WEBVTT
1
00:00:04.624 --> 00:00:06.651
Hi, this is the Week in Review for September 14th.
2
00:01:13.146 --> 00:01:13.607
Mmm.
3
00:03:06.742 --> 00:03:13.389
This is Technically Creative's Week in Review, the news, ideas, and occasional bullshit from where technology and the creative industries are colliding.
4
00:03:13.910 --> 00:03:21.678
I'm Orlando Wood, and every Monday, I'm gonna catch you up on the stories I think actually matter, what they mean, and occasionally why everybody is losing their minds about the wrong thing.
5
00:03:22.159 --> 00:03:23.880
This is Week in Review, September 14th, 2026.
6
00:03:25.977 --> 00:03:31.019
There's a fight happening right now in advertising production that on the surface sounds incredibly niche.
7
00:03:31.459 --> 00:03:33.460
It involves the triple bid scenario.
8
00:03:33.800 --> 00:03:36.301
So you lucky thing you're about to hear about procurement.
9
00:03:36.501 --> 00:03:36.961
Buckle up.
10
00:03:37.482 --> 00:03:46.105
UPP, the massive advertising holding company, is trying to push more commercial production through its own organization rather than independent production companies.
11
00:03:46.445 --> 00:03:49.088
The Flashpoint is the traditional triple bid process.
12
00:03:49.408 --> 00:03:53.792
For anyone who hasn't been in that particular Thunderdome, the basic idea is simple.
13
00:03:54.173 --> 00:04:05.244
An agency wants to make an ad, three independent production companies compete for it, directors submit treatments, everybody prices the work, and the client gets genuinely competitive, creative, and financial options.
14
00:04:05.744 --> 00:04:08.207
I used to do this day in and day out in my job.
15
00:04:08.387 --> 00:04:10.608
And it does have huge benefits to the client.
16
00:04:10.768 --> 00:04:24.517
And crucially for production companies, it means that there's a reasonable expectation that if you're any good at all, you're gonna get some of those triple bid opportunities, giving you a pretty clear path to the opportunities that allow you to create a thriving business.
17
00:04:25.257 --> 00:04:39.709
Well, Campaign Magazine published a leaked WPP production document that explicitly states that their goal is to avoid traditional commercial triple bidding by positioning WPP production as the default production partner.
18
00:04:39.729 --> 00:04:49.897
If a client insists on three bids, the documents contemplate whether WPP production could theoretically provide all three internally, including potentially from different markets.
19
00:04:50.337 --> 00:04:52.358
Which raises kind of an obvious clash.
20
00:04:52.458 --> 00:04:58.922
I mean, like, when I get my house remodeled, I don't want all my bids to come from the shady guy who's my cheap option.
21
00:04:59.142 --> 00:05:03.504
Sure, I'll probably go with him eventually and regret it, but at least I want the choice.
22
00:05:04.285 --> 00:05:08.147
And that's where the UK Advertising Producers Association stepped in.
23
00:05:08.567 --> 00:05:16.472
Steve Davies, its CEO, publicly criticized the proposal, arguing that three bids from within the same company doesn't count as competition.
24
00:05:16.492 --> 00:05:16.892
Yeah.
25
00:05:17.232 --> 00:05:23.635
Which is incredibly fair, and I'm not entirely sure this one required a whole CEO to weigh in on it.
26
00:05:23.975 --> 00:05:27.037
A central St. Martin's grad with a hangover could have worked that out.
27
00:05:27.677 --> 00:05:32.479
And full disclosure, I used to sit on the board of the APA, and I think that what they do is incredibly important.
28
00:05:32.999 --> 00:05:40.082
WPP has obviously disputed the suggestion that clients are being misled and says its processes are transparent and client-led.
29
00:05:40.602 --> 00:05:49.205
And I don't think that the interesting part of this is deciding that WPP is some Bond villain sitting inside a giant production facility stroking a cat.
30
00:05:49.646 --> 00:05:53.027
Because honestly, I do understand exactly why they're doing it.
31
00:05:53.527 --> 00:05:56.188
WPP has 10,000 production people.
32
00:05:56.748 --> 00:05:58.409
What did we think they were going to do?
33
00:06:01.170 --> 00:06:14.161
Earlier this year, WPP folded Hogarth, its production offering, into WPP production, creating a global production operation of roughly 10,000 people in more than 40 cities.
34
00:06:14.201 --> 00:06:21.507
The thin end of the wedge happened years ago when agencies started slowly responding to social media by saying, okay—
35
00:06:22.628 --> 00:06:26.231
These jobs are too bitty and too small to go out to production companies.
36
00:06:26.511 --> 00:06:30.715
So we need an internal offering that can capture some of this lower level stuff.
37
00:06:31.175 --> 00:06:36.059
Well, that lower level stuff is now all of the social media assets that a client needs.
38
00:06:36.319 --> 00:06:43.625
And the TV campaign that was the largest spend has now become a more and more relegated part of the asset deliverables.
39
00:06:44.125 --> 00:06:50.792
So that thin end of the wedge has been growing into what's happening now, and it's going to continue.
40
00:06:50.852 --> 00:06:54.116
What interests me is why this is increasingly making sense to them.
41
00:06:54.236 --> 00:06:57.219
The numbers say that the old advertising machine is under pressure.
42
00:06:57.359 --> 00:07:02.545
In the first half of 2026, the jewel in their crown, WPP Media, declined 5.4%.
