1
00:00:00,00 –> 00:00:04,560
Hey everybody and welcome to the Friday wrap up on Streamline Solopreneur, a short

2
00:00:04,560 –> 00:00:09,120
episode where I cover four things, what's on my mind this week? Recommended reading

3
00:00:09,120 –> 00:00:14,190
something fun and something to help your systems. This is the show that helps overwhelmed

4
00:00:14,190 –> 00:00:18,920
coaches, consultants, and service providers build systems so their business doesn't

5
00:00:18,930 –> 00:00:23,420
depend on them. Hopefully this week's curation gets you thinking about your own systems

6
00:00:23,660 –> 00:00:29,390
Before you close the laptop for the weekend, I'm your host, Joe Casabona, and here's

7
00:00:29,510 –> 00:00:33,570
what's on my mind for August 21st, 2026,

8
00:00:33,570 –> 00:00:37,40
what is conscientious AI use?

9
00:00:37,40 –> 00:00:43,430
As is usually the case, I wrote about AI on LinkedIn and it took off becoming my

10
00:00:43,460 –> 00:00:46,220
second most popular post in the last 90 days

11
00:00:46,220 –> 00:00:49,580
behind only a post to Doist paid to boost.

12
00:00:49,580 –> 00:00:54,110
It's a contrarian view that I don't feel should be contrarian.

13
00:00:54,110 –> 00:00:56,810
If you didn't write it, it's not yours.

14
00:00:56,810 –> 00:01:01,460
I'll touch more on this in the recommended reading segment, but ultimately it boils

15
00:01:01,460 –> 00:01:02,270
down to this.

16
00:01:02,270 –> 00:01:09,530
If you have an idea and outsource it to an LLM to write about that idea, it's not

17
00:01:09,530 –> 00:01:17,90
yours. Ideas are not executions. If they were, I'd get credit for thinking of Facebook

18
00:01:17,120 –> 00:01:22,250
or as I put it, MySpace, but only for college students back in 2003.

19
00:01:22,350 –> 00:01:25,680
But as I said, you'll get more of that in the next segment.

20
00:01:25,680 –> 00:01:31,650
This LinkedIn post was the catalyst for what's actually on my mind this week, conscientious

21
00:01:31,680 –> 00:01:37,140
AI use. This was a term proposed to me by my mastermind group. Shout out to,

22
00:01:37,140 –> 00:01:43,380
Jen, Liz, Rahul, Tanya, and Mike. And I've been thinking about it ever since.

23
00:01:43,380 –> 00:01:49,290
I've written a lot about AI over the last couple of years, but I've always been hesitant

24
00:01:49,350 –> 00:01:57,390
to brand myself as an AI person, even a contrarian one because I didn't want to be

25
00:01:57,390 –> 00:02:03,690
viewed as taking advantage of some ai. Gold rush tools come and go, methods come

26
00:02:03,690 –> 00:02:07,860
and go, but it's really the core principles that matter when you're doing something

27
00:02:07,860 –> 00:02:13,320
that said more people are asking me about it and it's becoming a legitimate part

28
00:02:13,380 –> 00:02:15,00
of systems.

29
00:02:15,00 –> 00:02:18,300
I can't really ignore it or avoid it anymore.

30
00:02:18,300 –> 00:02:26,460
The other event that is pushing me towards this thinking about conscientious AI usage

31
00:02:26,880 –> 00:02:32,640
is that Claude has broken spectacularly for me this week.

32
00:02:32,640 –> 00:02:37,560
Basically from when I started my weekly planning on Sunday up until right before

33
00:02:37,560 –> 00:02:43,500
I hit record every day. Claude has given me a different problem. Whatever changes

34
00:02:43,500 –> 00:02:49,840
they made in last week's updates has screwed up a bunch of my skills and projects.

35
00:02:49,840 –> 00:02:57,790
and for some this would put a halt to their work as they rely so heavily on AI to

36
00:02:57,790 –> 00:03:02,800
do the work for them. I know this because when Claude is down, there are people who

37
00:03:02,800 –> 00:03:09,550
legitimately cannot work a lot, say This must be a sign to go outside. I say a sign

38
00:03:09,550 –> 00:03:12,130
to go outside is you want to go outside,

39
00:03:12,130 –> 00:03:15,730
not the thing you've outsourced your thinking to is down.

