1
00:00:00,600 --> 00:00:04,080
This is the Elon Musk Podcast, 
your daily hit of what is really

2
00:00:04,080 --> 00:00:07,960
going on at Tesla, SpaceX X AI, 
and the rest of the Musk 

3
00:00:07,960 --> 00:00:10,200
universe. 
I'm your host Will Walden, and I

4
00:00:10,200 --> 00:00:13,560
have covered Elon Musk for more 
than five years, spent a year on

5
00:00:13,560 --> 00:00:16,160
the ground at SpaceX, Starbase 
during early Starship 

6
00:00:16,160 --> 00:00:19,480
development, and before this I 
spent my career as a software 

7
00:00:19,480 --> 00:00:22,200
developer working with billion 
dollar companies. 

8
00:00:22,520 --> 00:00:26,880
I've also built and sold my own 
businesses and now I make 

9
00:00:26,880 --> 00:00:29,320
content and help other people 
grow their companies. 

10
00:00:29,640 --> 00:00:32,320
Now on this show I used that 
experience to break down the 

11
00:00:32,320 --> 00:00:36,440
news, filter out all the noise, 
and give you clear context. 

12
00:00:36,680 --> 00:00:45,000
You can actually use a Gentek AI
gives an AI system the ability 

13
00:00:45,000 --> 00:00:47,880
to take small sensible steps 
towards a goal. 

14
00:00:47,880 --> 00:00:52,360
Instead of stopping after one 
single reply, think of it as 

15
00:00:52,360 --> 00:00:56,800
moving from a one time answer to
a finished task with receipts 

16
00:00:56,800 --> 00:01:00,800
that you can actually review. 
Businesses use it to book 

17
00:01:00,800 --> 00:01:04,680
travel, organize calendars, 
clean out spreadsheets, and 

18
00:01:04,680 --> 00:01:07,280
draft polite emails that follow 
company rules. 

19
00:01:07,880 --> 00:01:10,520
And you get outcomes rather than
lose suggestions with a record 

20
00:01:10,520 --> 00:01:12,800
of how each decision was 
actually happening. 

21
00:01:13,360 --> 00:01:15,240
And how does it actually work in
everyday life. 

22
00:01:15,240 --> 00:01:18,640
Though in this episode, you'll 
get a plain English definition, 

23
00:01:18,880 --> 00:01:21,960
a simple loop that you can 
picture in your head, and a few 

24
00:01:21,960 --> 00:01:25,200
real world stories you can adapt
at home or at work. 

25
00:01:25,760 --> 00:01:28,280
And we're going to talk about 
what it does well, where it 

26
00:01:28,280 --> 00:01:31,360
needs clear instructions of how 
to keep control with approvals, 

27
00:01:31,360 --> 00:01:34,880
spending caps, and time limits. 
And by the end, you'll know how 

28
00:01:34,880 --> 00:01:38,560
to start small, test safely, and
grow with confidence. 

29
00:01:38,560 --> 00:01:42,920
And we're going to get into that
right after this short 

30
00:01:43,000 --> 00:01:49,040
commercial break. 
Here is the short definition of 

31
00:01:49,200 --> 00:01:52,360
agentic AI. 
It is a system that pursues a 

32
00:01:52,400 --> 00:01:56,480
goal through several steps, 
using tools along the way, until

33
00:01:56,480 --> 00:01:59,800
it succeeds or runs out of 
allowed attempts. 

34
00:02:00,720 --> 00:02:02,840
Regular chat gives you one 
response. 

35
00:02:03,280 --> 00:02:06,760
Then you ask it again, and again
and again and again. 

36
00:02:06,760 --> 00:02:10,600
You get a bunch of responses. 
An agent continues on its own 

37
00:02:10,639 --> 00:02:12,880
within the rules that you set 
for it. 

38
00:02:13,480 --> 00:02:16,520
That shift from answers to 
outcomes changes how you design 

39
00:02:16,520 --> 00:02:20,560
tasks, how you approve actions, 
and how you judge success. 

40
00:02:21,440 --> 00:02:24,960
Now picture a helpful assistant 
with a checklist. 

41
00:02:26,080 --> 00:02:30,120
You say plan a family weekend 
that fits a budget, includes a 

42
00:02:30,120 --> 00:02:32,440
museum visit. 
Keep Sunday afternoon free. 