43
00:07:04.707 --> 00:07:06.707
WPP Creative declined 4.9%.
44
00:07:06.888 --> 00:07:13.269
And revenue-less pass-through costs across media, creative, and production fell 4.7%.
45
00:07:13.709 --> 00:07:14.769
What was the odd one out?
46
00:07:15.629 --> 00:07:18.070
WPP Production grew 1.6%.
47
00:07:18.710 --> 00:07:19.570
And there it is.
48
00:07:20.290 --> 00:07:31.232
If you're running WPP, I imagine there's a fairly short meeting where somebody points that slide and says, The bits selling traditional agency services are shrinking, but making the stuff is growing.
49
00:07:31.672 --> 00:07:33.653
Perhaps we should make more of the stuff.
50
00:07:34.633 --> 00:07:41.815
And this is where I think the independent production community is right to be worried, while WPP is also behaving completely rationally.
51
00:07:41.975 --> 00:07:44.775
Advertising used to operate in a more episodic world.
52
00:07:45.075 --> 00:07:48.696
You made the campaign, you made the commercial, maybe a few cutdowns.
53
00:07:49.016 --> 00:07:51.217
Everyone went to an obscenely lavish lunch.
54
00:07:51.297 --> 00:07:56.038
By the way, the lavishness of the lunch is part of the client's problem with the old system.
55
00:07:56.982 --> 00:08:12.271
Now a global advertiser needs TikTok, Instagram, YouTube, retail media, e-commerce, connected TV, six-second version, 15-second versions, vertical versions, square versions, translations, localizations, product variants, and whatever format was invented next.
56
00:08:12.431 --> 00:08:14.813
iPhone Duo just gave us a new one last week.
57
00:08:15.493 --> 00:08:17.055
There's no campaign anymore.
58
00:08:17.476 --> 00:08:21.101
There's a constantly on content supply chain.
59
00:08:21.281 --> 00:08:26.808
And clients don't want the cost of traditional bespoke production multiplied by 4,000 assets.
60
00:08:26.988 --> 00:08:30.173
And to be clear, that pressure predates generative AI.
61
00:08:30.633 --> 00:08:33.457
I was experiencing it when I was running Biscuit 10 years ago.
62
00:08:34.663 --> 00:08:37.666
AI simply makes the response more technically possible.
63
00:08:38.186 --> 00:08:45.732
Automated localization, generative adaptation, virtual production, synthetic assets, faster posts, and more work performed internally.
64
00:08:46.033 --> 00:08:51.257
That's why framing this as WPP versus independent production companies slightly misses the bigger story.
65
00:08:51.797 --> 00:08:54.440
The holding company model itself is being squeezed.
66
00:08:54.860 --> 00:08:57.983
Yes, they're huge and much bigger than independent production companies.
67
00:08:58.643 --> 00:09:01.226
but they're also increasingly fucked themselves.
68
00:09:01.967 --> 00:09:03.349
So let's learn more about that.
69
00:09:03.609 --> 00:09:10.878
Because Omnicom completed its acquisition of IPG late last year, creating an absolute monster of an advertising holding company.
70
00:09:11.278 --> 00:09:13.741
And it's done a few things that people object to as well.
71
00:09:14.402 --> 00:09:14.602
It's...
72
00:09:15.203 --> 00:09:18.486
Changing everybody's emails to OMC email addresses.
73
00:09:19.006 --> 00:09:21.188
It's creating a centralized production hub.
74
00:09:21.208 --> 00:09:22.629
So all of these things are happening.
75
00:09:22.669 --> 00:09:24.231
They're just messaging it differently.
76
00:09:24.451 --> 00:09:28.234
The merger has naturally been followed by restructuring and cost reduction.
77
00:09:28.634 --> 00:09:36.621
At exactly the same time, clients want more content, more data, more technology, and faster turnaround without volunteering to double the agency fee.
78
00:09:37.061 --> 00:09:41.865
This week gave us a perfect illustration of how fluid those client relationships have become.
79
00:09:42.045 --> 00:09:57.938
PepsiCo moved roughly $1.7 billion of global media, data, and technology business from Omnicom to Publicis after approximately 25 years with Omnicom, reportedly without the giant conventional global pitch circus we would normally expect for something of this scale.
80
00:09:58.138 --> 00:10:00.879
Publicis now has Pepsi, which creates a small issue.
81
00:10:01.199 --> 00:10:07.321
It also handles Coca-Cola's North American media business and had been participating in Coke's global media review.
82
00:10:08.001 --> 00:10:12.482
So Publicis is stepping away from that review and giving up the North American Coke business.
83
00:10:12.762 --> 00:10:15.123
Are you thinking that Omnicom might swoop into that?
84
00:10:15.923 --> 00:10:18.205
It looks like they're pretty much going to have to to fill the gap.
85
00:10:18.665 --> 00:10:21.347
The ad industry is not in a period of calm.
86
00:10:21.948 --> 00:10:24.710
And it makes WPP's behavior easier to understand.
87
00:10:25.130 --> 00:10:35.298
If client relationships are less permanent, service margins are under pressure, clients need exponentially more output, and technology increasingly allows production to be consolidated.
88
00:10:35.778 --> 00:10:39.161
Then owning more of the value chain starts to look awfully attractive.
89
00:10:39.281 --> 00:10:45.306
And technology, especially generative AI, increasingly allows production to be consolidated.
