WEBVTT

1
00:00:00.120 --> 00:00:04.519
<v Speaker 1>Welcome to AI Assembled, the top AI news for small business.

2
00:00:05.040 --> 00:00:07.759
<v Speaker 1>I'm Chris daily and each week I bring you the

3
00:00:07.799 --> 00:00:13.160
<v Speaker 1>most important AI developments that can transform your business. Let's

4
00:00:13.199 --> 00:00:15.080
<v Speaker 1>dive into this week's top stories.

5
00:00:16.760 --> 00:00:19.480
<v Speaker 2>Washington is telling its allies they have to pick a

6
00:00:19.559 --> 00:00:24.320
<v Speaker 2>side in the artificial intelligence race with China. A State

7
00:00:24.399 --> 00:00:28.640
<v Speaker 2>Department framework called Pax Silica is now being presented to

8
00:00:28.719 --> 00:00:32.640
<v Speaker 2>dozens of US partners, pressing them to join American led

9
00:00:32.719 --> 00:00:37.679
<v Speaker 2>joint projects on semiconductors and the chips that power frontier AI,

10
00:00:38.439 --> 00:00:41.960
<v Speaker 2>or risk finding themselves on the outside of the US

11
00:00:41.960 --> 00:00:46.759
<v Speaker 2>technology ecosystem. A leaked draft of the agreement makes the

12
00:00:46.880 --> 00:00:51.960
<v Speaker 2>choice explicit. Countries that participate in the American AI partnership

13
00:00:52.359 --> 00:00:56.399
<v Speaker 2>are expected to align with US export controls aimed at

14
00:00:56.479 --> 00:00:59.960
<v Speaker 2>keeping advanced chips and chip making equipment out of sad

15
00:01:00.159 --> 00:01:04.040
<v Speaker 2>Chinese hands and to steer clear of a rival China

16
00:01:04.120 --> 00:01:08.920
<v Speaker 2>led initiative for small and mid sized businesses, this is

17
00:01:09.000 --> 00:01:14.120
<v Speaker 2>not abstract geopolitics. If your supply chain, your cloud providers,

18
00:01:14.439 --> 00:01:18.359
<v Speaker 2>or your customers touch Asia, the chipwar is about to

19
00:01:18.359 --> 00:01:23.280
<v Speaker 2>get more rigid. Expect tighter controls on where advanced computing

20
00:01:23.319 --> 00:01:28.120
<v Speaker 2>can be sold, more compliance questions from vendors and potentially

21
00:01:28.560 --> 00:01:35.040
<v Speaker 2>diverging technology stacks between the American and Chinese spheres. Businesses

22
00:01:35.079 --> 00:01:39.840
<v Speaker 2>that source hardware, sell internationally, or rely on global cloud

23
00:01:39.879 --> 00:01:43.840
<v Speaker 2>regions should start asking their providers a simple question, now,

24
00:01:44.640 --> 00:01:47.359
<v Speaker 2>which side of this line are your services built on?

25
00:01:49.120 --> 00:01:51.920
<v Speaker 2>Now to the Vatican, where the leader of the Catholic

26
00:01:52.000 --> 00:01:56.159
<v Speaker 2>Church just told lawmakers exactly where he thinks AI guard

27
00:01:56.200 --> 00:02:00.760
<v Speaker 2>rails should go. On August twenty first, Hope Leo the

28
00:02:00.799 --> 00:02:05.319
<v Speaker 2>fourteenth addressed the seventeenth Annual Meeting of the International Catholic

29
00:02:05.439 --> 00:02:09.960
<v Speaker 2>Legislator's Network, a gathering of Catholic lawmakers from around the world,

30
00:02:10.560 --> 00:02:15.120
<v Speaker 2>on the theme of human dignity and artificial intelligence. His

31
00:02:15.240 --> 00:02:19.599
<v Speaker 2>message was direct. AI must never be allowed to erode