43
00:02:33,320 --> 00:02:36,960
And the assistant breaks the 
goal into steps, chooses the 

44
00:02:36,960 --> 00:02:40,440
next step, uses the right apps, 
writes down what it found, and 

45
00:02:40,920 --> 00:02:45,040
it decides what to do next. 
In agentic AI, the assistant is 

46
00:02:45,040 --> 00:02:47,200
the model. 
The apps are tools like a 

47
00:02:47,200 --> 00:02:50,200
calendar, a map, a price look 
up, or your notes. 

48
00:02:50,320 --> 00:02:52,760
In the written record is a log 
you can read later. 

49
00:02:53,600 --> 00:02:56,880
System repeats this plan, act 
and check pattern until it 

50
00:02:56,880 --> 00:02:59,720
reaches the goal or asks you for
help. 

51
00:02:59,720 --> 00:03:02,560
It may fail and the main parts 
are easy to remember. 

52
00:03:02,560 --> 00:03:06,680
First, the goal, which states 
what done looks like in simple 

53
00:03:06,680 --> 00:03:10,720
terms such as stay under $600, 
keep walking time short, and 

54
00:03:10,720 --> 00:03:15,520
pick refundable options. 
Second, a planner which turns 

55
00:03:15,520 --> 00:03:17,920
the goal into a short list of 
small steps. 

56
00:03:18,520 --> 00:03:22,400
Third, tools which are safe 
buttons the agent can push like 

57
00:03:22,640 --> 00:03:26,040
search hotels, read a 
spreadsheet, or add an event. 

58
00:03:26,560 --> 00:03:29,240
4th. 
Memory, which stores notes from 

59
00:03:29,240 --> 00:03:32,320
each step so the next step does 
not forget what just happened. 

60
00:03:32,800 --> 00:03:35,360
5th. 
A checker which asks if the last

61
00:03:35,360 --> 00:03:39,400
step helped or hurt. 6th. 
A stop rule which ends the run 

62
00:03:39,400 --> 00:03:43,080
when the job is done, when a 
limit is hit, or when the system

63
00:03:43,080 --> 00:03:45,520
needs your approval, or when it 
completely fails. 

64
00:03:46,480 --> 00:03:51,520
Now let us walk through a travel
example without any jargon. 

65
00:03:51,760 --> 00:03:54,800
You tell the agent plan a 2 day 
Austin trip for two adults, keep

66
00:03:54,800 --> 00:03:58,800
lodging out of $200 per night, 
include a live music event, and 

67
00:03:58,800 --> 00:04:02,240
leave time for some nice Texas 
BBQ. 

68
00:04:03,680 --> 00:04:07,680
Now the planner lays out steps 
like find dates, look up hotels,

69
00:04:08,680 --> 00:04:11,280
check refund policies, search 
music events, draft the 

70
00:04:11,280 --> 00:04:13,440
schedule, and prepare a summary 
for you. 

71
00:04:14,800 --> 00:04:20,120
The agent goes to work. 
It uses a hotel lookup tool, 

72
00:04:20,839 --> 00:04:24,200
writes down choices for you, 
compares prices for your budget,

73
00:04:25,280 --> 00:04:27,840
and drops the top pick onto a 
calendar for you. 

74
00:04:28,600 --> 00:04:31,560
And after each step, the checker
asks if the plan still fits the 

75
00:04:31,560 --> 00:04:33,960
rules. 
Now the stop rule ends the loop 

76
00:04:33,960 --> 00:04:37,800
when the schedule, budget, and 
refund notes meet that goal. 

77
00:04:38,320 --> 00:04:41,040
Then the agent hands you a clear
summary and a log of every 

78
00:04:41,040 --> 00:04:42,400
single action that happened 
there. 

79
00:04:43,760 --> 00:04:47,760
And the style of AI works best 
on chores with clear rules, 

80
00:04:48,440 --> 00:04:50,520
repeatable steps, and measurable
results. 