90
00:10:45.586 --> 00:10:48.268
allowing them to own more of the value chain.
91
00:10:48.948 --> 00:10:52.251
The problem, of course, is that everyone else already lives in that value chain.
92
00:10:52.571 --> 00:11:03.118
Directors, independent producers, editors, posthouses, VFX companies, and specialist shops built around the assumption that the agency orchestrates the process rather than becomes the process.
93
00:11:03.679 --> 00:11:05.640
And that, I think, is the actual fight.
94
00:11:05.780 --> 00:11:14.967
Not whether three internal WPP bids technically constitute a triple bid, but whether the traditional ecosystem of advertising production survives the transformation
95
00:11:15.367 --> 00:11:19.510
of agencies into vertically integrated content factories.
96
00:11:20.331 --> 00:11:22.373
I genuinely don't know the answer, but I know this.
97
00:11:22.753 --> 00:11:40.928
If you build a 10,000 person production company and your production business is growing while much of the rest of your company is shrinking, you probably aren't gonna say, wonderful, let's send the work somewhere else, especially as generative AI is getting better every day and has the potential to make you look like you're actually capable of doing this.
98
00:11:42.062 --> 00:11:47.644
And although generative AI didn't create this mess, it may be the thing that lets WPP exploit it.
99
00:11:47.864 --> 00:11:52.226
I've said this before, but digitization disrupted distribution first.
100
00:11:52.666 --> 00:11:56.608
It's why I'm able to put this out and get global distribution immediately.
101
00:11:57.383 --> 00:11:58.804
Now production is next.
102
00:11:59.224 --> 00:12:00.945
That's what Gen. AI is going to do.
103
00:12:01.005 --> 00:12:02.947
So I'm interested to see how this rolls out.
104
00:12:03.067 --> 00:12:10.272
I actually have Steve Davies and Matt Miller coming on the podcast soon to talk about this exact same thing, and I'm really excited about it.
105
00:12:10.792 --> 00:12:11.893
So on to other news.
106
00:12:12.433 --> 00:12:17.657
Adobe put AI inside Premiere and then crushed earnings afterwards.
107
00:12:18.592 --> 00:12:25.877
They made one of the week's most important creative AI announcements, and they gave us a pretty compelling piece of evidence that its approach might be working.
108
00:12:26.157 --> 00:12:40.406
Premiere 26.5 now lets editors generate video and sound effects inside the timeline, not in another website, not generate, export, download, discover you downloaded the wrong clip, import, swear, go back to the website and start again.
109
00:12:40.806 --> 00:12:48.251
You select a range in Premiere, describe what you want, choose a model, and then the generated material arrives as an editable clip in the sequence.
110
00:12:48.711 --> 00:12:55.760
You can also use footage already in the edit as a visual reference, and Adobe is not forcing editors to exclusively use Firefly.
111
00:12:56.160 --> 00:12:59.484
They support third-party models, and they can live inside the workflow too.
112
00:13:00.145 --> 00:13:03.509
This is much more interesting to me than another generative video demo.
113
00:13:03.529 --> 00:13:03.610
Wow.
114
00:13:04.090 --> 00:13:08.653
We've spent years watching increasingly impressive models make increasingly impressive shots.
115
00:13:09.013 --> 00:13:12.435
The next stage is making the technology disappear into the workflow.
116
00:13:12.535 --> 00:13:19.379
They just want to look at the cut and think, I just need three more seconds of rain hitting that window, generate it, adjust it, and keep working.
117
00:13:20.040 --> 00:13:22.041
That's how technology actually gets adopted.
118
00:13:22.361 --> 00:13:26.203
The miraculous new technology eventually becomes a button nobody thinks about.
119
00:13:26.604 --> 00:13:27.604
Do you remember dial-up?
120
00:13:28.004 --> 00:13:31.426
It was impressive and slow, but now, do you even think about Wi-Fi?
121
00:13:31.486 --> 00:13:32.207
Of course not.
122
00:13:32.527 --> 00:13:40.013
If Adobe becomes the professional layer through which creatives access whichever model is best, it can benefit from the model war without winning every battle.
123
00:13:40.213 --> 00:13:49.240
And then Adobe reported record quarterly revenue of $6.76 billion, beating expectations on earnings, and raised its full-year targets.
124
00:13:49.440 --> 00:13:54.124
Annual recurring revenue from its AI-first products has more than doubled year over year.
125
00:13:54.444 --> 00:13:59.408
It's an interesting proof point for a company whose stock tanked when Nano Banana launched—
126
00:13:59.848 --> 00:14:03.850
Proof that Adobe knows how creatives like to work, and that has value.
127
00:14:04.270 --> 00:14:10.513
Wall Street still found something to complain about because the Q4 revenue outlook was softer than analysts wanted, but whatever.
128
00:14:10.793 --> 00:14:13.294
From where I'm sitting, the interesting signal is simpler.
129
00:14:13.654 --> 00:14:18.957
Adobe is putting AI where professional creatives already work, and people appear to be paying for it.
130
00:14:19.197 --> 00:14:22.398
And then routers and ITN announced something this week that I love.
131
00:14:22.598 --> 00:14:26.280
ITN has more than 70 years of broadcast journalism sitting in its archive.
132
00:14:26.840 --> 00:14:39.666
Nelson Mandela being released, the Berlin Wall falling, royal weddings, political upheaval, wards, culture, millions of moments where somebody pointed an actual camera at something that actually happened.