32
00:02:19.639 --> 00:02:24.159
<v Speaker 2>the family, whether by reducing people and relationships to data

33
00:02:24.319 --> 00:02:29.560
<v Speaker 2>and simulations, by flooding young minds with content that distorts desire,

34
00:02:30.400 --> 00:02:35.560
<v Speaker 2>or through economic models that make family life precarious. He

35
00:02:35.639 --> 00:02:39.280
<v Speaker 2>told the legislators their job is not to hinder innovation,

36
00:02:39.879 --> 00:02:42.800
<v Speaker 2>but to guide it wisely, and he laid out what

37
00:02:42.919 --> 00:02:49.840
<v Speaker 2>sound legislation should do protect users from exploitation, preserve personal privacy,

38
00:02:50.639 --> 00:02:56.199
<v Speaker 2>insure transparency in how emerging technologies are used and guarantee

39
00:02:56.240 --> 00:03:02.080
<v Speaker 2>that they strengthen democratic institutions. Follows his first encyclical earlier

40
00:03:02.120 --> 00:03:05.800
<v Speaker 2>this year, which called for AI to be, in his words,

41
00:03:06.240 --> 00:03:10.159
<v Speaker 2>disarmed and to serve human dignity and the common good.

42
00:03:11.319 --> 00:03:13.759
<v Speaker 2>Why should a business owner care what the Pope says

43
00:03:13.800 --> 00:03:19.000
<v Speaker 2>about AI? Because moral framing from institutions with global reach

44
00:03:19.439 --> 00:03:23.800
<v Speaker 2>has a way of becoming regulatory framing. The principles he

45
00:03:23.879 --> 00:03:30.639
<v Speaker 2>named transparency, privacy, protection from exploitation are the same pillars

46
00:03:30.680 --> 00:03:34.159
<v Speaker 2>appearing in the EU AI Act and in state level

47
00:03:34.199 --> 00:03:38.159
<v Speaker 2>bills across America. When you build or buy AI tools,

48
00:03:38.560 --> 00:03:42.439
<v Speaker 2>these are the criteria your future regulators and your customers

49
00:03:42.759 --> 00:03:43.879
<v Speaker 2>will be judging them by.

50
00:03:44.039 --> 00:03:49.879
<v Speaker 3>AI literacy is officially becoming homework. State lawmakers across America

51
00:03:49.919 --> 00:03:53.400
<v Speaker 3>are moving fast to put artificial intelligence into the high

52
00:03:53.439 --> 00:03:58.439
<v Speaker 3>school curriculum. During this year's legislative session, seventy seven bills

53
00:03:58.479 --> 00:04:02.879
<v Speaker 3>across twenty seven states have been introduced addressing AI in

54
00:04:03.000 --> 00:04:07.400
<v Speaker 3>classroom instruction, and ten have already been signed into law.

55
00:04:08.159 --> 00:04:12.199
<v Speaker 3>Alabama now requires every student to complete an approved computer

56
00:04:12.319 --> 00:04:16.160
<v Speaker 3>science course that includes AI instruction in order to graduate.

57
00:04:16.839 --> 00:04:20.959
<v Speaker 3>Idaho has directed its education department to build a comprehensive

58
00:04:21.160 --> 00:04:28.399
<v Speaker 3>generative AI framework covering privacy, transparency, academic integrity, and literacy standards.

59
00:04:29.120 --> 00:04:34.480
<v Speaker 3>Maryland's Artificial Intelligence Ready Schools Act requires every district to

60
00:04:34.519 --> 00:04:39.560
<v Speaker 3>adopt AI policies and designate an AI coordinator, and UTAH

61
00:04:39.839 --> 00:04:43.519
<v Speaker 3>is establishing a required digital skills course for seventh and

62
00:04:43.720 --> 00:04:49.839
<v Speaker 3>eighth graders covering AI literacy alongside cybersecurity and digital privacy.