40
00:03:15,730 –> 00:03:22,810
One tenant of conscientious AI usage is that you're using it for specific tasks

41
00:03:22,810 –> 00:03:24,100
and not craft

42
00:03:24,100 –> 00:03:28,630
meaning that when the large language model breaks your whole business

43
00:03:28,630 –> 00:03:31,180
or your whole work day isn't

44
00:03:32,50 –> 00:03:32,560
upended

45
00:03:32,560 –> 00:03:36,910
that you don't have to spend a bunch of time fixing the thing so the work can get

46
00:03:36,910 –> 00:03:37,240
done

47
00:03:37,240 –> 00:03:40,600
that you should be able to continue working

48
00:03:40,600 –> 00:03:45,10
or make small tweaks like I did with my copy editor skill

49
00:03:45,10 –> 00:03:49,90
or like I did with another project I use for podcasting.

50
00:03:49,90 –> 00:03:55,290
I will have more to say about this in the coming weeks, but in the meantime, I have

51
00:03:55,290 –> 00:04:02,70
set up a new hub for all of my AI content. You can find it over at Love-hate dot

52
00:04:02,100 –> 00:04:08,100
ai. That's love-hate dot ai. I am actually shocked that domain was available.

53
00:04:08,100 –> 00:04:15,660
that is going to be where all of my AI writing podcast episodes and YouTube videos

54
00:04:15,780 –> 00:04:20,190
will be. By the way, when I say AI writing, I mean things I wrote about ai, not things

55
00:04:20,190 –> 00:04:21,840
I had AI write.

56
00:04:21,840 –> 00:04:26,140
And of course, if you have something on your mind, what do you think about conscientious

57
00:04:26,140 –> 00:04:32,50
AI usage? You can let me know over@streamlinedfeedback.com. You can write in or leave

58
00:04:32,50 –> 00:04:35,320
a voice note. I might even play the voice note on the show.

59
00:04:35,320 –> 00:04:40,270
Okay, let's get to the recommended reading. This is from Dan Morin over at Six Colors.

60
00:04:40,270 –> 00:04:42,40
It's called LLMs RT Writing.

61
00:04:42,40 –> 00:04:46,660
And this is another thing that was pushing me towards conscientious AI usage.

62
00:04:46,660 –> 00:04:52,810
Several people have asked me what I think about the lexical watermark that Anthropic

63
00:04:52,810 –> 00:04:55,450
is adding to writing in its new models.

64
00:04:55,450 –> 00:05:00,760
my most glib answer has been, I'm not worried about it because I'm not using LLMs

65
00:05:00,760 –> 00:05:05,500
to write for me. I don't think anyone should. So this is actually, Annette, good

66
00:05:05,500 –> 00:05:05,890
for me,

67
00:05:05,890 –> 00:05:12,490
but I was also very confused to read John Gruber's very angry take about it.

68
00:05:12,490 –> 00:05:19,130
He is so mad that this watermarking is happening. He is so mad that AI has stolen

69
00:05:19,130 –> 00:05:26,00
writing, which I am too. But the screed reads differently from what I would expect

70
00:05:26,00 –> 00:05:30,20
because my understanding is that he doesn't use AI to write either. So like

71
00:05:30,20 –> 00:05:32,900
shouldn't just just push more people

72
00:05:32,900 –> 00:05:34,130
to not use it.

73
00:05:34,130 –> 00:05:38,960
And truth be told, I've never been a huge fan of John Gruber's work, though I do

74
00:05:38,960 –> 00:05:40,850
respect what he's done. I mean, he's like an

75
00:05:41,480 –> 00:05:41,540
OG

76
00:05:41,540 –> 00:05:44,240
actually get paid to blog Blogger.

77
00:05:44,240 –> 00:05:49,820
My take is actually much closer to Dan Morin's at Six Colors and it is today's recommended

78
00:05:49,820 –> 00:05:51,770
reading. So I'm gonna quote this directly here.

79
00:05:51,770 –> 00:05:56,990
If this does truly make AI generated text worse, well good.

80
00:05:56,990 –> 00:06:01,610
A lot of people are already willing to accept what an LLM turns out as good enough.

81
00:06:02,00 –> 00:06:05,780
And if I'm being realistic, I don't think this will change anything.