133
00:14:40.107 --> 00:14:46.350
Well, starting in January, Reuters is going to make that archive substantially easier to search and license through Reuters Connect.
134
00:14:46.930 --> 00:14:48.671
AI will handle a lot of the boring work.
135
00:14:48.791 --> 00:14:50.572
This is one of my favorite uses of AI.
136
00:14:50.912 --> 00:14:53.534
Nobody needed to generate a fake Berlin Wall falling.
137
00:14:53.654 --> 00:14:55.015
We have the fucking footage.
138
00:14:55.475 --> 00:14:57.136
The problem is finding it.
139
00:14:57.616 --> 00:15:07.022
Somewhere inside of gigantic media archives around the world are extraordinary images, interviews, and moments that barely exist commercially because nobody can locate them officially.
140
00:15:07.202 --> 00:15:11.785
Suddenly, something that spent 50 years sitting on tape becomes economically alive.
141
00:15:12.306 --> 00:15:14.027
That's not replacing human creativity.
142
00:15:14.367 --> 00:15:17.049
That's giving old human creativity a better search engine.
143
00:15:17.329 --> 00:15:18.830
So more of this, please.
144
00:15:19.886 --> 00:15:24.690
And in other news, AI is looking for a pretty substantial credit rating.
145
00:15:25.270 --> 00:15:32.276
OpenAI and Anthropic are talking to ratings agencies about obtaining investment-grade credit ratings after their IPOs.
146
00:15:32.396 --> 00:15:37.320
Investment-grade status would give the companies cheaper access to the enormous corporate bond market.
147
00:15:37.460 --> 00:15:41.584
It's another step in AI's transformation from software story to industrial story.
148
00:15:42.144 --> 00:15:46.308
But neither OpenAI nor Anthropic currently generates positive free cash flow.
149
00:15:46.748 --> 00:15:52.473
Yet they're discussing the kinds of financing structures we associate with giant industrial businesses.
150
00:15:53.133 --> 00:15:54.094
Safe businesses.
151
00:15:54.534 --> 00:15:55.715
Profitable businesses.
152
00:15:55.916 --> 00:16:00.660
And that access only comes usually with a certain amount of safety and provability.
153
00:16:01.300 --> 00:16:03.162
And the ratings agencies are what give it.
154
00:16:03.763 --> 00:16:05.926
So they're asking for it ahead of time.
155
00:16:06.126 --> 00:16:09.610
And this is something we should all be reasonably scared about.
156
00:16:09.910 --> 00:16:12.513
We don't know if they're going to get it, but they're asking for it.
157
00:16:12.653 --> 00:16:16.778
So they're basically asking pension funds to bet on the AI bubble.
158
00:16:16.979 --> 00:16:19.141
And that brings us to my next question is –
159
00:16:20.182 --> 00:16:21.603
Is AI going to kill us all?
160
00:16:22.263 --> 00:16:24.105
The answer is not no.
161
00:16:24.625 --> 00:16:26.226
It's a resounding possibly.
162
00:16:26.866 --> 00:16:37.353
Anthropic researcher Jacob Coxon resigned this week and publicly warned that people building frontier AI genuinely believe the technology could pose an existential threat to humanity.
163
00:16:37.913 --> 00:16:47.480
His main post went properly nuclear online, passing 100 million views, and Axios reported that he left about two months before his anthropic equity would have vested.
164
00:16:47.981 --> 00:16:54.246
Walking away from meaningful equity doesn't prove the fear is correct, but it makes the warning harder to dismiss as cheap attention-seeking.
165
00:16:54.966 --> 00:16:58.091
Other people at Anthropic have expressed similarly alarming views.
166
00:16:58.271 --> 00:17:05.982
Alignment researcher Evan Hubbinger has put the probability of AI causing human extinction within the next decade at greater than 10%.
167
00:17:06.643 --> 00:17:07.785
Which do I say, 10%!
168
00:17:09.773 --> 00:17:14.977
I don't, I won't get on a plane if somebody tells me there's a 10% chance there isn't Diet Coke on the plane.
169
00:17:15.558 --> 00:17:20.201
So I don't want to wave this away, but these are very smart people working very close to the technology.
170
00:17:20.241 --> 00:17:24.585
And when people with that level of access say they're genuinely frightened, we should listen to them.
171
00:17:25.085 --> 00:17:27.449
But historical context also matters.
172
00:17:27.769 --> 00:17:38.905
When OpenAI announced GPT-2 in 2029, it withheld the full model because of concerns around malicious use, then released increasingly capable versions before eventually releasing the whole thing.
173
00:17:39.105 --> 00:17:40.668
GPT-2 today...
174
00:17:40.948 --> 00:17:43.431
feels about as threatening as an ambitious toaster.
175
00:17:43.672 --> 00:17:45.514
None of this means today's warnings are fake.
176
00:17:45.714 --> 00:17:55.567
And to be honest, the thing that gave me more pause was that Dario Amadei wrote on his blog that we should pause or slow down the rate of progress of these models.
177
00:17:56.028 --> 00:17:56.869
His fear was...
178
00:17:57.710 --> 00:18:03.192
A really highly capable new agent could create bot swarms that could literally take down the internet.
179
00:18:03.272 --> 00:18:04.792
And where would we be without the internet?