63
00:04:50.560 --> 00:04:54.240
<v Speaker 3>The driver behind much of this AI literacy will be

64
00:04:54.279 --> 00:04:57.160
<v Speaker 3>assessed for the first time on the twenty twenty nine

65
00:04:57.279 --> 00:05:02.120
<v Speaker 3>PISA Exam, the global test compared fifteen year olds across countries.

66
00:05:02.879 --> 00:05:07.040
<v Speaker 3>For small businesses, this matters in two ways. First, the

67
00:05:07.199 --> 00:05:10.319
<v Speaker 3>entry level employees you hire three to five years from

68
00:05:10.360 --> 00:05:14.680
<v Speaker 3>now will arrive AI literate by law, not by chance,

69
00:05:15.279 --> 00:05:17.560
<v Speaker 3>and that changes what you can expect from them on

70
00:05:17.680 --> 00:05:22.199
<v Speaker 3>day one. Second, if you sell products or services to schools,

71
00:05:22.680 --> 00:05:29.040
<v Speaker 3>AI curriculum requirements are creating a brand new buying category overnight. Next,

72
00:05:29.439 --> 00:05:33.759
<v Speaker 3>the technology plumbing that is quietly making agentic AI reeal.

73
00:05:34.519 --> 00:05:38.920
<v Speaker 3>Agentic AI means AI systems that do not just answer questions,

74
00:05:39.319 --> 00:05:43.439
<v Speaker 3>but take actions on your behalf, reading your data, updating

75
00:05:43.439 --> 00:05:48.439
<v Speaker 3>your systems, completing tasks. The breakout standard making that possible

76
00:05:48.600 --> 00:05:54.040
<v Speaker 3>is called the Model Context Protocol or MCP. Think of

77
00:05:54.040 --> 00:05:58.800
<v Speaker 3>it as the USBC of AI, a universal connector that

78
00:05:58.879 --> 00:06:03.040
<v Speaker 3>lets AIAH plug into your existing tools and data sources

79
00:06:03.399 --> 00:06:08.639
<v Speaker 3>without expensive custom integration work. Adoption has been staggering, with

80
00:06:08.839 --> 00:06:14.439
<v Speaker 3>MCP's software development kits hitting ninety seven million downloads per month.

81
00:06:14.560 --> 00:06:18.120
<v Speaker 3>This spring up from roughly two million at launch less

82
00:06:18.160 --> 00:06:21.399
<v Speaker 3>than two years earlier, and it is now suddenly on

83
00:06:21.639 --> 00:06:27.879
<v Speaker 3>executive agendas everywhere. But there is a catch. Security researchers

84
00:06:27.959 --> 00:06:33.120
<v Speaker 3>warn that MCP connectors can be over permissioned, untrusted connectors

85
00:06:33.160 --> 00:06:37.079
<v Speaker 3>can leak data or enable attacks, and governance is lagging

86
00:06:37.199 --> 00:06:41.680
<v Speaker 3>far behind adoption. Fewer than four percent of security conference

87
00:06:41.720 --> 00:06:46.720
<v Speaker 3>submissions about MCP focus on opportunity, the rest focus on

88
00:06:46.879 --> 00:06:50.639
<v Speaker 3>risk and a personal note here from our host, he

89
00:06:50.800 --> 00:06:55.199
<v Speaker 3>is putting this exact idea into practice, building a company

90
00:06:55.240 --> 00:06:59.240
<v Speaker 3>knowledge base that multiple AI systems can access and understand,

91
00:07:00.000 --> 00:07:04.399
<v Speaker 3>one shared, structured source of truth that any AI assistant

92
00:07:04.439 --> 00:07:08.879
<v Speaker 3>can plug into, rather than knowledge trapped in scattered documents.