82
00:06:05,780 –> 00:06:11,90
But if it does lead to more people being dissatisfied with the plum they're being

83
00:06:11,90 –> 00:06:15,80
fed and turning instead to writing and editing their own text,

84
00:06:15,80 –> 00:06:17,360
then that would actually be a positive outcome.

85
00:06:17,360 –> 00:06:21,680
I agree wholeheartedly with this quote, and I feel a lot of the backlash is revolving

86
00:06:21,680 –> 00:06:27,500
around two aspects of AI writing. The first is that AI is changing the words to make

87
00:06:27,500 –> 00:06:29,60
its output worse, which

88
00:06:29,60 –> 00:06:33,500
seems to not really be the case, and maybe that's going to get worse. But again,

89
00:06:33,500 –> 00:06:39,950
if you are so concerned about the word choices, which you should be,

90
00:06:39,950 –> 00:06:47,60
we need to think about the words we use. I tell my children, if they don't know the

91
00:06:47,60 –> 00:06:53,450
meaning of a word, they cannot say it. This is what communication is. It's why I'm

92
00:06:53,450 –> 00:06:54,560
so passionate

93
00:06:54,560 –> 00:06:57,110
that we shouldn't let AI write for us.

94
00:06:57,110 –> 00:06:59,150
'cause when we are letting a machine

95
00:06:59,150 –> 00:07:02,870
communicate what we feel are important thoughts,

96
00:07:02,870 –> 00:07:07,910
We are robbing people of our actual thoughts.

97
00:07:07,910 –> 00:07:13,430
And so we should be concerned about the words that are being used on our behalf.

98
00:07:13,430 –> 00:07:20,390
And so if this watermarking thing gets worse and more, more audacious, great, we

99
00:07:20,390 –> 00:07:22,370
should all use AI writing less.

100
00:07:22,370 –> 00:07:23,390
But the other side of it

101
00:07:23,390 –> 00:07:27,980
is that people are being outed for using AI in their writing. Like, why are we mad

102
00:07:27,980 –> 00:07:30,530
about this? This is a good thing.

103
00:07:30,530 –> 00:07:39,260
If you are unashamedly using AI to do your writing, then you should disclose that

104
00:07:39,260 –> 00:07:40,200
you used it.

105
00:07:40,200 –> 00:07:47,610
If you think it's so great that I just give Claude an idea and my team of agents

106
00:07:47,610 –> 00:07:52,110
makes it sound just like me, which it doesn't sound like

107
00:07:52,110 –> 00:07:52,590
you,

108
00:07:52,590 –> 00:07:55,980
you are diluting yourself if you believe that.

109
00:07:55,980 –> 00:07:58,20
But if you really feel that way,

110
00:07:58,20 –> 00:08:01,350
give the large language model a byline on the work.

111
00:08:01,350 –> 00:08:03,900
Give credit where credit is due,

112
00:08:03,900 –> 00:08:08,340
because otherwise, in my view, it is

113
00:08:08,340 –> 00:08:12,60
so close to the line of plagiarism

114
00:08:12,60 –> 00:08:16,80
I suspect I'll think about this a a a bit more, a lot more.

115
00:08:16,80 –> 00:08:20,550
Writing is one of the most important things that anyone can do. It's not just a mechanism.

116
00:08:20,550 –> 00:08:23,130
It's not just a means to an end.

117
00:08:23,130 –> 00:08:25,680
It's a universally important skill.

118
00:08:25,680 –> 00:08:28,410
Not everyone needs to be an incredible writer.

119
00:08:28,410 –> 00:08:31,680
Not everyone needs to be Shakespeare or whoever,

120
00:08:31,680 –> 00:08:35,880
but we should all be able to state our own thoughts.

121
00:08:35,880 –> 00:08:40,470
We should all be able to think through an opinion,

122
00:08:40,470 –> 00:08:45,390
not just give AI a vague idea and have it write it for us.

123
00:08:45,390 –> 00:08:46,250
Okay,

124
00:08:46,820 –> 00:08:50,150
well, with all that, let's move on to something fun.

125
00:08:50,150 –> 00:08:51,740
That's something fun as Halo three.

126
00:08:51,740 –> 00:08:54,140
Last weekend, it was a rainy, sleepy weekend.

127
00:08:54,140 –> 00:08:59,660
As we in the household in southeastern pa as we prepare to go back to school,

128
00:08:59,660 –> 00:09:01,430
we're tired from the summer,

129
00:09:01,430 –> 00:09:04,520
so we decided to have a family screen day.