180
00:18:04.992 --> 00:18:12.174
What really gave me pause was that Sam Altman and Elon Musk all agreed with him.
181
00:18:12.954 --> 00:18:15.595
Elon Musk tweeted, Dario's right.
182
00:18:15.955 --> 00:18:22.937
The notion of modern CEOs in tech all calling to slow down...
183
00:18:24.265 --> 00:18:32.895
is either incredibly self-serving because they're trying to save on the R&D costs, or it is actually a harbinger of something.
184
00:18:33.135 --> 00:18:33.956
I don't know what.
185
00:18:34.749 --> 00:18:38.272
But we can't slow down too much because China's doing this too.
186
00:18:38.533 --> 00:18:51.445
But when three of the biggest guys in tech, two of whom we know are pretty cavalier, are all calling for a slowdown in lines with what Bernie Sanders is suggesting, that's a moment we should all look at.
187
00:18:51.665 --> 00:18:54.127
And this is a perfect moment for government to step in.
188
00:18:54.227 --> 00:18:54.868
So I don't know.
189
00:18:54.888 --> 00:18:56.650
I don't know what to do about that news.
190
00:18:57.310 --> 00:19:06.078
I think it's great, but these CEOs, especially when they go public, they're not really in control of things like safety and governance anymore.
191
00:19:06.319 --> 00:19:11.463
The responsibility to deliver shareholder value becomes their absolute North Star.
192
00:19:12.004 --> 00:19:14.186
So this does seem like a job for the government.
193
00:19:14.306 --> 00:19:16.688
And if the people running the companies at least accept that—
194
00:19:17.469 --> 00:19:19.690
that might be a really good step in the right direction.
195
00:19:19.790 --> 00:19:29.436
But it is interesting that after dooming and glooming every single time there's a model release, almost as a marketing strategy, it's hard to know when to take these guys seriously.
196
00:19:29.996 --> 00:19:38.060
Because after so many years of this model maybe too powerful to release, I think the public is entitled to increasingly specific answers.
197
00:19:38.521 --> 00:19:39.581
What can it actually do?
198
00:19:39.681 --> 00:19:40.822
What happened in testing?
199
00:19:40.922 --> 00:19:41.582
What failed?
200
00:19:41.943 --> 00:19:43.003
What safeguards worked?
201
00:19:43.103 --> 00:19:44.164
What safeguard didn't?
202
00:19:44.744 --> 00:19:45.825
What safeguards didn't?
203
00:19:46.583 --> 00:19:51.687
I mean, show us, because trust us, it's terrifying is not a safety framework.
204
00:19:51.767 --> 00:19:55.169
I think that what connects these stories is that the contracts are changing.
205
00:19:55.610 --> 00:19:57.051
And that's why this feels weird.
206
00:19:57.731 --> 00:19:59.913
Technology doesn't just change what we can do.
207
00:20:00.033 --> 00:20:07.499
It changes the little agreements we make about how the world is supposed to work, the social and corporate contracts we have with one another.
208
00:20:08.019 --> 00:20:08.939
Advertising had one.
209
00:20:09.479 --> 00:20:11.040
Agencies did agency things.
210
00:20:11.440 --> 00:20:13.081
Production companies made the films.
211
00:20:13.521 --> 00:20:19.642
Great directors and producers went to production companies because that was where the craft, the mentorship, and the opportunity lived.
212
00:20:20.043 --> 00:20:26.064
But the next generation is coming out of universities right now, and most of them are not protecting an old career model.
213
00:20:26.504 --> 00:20:27.585
They want to make things.
214
00:20:27.865 --> 00:20:28.705
They want to get better.
215
00:20:28.745 --> 00:20:31.346
They want to build a body of work, and they want to prove themselves.
216
00:20:31.586 --> 00:20:31.906
Right.
217
00:20:31.966 --> 00:20:32.807
They want a shot.
218
00:20:33.147 --> 00:20:41.433
They want reps. And if those opportunities increasingly sit inside agencies or on social media, then that's where they'll go.
219
00:20:42.014 --> 00:20:42.874
They'll grow there.
220
00:20:42.934 --> 00:20:44.576
Their skill sets will develop there.
221
00:20:45.116 --> 00:20:47.158
And over time, the center of gravity shifts.
222
00:20:47.818 --> 00:20:49.699
Talent flows towards opportunity.
223
00:20:49.979 --> 00:20:51.859
That's why this moment feels so strange.
224
00:20:52.099 --> 00:20:54.180
Dial-up became the internet over decades.
225
00:20:54.520 --> 00:20:56.960
Social media rewrote the culture much faster.
226
00:20:57.380 --> 00:21:04.862
Generative AI feels like it's renegotiating authorship, truth, expertise, and work within seven years.
227
00:21:05.282 --> 00:21:08.083
And we're still arguing about the social media revolution.
228
00:21:08.283 --> 00:21:10.984
Technology is moving faster than the conventions around it.
229
00:21:11.564 --> 00:21:13.425
And eventually we'll write new rules.
230
00:21:13.485 --> 00:21:16.287
Then somebody will invent something else and fuck them up again.
231
00:21:16.327 --> 00:21:19.049
But I think we have to get faster at writing the new rules.
232
00:21:19.369 --> 00:21:24.252
Because we've all bought into a society where corporations have nearly unchecked power.
233
00:21:24.312 --> 00:21:26.153
We can't be surprised when they use it.