93
00:07:09.759 --> 00:07:14.680
<v Speaker 3>That is precisely the architecture. This new standard enables the

94
00:07:14.759 --> 00:07:19.120
<v Speaker 3>takeaway for small businesses the era of AI that can

95
00:07:19.240 --> 00:07:22.959
<v Speaker 3>actually do things inside. Your company has arrived, and the

96
00:07:23.000 --> 00:07:25.879
<v Speaker 3>winners will be the ones who get their knowledge organized

97
00:07:26.199 --> 00:07:30.240
<v Speaker 3>and their permissions governed before they plug agents in. New

98
00:07:30.319 --> 00:07:35.199
<v Speaker 3>data confirms a hard truth about AI adoption. Most pilots fail,

99
00:07:35.800 --> 00:07:39.079
<v Speaker 3>and the way you start largely determines whether yours does.

100
00:07:40.199 --> 00:07:44.360
<v Speaker 3>Research now circulating among chief financial officers finds that roughly

101
00:07:44.519 --> 00:07:49.079
<v Speaker 3>ninety five percent of enterprise generative AI pilots deliver no

102
00:07:49.360 --> 00:07:53.759
<v Speaker 3>measurable return, and separate analysis has long put the failure

103
00:07:53.839 --> 00:07:59.079
<v Speaker 3>rate for AI projects overall above eighty percent, about twice

104
00:07:59.120 --> 00:08:03.800
<v Speaker 3>the failure rate of traditional IT projects. The pattern is

105
00:08:03.879 --> 00:08:09.360
<v Speaker 3>remarkably consistent. Pilots launched without structure, without a defined problem,

106
00:08:09.759 --> 00:08:14.160
<v Speaker 3>without integration into real workflows, and without an owner who

107
00:08:14.199 --> 00:08:18.399
<v Speaker 3>is accountable for results are the ones that die. The

108
00:08:18.439 --> 00:08:23.279
<v Speaker 3>research points to flawed enterprise integration, not model quality, as

109
00:08:23.319 --> 00:08:28.680
<v Speaker 3>the core issue. Generic tools impress individuals but stall inside

110
00:08:28.759 --> 00:08:32.480
<v Speaker 3>organizations because they do not learn from or adapt to

111
00:08:32.840 --> 00:08:35.879
<v Speaker 3>actual workflows. What works.

112
00:08:36.600 --> 00:08:41.320
<v Speaker 4>Buying specialized tools and partnering with vendors succeeds about sixty

113
00:08:41.399 --> 00:08:44.840
<v Speaker 4>seven percent of the time, while building your own internally

114
00:08:45.080 --> 00:08:49.080
<v Speaker 4>succeeds only about a third as often, and the biggest

115
00:08:49.159 --> 00:08:53.559
<v Speaker 4>return on investment is not where most budgets go. More

116
00:08:53.600 --> 00:08:57.120
<v Speaker 4>than half of generative AI budgets flow to sales and

117
00:08:57.200 --> 00:09:01.159
<v Speaker 4>marketing tools, Yet the strongest return are showing up in

118
00:09:01.320 --> 00:09:07.960
<v Speaker 4>back office automation, eliminating outsourced processes, cutting agency costs, and

119
00:09:08.039 --> 00:09:13.279
<v Speaker 4>streamlining operations For a small business. The lesson is blunt,

120
00:09:14.120 --> 00:09:18.240
<v Speaker 4>pick one pain point, buy rather than build, put a

121
00:09:18.320 --> 00:09:22.919
<v Speaker 4>line manager in charge, and measure the result. The eighty

122
00:09:22.960 --> 00:09:26.519
<v Speaker 4>percent who fail are overwhelmingly the ones who started with

123
00:09:26.600 --> 00:09:31.399
<v Speaker 4>a vague ambition instead of a structured plan. Finally, the

124
00:09:31.440 --> 00:09:35.879
<v Speaker 4>staggering price tag of the AI arms race. This week

125
00:09:35.960 --> 00:09:40.840
<v Speaker 4>brought two massive infrastructure deals that underscore just how expensive

126
00:09:41.000 --> 00:09:45.799
<v Speaker 4>powering frontier AI has become. Google signed an agreement with

127
00:09:45.960 --> 00:09:50.679
<v Speaker 4>chip maker Marvel Technology to co develop custom AI chips,