81
00:04:50,680 --> 00:04:54,000
Reading bills, copying key 
numbers into a tracker, and 

82
00:04:54,000 --> 00:04:58,320
writing a short status note fits
well because each step has a 

83
00:04:58,400 --> 00:05:01,280
right or wrong outcome. 
Drafting polite replies to 

84
00:05:01,280 --> 00:05:04,480
common messages fits well 
because examples teach the tone 

85
00:05:04,480 --> 00:05:08,600
and structure. open-ended tasks 
with fuzzy goals like make 

86
00:05:08,600 --> 00:05:12,320
something creative for the 
website isn't a good one. 

87
00:05:12,320 --> 00:05:15,520
It's not a good prompt for that.
It can drift because the success

88
00:05:15,520 --> 00:05:19,560
target is not clear. 
You guide the agent by setting a

89
00:05:19,560 --> 00:05:24,120
very narrow goal. 
List rules in plain English and 

90
00:05:24,200 --> 00:05:29,480
name the tools it may use. 
Now you got to think about 

91
00:05:29,480 --> 00:05:32,600
control and safety. 
It's all within this loop, but 

92
00:05:32,600 --> 00:05:35,080
not at the end. 
You give the agent only approved

93
00:05:35,080 --> 00:05:38,920
tools like a read only calendar 
or hotel search that cannot buy 

94
00:05:38,920 --> 00:05:41,560
anything. 
You add spending caps, time 

95
00:05:41,560 --> 00:05:43,560
limits, and a maximum number of 
steps. 

96
00:05:43,960 --> 00:05:47,320
You require human approval for 
anything irreversible, such as 

97
00:05:47,320 --> 00:05:51,760
purpose purchases, deletions, or
messages to customers. 

98
00:05:52,200 --> 00:05:55,120
You keep private information 
safe by redacting secrets, 

99
00:05:55,200 --> 00:05:58,720
limiting who the agent can 
contact, and preventing it from 

100
00:05:58,720 --> 00:06:02,600
pasting data into public sites. 
These basic moves act like 

101
00:06:02,600 --> 00:06:06,680
seatbelts, locks, and curfews 
for a very fast helper, and the 

102
00:06:06,680 --> 00:06:10,160
helper can do these tasks within
seconds sometimes. 

103
00:06:10,840 --> 00:06:14,280
So you have to have rules. 
Now there are a few simple 

104
00:06:14,280 --> 00:06:18,120
shapes for how agents works. 
One helper handles a straight 

105
00:06:18,120 --> 00:06:21,880
path, like reading a form, 
filling a tracker, and writing a

106
00:06:21,880 --> 00:06:25,520
summary about it. 
A planner and a doer split the 

107
00:06:25,520 --> 00:06:28,800
job, which reduces trial and 
error and keeps the log knee. 

108
00:06:29,440 --> 00:06:33,120
A small team uses a supervisor 
to route tasks to specialists, 

109
00:06:33,120 --> 00:06:36,480
like a calendar specialist, a 
data specialist, and a writing 

110
00:06:36,480 --> 00:06:38,920
specialist. 
Many teams do well with the 

111
00:06:38,920 --> 00:06:42,080
planner and do repair because it
stays simple, exposes the 

112
00:06:42,080 --> 00:06:44,560
decision points, and makes 
reviews really fast. 

113
00:06:45,400 --> 00:06:47,440
Now you do not need a big 
platform to start. 

114
00:06:47,520 --> 00:06:50,960
Think of tools as small safe 
buttons with clear labels and 

115
00:06:50,960 --> 00:06:53,720
clear limits. 
One button reads a file, 1 

116
00:06:53,720 --> 00:06:57,280
button looks up a price, 1 
button adds a calendar entry. 

117
00:06:57,560 --> 00:07:00,040
Each button returns a simple 
result that the agent can 

118
00:07:00,040 --> 00:07:03,360
understand, and the log shows 
which button the agent pressed, 

119
00:07:03,880 --> 00:07:06,680
which words it sent into the 
button, and what came back. 

120
00:07:07,240 --> 00:07:10,800
If something looks odd, you can 
replay the run and see exactly 

121
00:07:10,800 --> 00:07:14,200
where it went off track and you 
can figure out how to fix that. 

122
00:07:15,120 --> 00:07:17,880
Humans are still needed at this 
point with AI. 

123
00:07:18,320 --> 00:07:23,200
Some are really good. 
I run an agent sometime to get 

124
00:07:23,200 --> 00:07:27,440
me the latest news on Elon Musk 
and it does wonders for me. 