234
00:21:26.493 --> 00:21:29.815
So on that shiny note, thank you for listening this week.
235
00:21:29.955 --> 00:21:34.098
Tomorrow's interview episode will be with Toby Walsh of Made by Humans and Imagine This Studio.
236
00:21:34.378 --> 00:21:38.660
Doing some amazing AI stuff in the production company space.
237
00:21:38.881 --> 00:21:40.842
And crucially not doing it inside an agency.
238
00:21:41.302 --> 00:21:44.367
So anyway, thank you for listening, and we'll see you next week.
1
00:00:04.624 --> 00:00:06.651
Hi, this is the Week in Review for September 14th.
2
00:01:13.146 --> 00:01:13.607
Mmm.
3
00:03:06.742 --> 00:03:13.389
This is Technically Creative's Week in Review, the news, ideas, and occasional bullshit from where technology and the creative industries are colliding.
4
00:03:13.910 --> 00:03:21.678
I'm Orlando Wood, and every Monday, I'm gonna catch you up on the stories I think actually matter, what they mean, and occasionally why everybody is losing their minds about the wrong thing.
5
00:03:22.159 --> 00:03:23.880
This is Week in Review, September 14th, 2026.
6
00:03:25.977 --> 00:03:31.019
There's a fight happening right now in advertising production that on the surface sounds incredibly niche.
7
00:03:31.459 --> 00:03:33.460
It involves the triple bid scenario.
8
00:03:33.800 --> 00:03:36.301
So you lucky thing you're about to hear about procurement.
9
00:03:36.501 --> 00:03:36.961
Buckle up.
10
00:03:37.482 --> 00:03:46.105
UPP, the massive advertising holding company, is trying to push more commercial production through its own organization rather than independent production companies.
11
00:03:46.445 --> 00:03:49.088
The Flashpoint is the traditional triple bid process.
12
00:03:49.408 --> 00:03:53.792
For anyone who hasn't been in that particular Thunderdome, the basic idea is simple.
13
00:03:54.173 --> 00:04:05.244
An agency wants to make an ad, three independent production companies compete for it, directors submit treatments, everybody prices the work, and the client gets genuinely competitive, creative, and financial options.
14
00:04:05.744 --> 00:04:08.207
I used to do this day in and day out in my job.
15
00:04:08.387 --> 00:04:10.608
And it does have huge benefits to the client.
16
00:04:10.768 --> 00:04:24.517
And crucially for production companies, it means that there's a reasonable expectation that if you're any good at all, you're gonna get some of those triple bid opportunities, giving you a pretty clear path to the opportunities that allow you to create a thriving business.
17
00:04:25.257 --> 00:04:39.709
Well, Campaign Magazine published a leaked WPP production document that explicitly states that their goal is to avoid traditional commercial triple bidding by positioning WPP production as the default production partner.
18
00:04:39.729 --> 00:04:49.897
If a client insists on three bids, the documents contemplate whether WPP production could theoretically provide all three internally, including potentially from different markets.
19
00:04:50.337 --> 00:04:52.358
Which raises kind of an obvious clash.
20
00:04:52.458 --> 00:04:58.922
I mean, like, when I get my house remodeled, I don't want all my bids to come from the shady guy who's my cheap option.
21
00:04:59.142 --> 00:05:03.504
Sure, I'll probably go with him eventually and regret it, but at least I want the choice.
22
00:05:04.285 --> 00:05:08.147
And that's where the UK Advertising Producers Association stepped in.
23
00:05:08.567 --> 00:05:16.472
Steve Davies, its CEO, publicly criticized the proposal, arguing that three bids from within the same company doesn't count as competition.
24
00:05:16.492 --> 00:05:16.892
Yeah.
25
00:05:17.232 --> 00:05:23.635
Which is incredibly fair, and I'm not entirely sure this one required a whole CEO to weigh in on it.
26
00:05:23.975 --> 00:05:27.037
A central St. Martin's grad with a hangover could have worked that out.
27
00:05:27.677 --> 00:05:32.479
And full disclosure, I used to sit on the board of the APA, and I think that what they do is incredibly important.
28
00:05:32.999 --> 00:05:40.082
WPP has obviously disputed the suggestion that clients are being misled and says its processes are transparent and client-led.
29
00:05:40.602 --> 00:05:49.205
And I don't think that the interesting part of this is deciding that WPP is some Bond villain sitting inside a giant production facility stroking a cat.
30
00:05:49.646 --> 00:05:53.027
Because honestly, I do understand exactly why they're doing it.
31
00:05:53.527 --> 00:05:56.188
WPP has 10,000 production people.
32
00:05:56.748 --> 00:05:58.409
What did we think they were going to do?
33
00:06:01.170 --> 00:06:14.161
Earlier this year, WPP folded Hogarth, its production offering, into WPP production, creating a global production operation of roughly 10,000 people in more than 40 cities.
34
00:06:14.201 --> 00:06:21.507
The thin end of the wedge happened years ago when agencies started slowly responding to social media by saying, okay—
35
00:06:22.628 --> 00:06:26.231
These jobs are too bitty and too small to go out to production companies.
36
00:06:26.511 --> 00:06:30.715
So we need an internal offering that can capture some of this lower level stuff.
37
00:06:31.175 --> 00:06:36.059
Well, that lower level stuff is now all of the social media assets that a client needs.