130
00:09:04,520 –> 00:09:10,940
The kids enjoyed movies and games and I dusted off the old Xbox, paid $24 for a month

131
00:09:10,940 –> 00:09:13,760
of Game Pass and I booted up Halo three.

132
00:09:13,760 –> 00:09:17,780
I don't play video games much these days. I mean, I do like the

133
00:09:17,780 –> 00:09:21,560
mobile games from New York Times. I play Immaculate Grid

134
00:09:21,560 –> 00:09:21,980
But

135
00:09:21,980 –> 00:09:27,710
true blue video games, right? These would consume large portions of my days in high

136
00:09:27,710 –> 00:09:28,790
school and college.

137
00:09:28,790 –> 00:09:29,900
Sometimes

138
00:09:29,900 –> 00:09:35,60
me and my dorm mates would play from early in the morning until past midnight,

139
00:09:35,60 –> 00:09:38,810
But as games evolved, they got more complicated

140
00:09:38,810 –> 00:09:44,00
and they require much more of a time commitment that I'm willing to give,

141
00:09:44,00 –> 00:09:47,210
especially 'cause there's a bunch of other stuff I want to do and I have to do.

142
00:09:47,210 –> 00:09:52,610
That's why I love old school games like Halo or Gears of War, these run and gun games,

143
00:09:52,610 –> 00:09:56,330
you know you're going the right way as long as you see more bad guys

144
00:09:56,330 –> 00:10:01,490
and there are no side quests or massive customizations. I knew Call of Duty left

145
00:10:01,490 –> 00:10:07,940
me behind when I had to like pick all the gear I had and the type of soldier I wanted

146
00:10:07,940 –> 00:10:14,540
to be. I just wanna pick up a gun that I run over and and use it. I don't wanna have

147
00:10:14,540 –> 00:10:17,900
to think about every single attribute of my character.

148
00:10:17,900 –> 00:10:21,50
And so I had a lot of fun playing Halo three last weekend.

149
00:10:21,50 –> 00:10:25,880
I don't know how much more I'll revisit it before school and travel startups. September

150
00:10:25,880 –> 00:10:28,670
and October are going to be very busy for me,

151
00:10:28,670 –> 00:10:34,40
but I do know that it's supposed to be another rainy weekend this weekend here in

152
00:10:34,40 –> 00:10:35,450
southeastern pa.

153
00:10:35,450 –> 00:10:40,160
Okay, let's move on to something to help your systems.

154
00:10:40,160 –> 00:10:42,470
And I wanna talk about something I did this week with

155
00:10:42,470 –> 00:10:43,10
AI

156
00:10:43,10 –> 00:10:44,90
and it's creating

157
00:10:44,90 –> 00:10:45,260
a quiz with Claude.

158
00:10:45,260 –> 00:10:50,60
I honestly don't know how much longer quizzes as lead magnets will be useful.

159
00:10:50,60 –> 00:10:50,930
I know

160
00:10:50,930 –> 00:10:52,670
like upsettingly little

161
00:10:52,670 –> 00:10:57,410
about what makes a good lead magnet, but I worry that they'll saturate the market

162
00:10:57,410 –> 00:10:58,520
soon. And here's why.

163
00:10:58,520 –> 00:11:02,660
Generally, quizzes have been something aspirational for me

164
00:11:02,660 –> 00:11:06,590
due to the time constraints and the fact that I didn't want to pay for all of the

165
00:11:06,590 –> 00:11:11,420
industry leading services, right? I think like SCORE app is, is

166
00:11:11,420 –> 00:11:13,970
50 or a hundred bucks a month or something like that,

167
00:11:13,970 –> 00:11:16,340
but large language models have changed that

168
00:11:16,340 –> 00:11:21,480
Earlier this week in conjunction with my thinking around conscientious AI usage,

169
00:11:21,900 –> 00:11:26,430
I decided to make an Are you suffering from AI Brain Rot quiz.

170
00:11:26,430 –> 00:11:32,850
I have the aforementioned hub of content over@lovehate.ai, which I fed into Claude

171
00:11:32,850 –> 00:11:39,780
to help me make a short eight question quiz evaluating my four dimensions of AI brain

172
00:11:39,780 –> 00:11:44,700
rot, which is using it reflexively. That is to say, not thinking of a problem for

173
00:11:44,700 –> 00:11:47,280
more than a few minutes before asking a large language model,

174
00:11:47,280 –> 00:11:51,990
talking to AI instead of people. So asking like, what does my audience want? Instead

175
00:11:51,990 –> 00:11:54,30
of asking your audience, what do you want?

