (876.24s)
afterwards. How did you build this
[14:39] (879.36s)
in the initial version? I mean the 0.1
[14:43] (883.20s)
version I created only a couple of hours
[14:46] (886.88s)
using cloud code. I remember the event
[14:50] (890.24s)
hosted by Anthropic. I must be I I think
[14:55] (895.20s)
it's on 21st of May actually and they
[14:58] (898.72s)
announced the Sonet 4 and Opus 4 and I
[15:03] (903.04s)
was shocked the how smart the model is
[15:08] (908.56s)
but Opus 4 and Sonet 4 is a game changer
[15:13] (913.04s)
right so and Cloud's experience is
[15:16] (916.32s)
really good. So I was I was really into
[15:20] (920.00s)
it. I saw a Japanese blog how to
[15:22] (922.72s)
calculate the cost at that time. So
[15:25] (925.36s)
there is a log file of code. We have a
[15:29] (929.76s)
property called cost USD.
[15:32] (932.64s)
So at that time we can just uh sum up
[15:35] (935.84s)
the cost USD and and calculate the total
[15:39] (939.60s)
cost. It looks interesting. Why don't I
[15:42] (942.08s)
create a new tool and why don't I try
[15:45] (945.28s)
the new vibing style? So I I decided to
[15:50] (950.64s)
create a CC usage and I didn't expect it
[15:54] (954.00s)
goes seriously and it goes so viral.
[15:57] (957.20s)
I noticed at the start of this you say
[15:59] (959.52s)
that you're fetching the latest model
[16:01] (961.04s)
pricing from light LLM. At the end of
[16:03] (963.84s)
May we can calculate the cost easily by
[16:07] (967.36s)
some of the cost USD properties but also
[16:12] (972.16s)
deleted the property.
[16:14] (974.32s)
Oh really?
[16:15] (975.52s)
I was really shocked and yeah everyone
[16:18] (978.96s)
in our community was so shocked. So the
[16:21] (981.52s)
issue came up like I don't know 1.010
[16:26] (986.48s)
doesn't support cost USD anymore. We
[16:29] (989.04s)
need to do we need to find alternative
[16:31] (991.68s)
solutions. So I came up idea to like
[16:34] (994.88s)
just calculate the cost by using the to
[16:39] (999.12s)
consumed the token and the price table.
[16:43] (1003.44s)
I changed the spec in just four hours I
[16:47] (1007.44s)
think.
[16:48] (1008.64s)
Is there anything in particular that you
[16:51] (1011.04s)
are looking for help on with CC usage? I
[16:54] (1014.48s)
know you have a lot of contributors. Um
[16:57] (1017.28s)
yeah, some people ask me uh about Gemini
[17:01] (1021.84s)
CI but I don't know because Gemini CI
[17:07] (1027.68s)
provides the cost usage when you finish
[17:10] (1030.80s)
finish using Gemini. So I don't I don't
[17:13] (1033.68s)
think I created. So there are a lots of
[17:17] (1037.92s)
people uh started to creating something
[17:21] (1041.28s)
cool using CC usage, right? So like some
[17:25] (1045.60s)
so some people created the raycast
[17:28] (1048.56s)
extension. Yeah, you can see the CC
[17:30] (1050.80s)
usage result on not on the raycast
[17:33] (1053.52s)
window but also you can see on the menu
[17:35] (1055.68s)
bar, right?
[17:36] (1056.72s)
Oh, that's cool. So I would be able to
[17:38] (1058.56s)
see it right up here.
[17:40] (1060.16s)
Yeah. And also also I also another
[17:43] (1063.20s)
people is creating the UIs right. So uh
[17:48] (1068.08s)
like like dashboard or something and and
[17:51] (1071.84s)
I heard somebody started to create a
[17:54] (1074.16s)
leaderboard for
[17:55] (1075.44s)
CC. Yes. So this is this was funny. I
[17:59] (1079.20s)
noticed on the Claude AAI subreddit
[18:01] (1081.76s)
there were people posting their their CC
[18:05] (1085.44s)
usage and then you know I think I was
[18:09] (1089.36s)
surprised a little bit about how
[18:11] (1091.44s)
negative the comments some of the
[18:13] (1093.44s)
comments were but I guess it kind of
[18:14] (1094.96s)
makes sense right like it does feel like
[18:17] (1097.84s)
this $200 a month plan to get $1,200 of
[18:20] (1100.80s)
tokens is a great gift from Anthropic
[18:23] (1103.76s)
that probably can't last forever but
[18:26] (1106.08s)
almost certainly is not going to last
[18:27] (1107.76s)
forever. or if a bunch of people on the
[18:29] (1109.68s)
internet just are like, "How do I spend
[18:31] (1111.68s)
as many tokens as possible as quickly as
[18:33] (1113.84s)
possible so I can like get on top of the
[18:36] (1116.24s)
Reddit leaderboard?" Or this acts as
[18:38] (1118.72s)
this measure of conspicuous consumption
[18:42] (1122.24s)
slash nerdcred.
[18:44] (1124.32s)
But yeah, then there's also like all the
[18:46] (1126.48s)
environmental and wasteful concerns all
[18:48] (1128.32s)
wrapped up in that. And it's really
[18:49] (1129.68s)
interesting just all the feelings that
[18:52] (1132.16s)
uh that this tool evokes. I bet you were
[18:54] (1134.24s)
surprised by some of the reactions that
[18:56] (1136.08s)
folks had. So I saw another tweet and
[18:59] (1139.68s)
and he show the table and he see he saw
[19:03] (1143.36s)
the tokens and cost and he realized okay
[19:08] (1148.24s)
I worked so hard with clothing for for
[19:11] (1151.68s)
this m so I I want to cheer me up like
[19:15] (1155.12s)
that he said like that. So it was really
[19:18] (1158.16s)
heartwarming. I love that. I spent most
[19:20] (1160.80s)
of my, you know, the last couple years
[19:23] (1163.36s)
just building a lot of stuff with AI.