38
00:06:36.319 --> 00:06:43.625
And the TV campaign that was the largest spend has now become a more and more relegated part of the asset deliverables.
39
00:06:44.125 --> 00:06:50.792
So that thin end of the wedge has been growing into what's happening now, and it's going to continue.
40
00:06:50.852 --> 00:06:54.116
What interests me is why this is increasingly making sense to them.
41
00:06:54.236 --> 00:06:57.219
The numbers say that the old advertising machine is under pressure.
42
00:06:57.359 --> 00:07:02.545
In the first half of 2026, the jewel in their crown, WPP Media, declined 5.4%.
43
00:07:04.707 --> 00:07:06.707
WPP Creative declined 4.9%.
44
00:07:06.888 --> 00:07:13.269
And revenue-less pass-through costs across media, creative, and production fell 4.7%.
45
00:07:13.709 --> 00:07:14.769
What was the odd one out?
46
00:07:15.629 --> 00:07:18.070
WPP Production grew 1.6%.
47
00:07:18.710 --> 00:07:19.570
And there it is.
48
00:07:20.290 --> 00:07:31.232
If you're running WPP, I imagine there's a fairly short meeting where somebody points that slide and says, The bits selling traditional agency services are shrinking, but making the stuff is growing.
49
00:07:31.672 --> 00:07:33.653
Perhaps we should make more of the stuff.
50
00:07:34.633 --> 00:07:41.815
And this is where I think the independent production community is right to be worried, while WPP is also behaving completely rationally.
51
00:07:41.975 --> 00:07:44.775
Advertising used to operate in a more episodic world.
52
00:07:45.075 --> 00:07:48.696
You made the campaign, you made the commercial, maybe a few cutdowns.
53
00:07:49.016 --> 00:07:51.217
Everyone went to an obscenely lavish lunch.
54
00:07:51.297 --> 00:07:56.038
By the way, the lavishness of the lunch is part of the client's problem with the old system.
55
00:07:56.982 --> 00:08:12.271
Now a global advertiser needs TikTok, Instagram, YouTube, retail media, e-commerce, connected TV, six-second version, 15-second versions, vertical versions, square versions, translations, localizations, product variants, and whatever format was invented next.
56
00:08:12.431 --> 00:08:14.813
iPhone Duo just gave us a new one last week.
57
00:08:15.493 --> 00:08:17.055
There's no campaign anymore.
58
00:08:17.476 --> 00:08:21.101
There's a constantly on content supply chain.
59
00:08:21.281 --> 00:08:26.808
And clients don't want the cost of traditional bespoke production multiplied by 4,000 assets.
60
00:08:26.988 --> 00:08:30.173
And to be clear, that pressure predates generative AI.
61
00:08:30.633 --> 00:08:33.457
I was experiencing it when I was running Biscuit 10 years ago.
62
00:08:34.663 --> 00:08:37.666
AI simply makes the response more technically possible.
63
00:08:38.186 --> 00:08:45.732
Automated localization, generative adaptation, virtual production, synthetic assets, faster posts, and more work performed internally.
64
00:08:46.033 --> 00:08:51.257
That's why framing this as WPP versus independent production companies slightly misses the bigger story.
65
00:08:51.797 --> 00:08:54.440
The holding company model itself is being squeezed.
66
00:08:54.860 --> 00:08:57.983
Yes, they're huge and much bigger than independent production companies.
67
00:08:58.643 --> 00:09:01.226
but they're also increasingly fucked themselves.
68
00:09:01.967 --> 00:09:03.349
So let's learn more about that.
69
00:09:03.609 --> 00:09:10.878
Because Omnicom completed its acquisition of IPG late last year, creating an absolute monster of an advertising holding company.
70
00:09:11.278 --> 00:09:13.741
And it's done a few things that people object to as well.
71
00:09:14.402 --> 00:09:14.602
It's...
72
00:09:15.203 --> 00:09:18.486
Changing everybody's emails to OMC email addresses.
73
00:09:19.006 --> 00:09:21.188
It's creating a centralized production hub.
74
00:09:21.208 --> 00:09:22.629
So all of these things are happening.
75
00:09:22.669 --> 00:09:24.231
They're just messaging it differently.
76
00:09:24.451 --> 00:09:28.234
The merger has naturally been followed by restructuring and cost reduction.
77
00:09:28.634 --> 00:09:36.621
At exactly the same time, clients want more content, more data, more technology, and faster turnaround without volunteering to double the agency fee.
78
00:09:37.061 --> 00:09:41.865
This week gave us a perfect illustration of how fluid those client relationships have become.
79
00:09:42.045 --> 00:09:57.938
PepsiCo moved roughly $1.7 billion of global media, data, and technology business from Omnicom to Publicis after approximately 25 years with Omnicom, reportedly without the giant conventional global pitch circus we would normally expect for something of this scale.
80
00:09:58.138 --> 00:10:00.879
Publicis now has Pepsi, which creates a small issue.
81
00:10:01.199 --> 00:10:07.321
It also handles Coca-Cola's North American media business and had been participating in Coke's global media review.
82
00:10:08.001 --> 00:10:12.482
So Publicis is stepping away from that review and giving up the North American Coke business.
83
00:10:12.762 --> 00:10:15.123
Are you thinking that Omnicom might swoop into that?
84
00:10:15.923 --> 00:10:18.205
It looks like they're pretty much going to have to to fill the gap.