176
00:11:54,30 –> 00:11:56,490
Having no skin in the game, which is

177
00:11:56,490 –> 00:12:00,180
because you didn't come up with something, you're actually not really committed to

178
00:12:00,180 –> 00:12:04,680
it. You're just burning tokens and not really thinking about the the things that

179
00:12:04,680 –> 00:12:05,220
you're doing.

180
00:12:05,220 –> 00:12:11,230
And using AI as a coach, and we know that AI cannot advise because it cannot have

181
00:12:11,230 –> 00:12:13,180
opinions because it's not thinking,

182
00:12:13,180 –> 00:12:16,390
it is doing word association.

183
00:12:16,390 –> 00:12:21,130
So it looks at the quiz, looks at my four dimensions of AI brain rock, and then gives

184
00:12:21,130 –> 00:12:22,480
the user a score

185
00:12:22,480 –> 00:12:27,550
and then the user will get a resource and recommendations for conscientious AI usage.

186
00:12:28,30 –> 00:12:29,200
That resource

187
00:12:29,200 –> 00:12:34,990
is A PDF and it was generated from a markdown file that I wrote using my design system

188
00:12:34,990 –> 00:12:38,710
and brand guidelines that I gave to Claude Design.

189
00:12:38,710 –> 00:12:40,780
The most kind of,

190
00:12:40,780 –> 00:12:44,560
I'll say craft work that Claude did here

191
00:12:44,560 –> 00:12:48,940
was come up with the scoring and questions based on my writing,

192
00:12:48,940 –> 00:12:52,480
and then I had it mark all of the places where it created placeholder copy so I can

193
00:12:52,480 –> 00:12:53,440
go in and change it.

194
00:12:53,440 –> 00:12:59,620
Claude Code then wrote the application and ran its automated tests, and then I tested

195
00:12:59,620 –> 00:13:05,500
it independently and wired it up to kit.com, which is my email service provider,

196
00:13:05,500 –> 00:13:07,90
and it's hosted on Netlify.

197
00:13:07,90 –> 00:13:10,930
All of this would've been in incredibly time intensive

198
00:13:10,930 –> 00:13:13,750
and it would not have looked nearly as good.

199
00:13:13,750 –> 00:13:19,870
Instead, it took me a couple of hours between meetings because Claude already had

200
00:13:19,870 –> 00:13:21,130
all of my content,

201
00:13:21,460 –> 00:13:23,980
knows my AP well, knows, quote unquote

202
00:13:23,980 –> 00:13:25,630
has. My approach in context,

203
00:13:25,630 –> 00:13:27,640
is connected to my kit account

204
00:13:27,640 –> 00:13:29,320
and is connected to Netlify.

205
00:13:29,320 –> 00:13:32,290
My last task in this project, which,

206
00:13:32,290 –> 00:13:34,360
should be done before you are hearing this,

207
00:13:34,360 –> 00:13:37,240
is to review all of the copy with fresh eyes,

208
00:13:37,220 –> 00:13:43,130
then I'll feel comfortable promoting it widely. But for now, you can find it

209
00:13:43,130 –> 00:13:45,500
at Brain Rot streamlined,

210
00:13:46,130 –> 00:13:46,610
FM

211
00:13:46,710 –> 00:13:51,720
But that's it for this episode of the Friday wrap up from Streamline Solopreneur.

212
00:13:51,750 –> 00:13:55,140
These episodes are about taking more time off, especially

213
00:13:55,140 –> 00:14:00,930
on the weekends, and this has been an AI intensive episode. So if you wanna be more

214
00:14:00,930 –> 00:14:07,50
conscientious about how you are using AI to save you time without replacing you

215
00:14:07,50 –> 00:14:09,90
in the process of your work,

216
00:14:09,90 –> 00:14:16,260
take the quiz over at Brain Rot streamlined FM to see where you land and how you

217
00:14:16,260 –> 00:14:18,180
can be using it more conscientiously.

218
00:14:18,180 –> 00:14:23,490
Thanks so much for listening to this episode, and until next time, I hope you find

219
00:14:23,490 –> 00:14:25,620
some space in your weekend.