[19:25] (1165.36s)
And, you know, some days it feels like
[19:29] (1169.52s)
you're learning a lot and building a
[19:31] (1171.04s)
lot. Some days it feels like you spent a
[19:33] (1173.76s)
whole bunch of hours doing something and
[19:35] (1175.20s)
the code never worked. Of course, it's a
[19:37] (1177.76s)
imperfect metric, but it is really cool
[19:40] (1180.64s)
to look back and see, oh no, I really
[19:44] (1184.00s)
was spending time trying to learn how
[19:45] (1185.92s)
this thing works. There is something
[19:47] (1187.44s)
very real about that of it is proof that
[19:49] (1189.76s)
you've been using the tools and I think
[19:51] (1191.36s)
that all of us are really trying to
[19:53] (1193.28s)
figure out like what do these things
[19:54] (1194.48s)
mean and how do I build software better
[19:56] (1196.56s)
because of the new tools that we have
[19:57] (1197.84s)
today.
[19:58] (1198.88s)
Yeah. Yeah. Yeah. I think so. I think
[20:00] (1200.96s)
so. So since usage is not not only the
[20:04] (1204.00s)
font to but also they
[20:06] (1206.88s)
enhance enhance the pipe coders or the
[20:10] (1210.24s)
developers. I'm really confident to say
[20:12] (1212.80s)
that I I I'm really happy to involve
[20:17] (1217.52s)
this awesome tour and also and yeah so
[20:21] (1221.20s)
this is really important things so I'm
[20:23] (1223.20s)
I'm taking a lot lots of time if you're
[20:26] (1226.88s)
happy about my tours uh please sponsor
[20:30] (1230.40s)
me and uh if you sponsor me so the all
[20:34] (1234.48s)
income is not only for me but also uh my
[20:38] (1238.16s)
contributors
[20:39] (1239.52s)
I think you are the first uh GitHub open
[20:42] (1242.48s)
source project that I've ever sponsored.
[20:44] (1244.08s)
And it was uh for me it was like, man,
[20:46] (1246.64s)
this it just is such a fun project that
[20:51] (1251.20s)
helps you understand
[20:53] (1253.68s)
the usage of these tools and and helps
[20:56] (1256.88s)
you understand the value you're getting
[20:58] (1258.40s)
out of a Cloud Max plan. Um but also I
[21:01] (1261.44s)
think helps me stay accountable.
[21:04] (1264.32s)
when you sign up for the Claude Max
[21:05] (1265.84s)
plan, you almost like feel bad walking
[21:08] (1268.16s)
away from your machine because you feel
[21:09] (1269.76s)
like if it's not working, I'm losing
[21:11] (1271.92s)
value, right? And so I actually really
[21:14] (1274.80s)
love starting my day by looking at CC
[21:16] (1276.64s)
usage. So I really appreciate you
[21:18] (1278.40s)
creating this. I've had so much fun with
[21:20] (1280.00s)
it. My friends and I have had a lot of
[21:21] (1281.92s)
fun and it has sparked some really great
[21:24] (1284.16s)
conversations between us and ultimately
[21:25] (1285.84s)
I think is helping us get better at
[21:27] (1287.20s)
using uh cloud code. So thank you for
[21:30] (1290.08s)
Thank you so much. I think I think CC
[21:31] (1291.92s)
usage has changed my life. So I got I
[21:35] (1295.12s)
got a lot of attention. I had so many
[21:38] (1298.80s)
conversation towards not only C uh
[21:41] (1301.68s)
contributors but also like customers or
[21:44] (1304.64s)
users including you. Yeah. So I guess so
[21:48] (1308.40s)
before before your email I I saw your
[21:52] (1312.32s)
video and Yeah. Now I'm here and I'm
[21:55] (1315.60s)
talking with you and I can connect with
[21:58] (1318.48s)
lots of people. It's really fun and I'm
[22:02] (1322.08s)
I'm really honored to have this cool
[22:05] (1325.76s)
project.
[22:06] (1326.80s)
Oh man, it there's so much fun stuff
[22:08] (1328.72s)
happening right now and I think you just
[22:10] (1330.56s)
never know what thing you're going to
[22:13] (1333.20s)
make that is going to uh you know land
[22:17] (1337.12s)
and and excite the internet. Um and and
[22:19] (1339.76s)
it's just I I feel like there's just
[22:22] (1342.24s)
never been a better time to be a
[22:24] (1344.00s)
software developer. And you know, you've
[22:27] (1347.04s)
got this project now that I I don't even
[22:30] (1350.08s)
know. I know you're at three and a half
[22:31] (1351.28s)
thousand stars. I don't even know how
[22:32] (1352.80s)
many downloads or installs. What is it
[22:34] (1354.40s)
at right now?
[22:35] (1355.92s)
It's it's 90k.
[22:38] (1358.00s)
90,000. Incredible, right? So, that's so
[22:40] (1360.72s)
cool. Like, what an amazing time uh to
[22:42] (1362.88s)
be able to to write code. So, thank you
[22:45] (1365.12s)
very much for for making it and thanks
[22:46] (1366.80s)
for uh hanging out with me here today.
[22:49] (1369.04s)
Yeah. Yeah. Thank you for having me.