128
00:09:51.080 --> 00:09:54.320
<v Speaker 4>a deal that extends to twenty thirty three and grants

129
00:09:54.360 --> 00:09:57.200
<v Speaker 4>Google a warrant to buy up to twelve point two

130
00:09:57.360 --> 00:10:02.279
<v Speaker 4>billion dollars of Marvell's stock nearly fifty nine million shares,

131
00:10:02.720 --> 00:10:06.000
<v Speaker 4>with the tie up potentially generating up to one hundred

132
00:10:06.039 --> 00:10:11.080
<v Speaker 4>and twenty billion dollars in custom chip sales. Days earlier,

133
00:10:11.320 --> 00:10:14.440
<v Speaker 4>in Nvidia agreed to guarantee up to one hundred and

134
00:10:14.600 --> 00:10:18.840
<v Speaker 4>five billion dollars in financing to help open ai lease

135
00:10:18.879 --> 00:10:22.759
<v Speaker 4>a sprawling data center being built in Ohio, while also

136
00:10:22.840 --> 00:10:27.600
<v Speaker 4>investing one point five billion dollars directly in sb Energy,

137
00:10:28.080 --> 00:10:32.279
<v Speaker 4>the soft bank link developer behind the project. That is

138
00:10:32.360 --> 00:10:37.879
<v Speaker 4>one company in Vidia effectively underwriting its own largest customer.

139
00:10:38.919 --> 00:10:42.200
<v Speaker 4>Two things for small businesses to take from this. First,

140
00:10:42.840 --> 00:10:46.960
<v Speaker 4>compute costs are not coming down anytime soon. When the

141
00:10:46.960 --> 00:10:49.879
<v Speaker 4>biggest players on Earth are locking in chip supply and

142
00:10:50.000 --> 00:10:53.759
<v Speaker 4>data center capacity a decade out, they're telling you demand

143
00:10:53.799 --> 00:10:57.960
<v Speaker 4>will keep outpacing supply, so budget accordingly, and watch your

144
00:10:57.960 --> 00:11:03.720
<v Speaker 4>AI subscription pricing. Second, circular deals like in Nvidia financing

145
00:11:03.799 --> 00:11:08.200
<v Speaker 4>OpenAI's infrastructure have some investors nervous about an AI bubble,

146
00:11:08.720 --> 00:11:12.360
<v Speaker 4>which means volatility in AI vendor pricing and even vendor

147
00:11:12.440 --> 00:11:16.840
<v Speaker 4>survival is a real risk. Diversify the AI tools your

148
00:11:16.879 --> 00:11:20.120
<v Speaker 4>business depends on, and make sure your data is portable

149
00:11:20.279 --> 00:11:23.799
<v Speaker 4>if one of them stumbles. And that is the week

150
00:11:23.840 --> 00:11:28.360
<v Speaker 4>in AI Assembled for your business. I'm James for all

151
00:11:28.399 --> 00:11:31.200
<v Speaker 4>of us here, thank you for listening, and we will

152
00:11:31.240 --> 00:11:32.120
<v Speaker 4>see you next week.

153
00:11:33.399 --> 00:11:36.759
<v Speaker 1>That's it for this week's AI Assembled. The top AI

154
00:11:36.879 --> 00:11:40.440
<v Speaker 1>news for small business. If you found this valuable, share

155
00:11:40.480 --> 00:11:42.879
<v Speaker 1>it with a fellow business owner. And if you want

156
00:11:42.879 --> 00:11:45.919
<v Speaker 1>to know whether AI tools can find and recommend your business,

157
00:11:46.200 --> 00:11:51.919
<v Speaker 1>get your free AI website optimization report at freshmediaworks dot com.

158
00:11:51.960 --> 00:11:54.679
<v Speaker 1>Slash ai ready until next week.

159
00:11:55.000 --> 00:11:56.799
<v Speaker 4>Keep innovating and keep it real