125
00:07:28,600 --> 00:07:32,720
Now. 
The process that it takes is 

126
00:07:32,720 --> 00:07:36,880
very straightforward. 
I say check these amount of news

127
00:07:36,880 --> 00:07:41,720
sources, check 20 news sources, 
etcetera, etcetera, and then 

128
00:07:41,720 --> 00:07:44,840
send me the links. 
It's very simple. 

129
00:07:45,160 --> 00:07:48,320
My my agent is very simple. 
Send me the links, give me a 

130
00:07:48,520 --> 00:07:52,640
like a three bullet point 
rundown of what the article is 

131
00:07:52,640 --> 00:07:54,520
about. 
See if it's worth my time. 

132
00:07:55,080 --> 00:08:00,520
I'm going to redevelop this 
agent into something a little 

133
00:08:00,520 --> 00:08:02,880
bit more robust. 
So it's easier for me to do this

134
00:08:02,880 --> 00:08:04,600
podcast. 
I've been doing this podcast for

135
00:08:04,600 --> 00:08:09,320
a very, very long time, five 
years or so thousand episodes 

136
00:08:09,320 --> 00:08:14,200
plus. 
So if you are a fan of the show,

137
00:08:14,200 --> 00:08:17,960
thank you, and if you aren't, 
thank you for stopping by and 

138
00:08:17,960 --> 00:08:21,640
listening. 
And also, since you are a new 

139
00:08:21,640 --> 00:08:24,640
fan to the show, please take a 
second and hit the follow 

140
00:08:24,640 --> 00:08:26,280
button. 
That'll be really helpful. 

141
00:08:26,280 --> 00:08:30,160
Or pick up some merch at 
starshipshirts.com. 

142
00:08:30,520 --> 00:08:36,559
That helps out tremendously. 
So agents are really cool 

143
00:08:36,559 --> 00:08:40,039
because you can measure an agent
the same way you measure a 

144
00:08:40,039 --> 00:08:44,120
person who handles a task. 
Then it finished the task, yes 

145
00:08:44,120 --> 00:08:46,240
or no. 
How long did it take? 

146
00:08:46,240 --> 00:08:49,960
How much did it cost to run it? 
How often did it ask for help? 

147
00:08:50,840 --> 00:08:53,240
Now those numbers give you a 
clear picture of whether the 

148
00:08:53,240 --> 00:08:56,720
system saves time or actually 
creates more work for you. 

149
00:08:57,360 --> 00:09:02,400
When I first started developing 
my agent, I was, it was taking 

150
00:09:02,400 --> 00:09:07,760
me more time than it than it was
before I had it just to get it 

151
00:09:07,760 --> 00:09:13,000
right. 
But by fine tuning it, I have a 

152
00:09:13,080 --> 00:09:17,720
clear path to, you know, the 
best results for the day. 

153
00:09:18,200 --> 00:09:21,320
I can't just Google things. 
You get a bunch of just junk in 

154
00:09:21,320 --> 00:09:23,720
there and you can't just chat 
TBT things because it doesn't 

155
00:09:23,720 --> 00:09:26,720
give you the right answers. 
So you have to make something. 

156
00:09:27,040 --> 00:09:29,760
I had to make something custom. 
I had to make a custom agent to 

157
00:09:29,760 --> 00:09:31,000
do the things that I wanted to 
do. 

158
00:09:31,000 --> 00:09:32,880
And it has a very simple clear 
task. 

159
00:09:34,200 --> 00:09:37,560
And then it doesn't fail anymore
because it knows exactly what to

160
00:09:37,560 --> 00:09:39,680
do, has like 4 things it needs 
to do. 

161
00:09:41,360 --> 00:09:46,920
Search the web on certain sites.
Send me back the headline. 

162
00:09:47,800 --> 00:09:50,680
Send me back three bullet points
about the article because it 

163
00:09:50,680 --> 00:09:53,000
reads the article. 
Send back three bullet points of

164
00:09:53,000 --> 00:09:56,600
the article. 
Send me a link to the news piece

165
00:09:56,600 --> 00:09:59,920
so I can check it out myself. 
And that's it. 

166
00:10:00,280 --> 00:10:03,360
That's all it does. 
And then I read everything I can

167
00:10:03,360 --> 00:10:04,960
and come up with stuff on my 
own. 

168
00:10:05,480 --> 00:10:08,120
So my agent is very simple. 
You can make something 

169
00:10:08,160 --> 00:10:11,920
absolutely complex. 
Crazy, you know? 

170
00:10:11,960 --> 00:10:16,640
How often does it ask for help? 
That's the important one, right?

171
00:10:17,560 --> 00:10:19,680
So those numbers give you a 
clear picture of whether the 

172
00:10:19,680 --> 00:10:23,280
system saves time or creates a 
huge headache for you. 

173
00:10:23,280 --> 00:10:28,120
And if it creates the headache, 
either work on it more or find 

174
00:10:28,120 --> 00:10:31,280
another process. 
I had to go through 5 or 6 

175
00:10:31,280 --> 00:10:34,560
different processes in order for
my agent to work properly. 

176
00:10:34,920 --> 00:10:38,320
If you have the time, great. 
If not, there are systems out 

177
00:10:38,320 --> 00:10:40,240
there that will make you an 
agent. 

178
00:10:40,800 --> 00:10:44,080
Or you can do like a drag and 
drop agent for this kind of 

179
00:10:44,080 --> 00:10:46,520
stuff. 
And you have to treat each run 

180
00:10:46,520 --> 00:10:49,400
like a transaction with a result
you can accept or reject because

181
00:10:49,400 --> 00:10:52,560
each one of these costs compute 
time or tokens. 

182
00:10:52,560 --> 00:10:56,720
If you do it on something, you 
know where somebody else hosts 

183
00:10:56,720 --> 00:10:59,400
the LLM. 
And when you reject a result, 

184
00:10:59,920 --> 00:11:02,400
add a quick note about what went
wrong and then turn that note 

185
00:11:02,400 --> 00:11:04,680
into a rule the agent can follow
the next time. 

186
00:11:07,440 --> 00:11:12,800
And let's talk about like a 
story from an office because 

187
00:11:12,800 --> 00:11:14,520
that might put it into 
perspective. 

188
00:11:16,440 --> 00:11:20,280
So customer support, we all hate
it, right? 

189
00:11:20,360 --> 00:11:21,880
We all hate going through 
customer support. 

190
00:11:21,880 --> 00:11:23,360
If something goes wrong, we want
to answer now. 

191
00:11:24,480 --> 00:11:27,440
But if you're on the other end 
where you're a customer support 

192
00:11:27,440 --> 00:11:31,520
agent, inbox assistance are 
important. 

193
00:11:32,840 --> 00:11:36,200
And if you have a goal, the goal
would say, read each new 

194
00:11:36,200 --> 00:11:39,280
message, match it to one of five
common issues. 

195
00:11:39,720 --> 00:11:43,840
Suggest a reply that follows the
playbook, which is, you know, 

196
00:11:43,960 --> 00:11:47,040
you have a playbook that you 
feed it and hand anything 

197
00:11:47,040 --> 00:11:52,160
unusual to a person. 
So the agent opens an e-mail, 

198
00:11:52,840 --> 00:11:58,560
finds key details, checks the 
playbook, drafts a reply and 

199
00:11:58,560 --> 00:12:01,080
puts it into a queue for a 
person to review. 

200
00:12:01,720 --> 00:12:04,480
And if the person reviews it and
it needs a little bit of help, 

201
00:12:05,000 --> 00:12:08,680
type in there, you know, sorry, 
sorry you're having this issue. 

202
00:12:09,120 --> 00:12:11,520
My name is Will. 
I'm here to help you. 

203
00:12:11,520 --> 00:12:16,560
And that person confirms that 
the draft uses the right issue 

204
00:12:16,560 --> 00:12:19,360
type and the right tone. 
So it's basically a huge 

205
00:12:19,360 --> 00:12:21,440
database of things like the 
playbook. 

206
00:12:21,960 --> 00:12:24,840
Now, a person gives a quick 
approval on the first week of 

207
00:12:24,840 --> 00:12:29,480
use, then reduces approvals to 
only high stakes messages once 

208
00:12:29,520 --> 00:12:32,560
the numbers look steady. 
So you're training this model 

209
00:12:33,720 --> 00:12:38,440
while it's doing its job. 
Now, another one we can think 

210
00:12:38,440 --> 00:12:44,040
about is where you mix tools and
approvals. 

211
00:12:45,000 --> 00:12:49,040
So if you're a finance helper 
and it could check small 

212
00:12:49,040 --> 00:12:53,400
purchase requests, the goal says
ensure the request sits within 

213
00:12:53,400 --> 00:12:57,240
the budget, confirm the vendor 
is on the approved list and 

214
00:12:57,320 --> 00:13:01,360
prepare a draft purchase order. 
The agent reads the request, 

215
00:13:01,360 --> 00:13:03,920
looks up the budget balance, 
matches the vendor name, and 

216
00:13:03,920 --> 00:13:08,200
fills a form. 
The stop rule blocks any e-mail 

217
00:13:08,200 --> 00:13:12,920
to a vendor until a human clicks
approve, and the log acts like a

218
00:13:13,040 --> 00:13:15,920
receipt for an audit, which 
keeps everyone comfortable with 

219
00:13:15,920 --> 00:13:17,840
the tool that touches their 
money. 

220
00:13:18,120 --> 00:13:19,400
You don't mess with people's 
money, man. 

221
00:13:21,760 --> 00:13:26,320
So if you want to start this, 
it's straightforward and there 

222
00:13:26,320 --> 00:13:27,560
are tools, you have to look them
up. 

223
00:13:27,560 --> 00:13:34,560
I don't want to suggest anything
because I'm not very, I'm not 

224
00:13:34,560 --> 00:13:37,520
very educated in what tools are 
out there other than the ones 

225
00:13:37,520 --> 00:13:40,120
that I build myself. 
I'm a coder, so I build these 

226
00:13:40,120 --> 00:13:43,600
things myself. 
But to start, you can pick a 

227
00:13:43,640 --> 00:13:49,040
tiny job that annoys you 
somebody like pulling a number 

228
00:13:49,040 --> 00:13:52,120
from a document and dropping it 
into a tracker or drafting a 

229
00:13:52,120 --> 00:13:53,600
follow up e-mail after a 
meeting. 

230
00:13:54,200 --> 00:13:57,480
Google does this a lot. 
You know the all the e-mail apps

231
00:13:57,480 --> 00:14:00,880
do this now, but a follow up 
e-mail after a meeting? 

232
00:14:01,480 --> 00:14:03,120
You could have your agent do 
that for you. 

233
00:14:04,480 --> 00:14:08,680
Write a goal in one sentence and
list 3 rules the agent must 

234
00:14:08,680 --> 00:14:11,400
follow. 
It's super simple, a goal. 

235
00:14:13,600 --> 00:14:19,840
You could tell the the AI agent 
every time I get an e-mail, send

236
00:14:19,840 --> 00:14:24,040
it to my phone or what you know,
like give me a give me an alert 

237
00:14:24,040 --> 00:14:25,600
on my phone, something like 
that. 

238
00:14:25,840 --> 00:14:29,640
Send an alert to my phone number
and text you, you know, text you

239
00:14:29,640 --> 00:14:33,960
a summary of the e-mail. 
And if it sees seems like a big 

240
00:14:33,960 --> 00:14:37,760
deal, then sure, answer it. 
But if not, just leave it alone.

241
00:14:38,720 --> 00:14:42,160
And I know there are e-mail apps
that do that by themselves, but 

242
00:14:42,600 --> 00:14:44,560
you know, we're just making a, a
silly model right now. 

243
00:14:45,160 --> 00:14:49,040
Or if certain person emails me, 
send me a text. 

244
00:14:49,600 --> 00:14:50,920
How about that? 
That would be great. 

245
00:14:51,000 --> 00:14:55,960
That's an actual like a list of 
a actual thing that could be 

246
00:14:55,960 --> 00:14:59,160
helpful. 
If my boss emails me, send me a 

247
00:14:59,160 --> 00:15:02,880
text because sometimes your 
e-mail alerts don't go through 

248
00:15:02,880 --> 00:15:06,600
or sometime it's after office 
hours, but your boss might need 

249
00:15:06,600 --> 00:15:08,360
to get a hold of you. 
You don't have your e-mail app 

250
00:15:08,360 --> 00:15:10,760
on your phone. 
Your your work e-mail app on 

251
00:15:10,760 --> 00:15:12,840
your phone. 
So you know you do you get a 

252
00:15:12,840 --> 00:15:16,360
text from your boss. 
It's 8:00 at night when you're 

253
00:15:16,360 --> 00:15:18,560
putting your kid to bed. 
That would be the worst app 

254
00:15:18,560 --> 00:15:19,680
ever. 
Don't do that. 

255
00:15:19,840 --> 00:15:25,280
I, I'll tell you this, do not 
have your work and your home 

256
00:15:25,560 --> 00:15:27,640
like and your personal phone as 
one phone. 

257
00:15:27,640 --> 00:15:29,920
That's the worst idea ever. 
Unless of course, you run your 

258
00:15:29,920 --> 00:15:31,200
own business and then you got to
do it. 

259
00:15:34,040 --> 00:15:36,920
So eventually you can just run 
it on autopilot and does all the

260
00:15:36,920 --> 00:15:41,800
stuff for you. 
You can have your agent do so 

261
00:15:41,800 --> 00:15:48,040
many different things and you 
can check out different 

262
00:15:48,040 --> 00:15:49,440
services. 
Like I said, I don't want to 

263
00:15:49,440 --> 00:15:52,480
recommend anything because I 
don't really use them myself 

264
00:15:52,480 --> 00:15:56,400
personally. 
I built them, so I build them 

265
00:15:56,400 --> 00:16:02,600
for my own personal use. 
So I hope that this has cleared 

266
00:16:02,600 --> 00:16:06,120
a few things up for you. 
It's like a helper, you know, if

267
00:16:06,120 --> 00:16:08,480
you, if you have an executive 
assistant, that's kind of what 

268
00:16:08,560 --> 00:16:14,320
an AI agent, agentic AI is all 
about. 

269
00:16:14,320 --> 00:16:18,200
It helps you do those tasks that
you just don't want to do. 

270
00:16:18,200 --> 00:16:23,240
It turns a chatty program into a
helper that completes tasks by 

271
00:16:23,800 --> 00:16:28,400
planning those steps using tools
that are safe and then checking 

272
00:16:28,400 --> 00:16:30,880
along the way so you can get the
results that you want. 

273
00:16:32,240 --> 00:16:36,160
You just got to give it a clear 
goal, a few rules and some 

274
00:16:36,720 --> 00:16:40,280
buttons that it can push. 
And you keep all the control 

275
00:16:40,280 --> 00:16:43,680
too. 
You get readable logs from it. 

276
00:16:44,600 --> 00:16:48,840
You measure success with finish 
rates, time to finish and 

277
00:16:49,120 --> 00:16:50,920
escalations. 
Like does it send it back to 

278
00:16:50,920 --> 00:16:54,800
you? 
So start small and confirm the 

279
00:16:54,800 --> 00:17:01,080
value and grow your agentic AI 
at your own pace. 

280
00:17:01,200 --> 00:17:05,200
You don't have to make a huge 
project, make something very 

281
00:17:05,200 --> 00:17:09,400
small that'll fix a task or fix 
a problem that you're having 

282
00:17:09,400 --> 00:17:14,720
right now. 
Hey, thank you so much for 

283
00:17:14,720 --> 00:17:16,880
listening today. 
I really do appreciate your 

284
00:17:16,880 --> 00:17:18,440
support. 
If you could take a second and 

285
00:17:18,440 --> 00:17:21,400
hit the subscribe or the follow 
button on whatever podcast 

286
00:17:21,400 --> 00:17:24,400
platform that you're listening 
on right now, I greatly 

287
00:17:24,400 --> 00:17:26,040
appreciate it. 
It helps out the show 

288
00:17:26,040 --> 00:17:28,680
tremendously and you'll never 
miss an episode. 

289
00:17:28,920 --> 00:17:32,280
And each episode is about 10 
minutes or less to get you 

290
00:17:32,280 --> 00:17:35,000
caught up quickly. 
And please, if you want to 

291
00:17:35,000 --> 00:17:41,120
support the show even more, go 
to patreon.com/stagezero and 

292
00:17:41,120 --> 00:17:43,560
please take care of yourselves 
and each other and I'll see you 

293
00:17:43,560 --> 00:17:44,080
tomorrow.