85
00:10:18.665 --> 00:10:21.347
The ad industry is not in a period of calm.
86
00:10:21.948 --> 00:10:24.710
And it makes WPP's behavior easier to understand.
87
00:10:25.130 --> 00:10:35.298
If client relationships are less permanent, service margins are under pressure, clients need exponentially more output, and technology increasingly allows production to be consolidated.
88
00:10:35.778 --> 00:10:39.161
Then owning more of the value chain starts to look awfully attractive.
89
00:10:39.281 --> 00:10:45.306
And technology, especially generative AI, increasingly allows production to be consolidated.
90
00:10:45.586 --> 00:10:48.268
allowing them to own more of the value chain.
91
00:10:48.948 --> 00:10:52.251
The problem, of course, is that everyone else already lives in that value chain.
92
00:10:52.571 --> 00:11:03.118
Directors, independent producers, editors, posthouses, VFX companies, and specialist shops built around the assumption that the agency orchestrates the process rather than becomes the process.
93
00:11:03.679 --> 00:11:05.640
And that, I think, is the actual fight.
94
00:11:05.780 --> 00:11:14.967
Not whether three internal WPP bids technically constitute a triple bid, but whether the traditional ecosystem of advertising production survives the transformation
95
00:11:15.367 --> 00:11:19.510
of agencies into vertically integrated content factories.
96
00:11:20.331 --> 00:11:22.373
I genuinely don't know the answer, but I know this.
97
00:11:22.753 --> 00:11:40.928
If you build a 10,000 person production company and your production business is growing while much of the rest of your company is shrinking, you probably aren't gonna say, wonderful, let's send the work somewhere else, especially as generative AI is getting better every day and has the potential to make you look like you're actually capable of doing this.
98
00:11:42.062 --> 00:11:47.644
And although generative AI didn't create this mess, it may be the thing that lets WPP exploit it.
99
00:11:47.864 --> 00:11:52.226
I've said this before, but digitization disrupted distribution first.
100
00:11:52.666 --> 00:11:56.608
It's why I'm able to put this out and get global distribution immediately.
101
00:11:57.383 --> 00:11:58.804
Now production is next.
102
00:11:59.224 --> 00:12:00.945
That's what Gen. AI is going to do.
103
00:12:01.005 --> 00:12:02.947
So I'm interested to see how this rolls out.
104
00:12:03.067 --> 00:12:10.272
I actually have Steve Davies and Matt Miller coming on the podcast soon to talk about this exact same thing, and I'm really excited about it.
105
00:12:10.792 --> 00:12:11.893
So on to other news.
106
00:12:12.433 --> 00:12:17.657
Adobe put AI inside Premiere and then crushed earnings afterwards.
107
00:12:18.592 --> 00:12:25.877
They made one of the week's most important creative AI announcements, and they gave us a pretty compelling piece of evidence that its approach might be working.
108
00:12:26.157 --> 00:12:40.406
Premiere 26.5 now lets editors generate video and sound effects inside the timeline, not in another website, not generate, export, download, discover you downloaded the wrong clip, import, swear, go back to the website and start again.
109
00:12:40.806 --> 00:12:48.251
You select a range in Premiere, describe what you want, choose a model, and then the generated material arrives as an editable clip in the sequence.
110
00:12:48.711 --> 00:12:55.760
You can also use footage already in the edit as a visual reference, and Adobe is not forcing editors to exclusively use Firefly.
111
00:12:56.160 --> 00:12:59.484
They support third-party models, and they can live inside the workflow too.
112
00:13:00.145 --> 00:13:03.509
This is much more interesting to me than another generative video demo.
113
00:13:03.529 --> 00:13:03.610
Wow.
114
00:13:04.090 --> 00:13:08.653
We've spent years watching increasingly impressive models make increasingly impressive shots.
115
00:13:09.013 --> 00:13:12.435
The next stage is making the technology disappear into the workflow.
116
00:13:12.535 --> 00:13:19.379
They just want to look at the cut and think, I just need three more seconds of rain hitting that window, generate it, adjust it, and keep working.
117
00:13:20.040 --> 00:13:22.041
That's how technology actually gets adopted.
118
00:13:22.361 --> 00:13:26.203
The miraculous new technology eventually becomes a button nobody thinks about.
119
00:13:26.604 --> 00:13:27.604
Do you remember dial-up?
120
00:13:28.004 --> 00:13:31.426
It was impressive and slow, but now, do you even think about Wi-Fi?
121
00:13:31.486 --> 00:13:32.207
Of course not.
122
00:13:32.527 --> 00:13:40.013
If Adobe becomes the professional layer through which creatives access whichever model is best, it can benefit from the model war without winning every battle.
123
00:13:40.213 --> 00:13:49.240
And then Adobe reported record quarterly revenue of $6.76 billion, beating expectations on earnings, and raised its full-year targets.
124
00:13:49.440 --> 00:13:54.124
Annual recurring revenue from its AI-first products has more than doubled year over year.
125
00:13:54.444 --> 00:13:59.408
It's an interesting proof point for a company whose stock tanked when Nano Banana launched—
126
00:13:59.848 --> 00:14:03.850
Proof that Adobe knows how creatives like to work, and that has value.
127
00:14:04.270 --> 00:14:10.513
Wall Street still found something to complain about because the Q4 revenue outlook was softer than analysts wanted, but whatever.
128
00:14:10.793 --> 00:14:13.294
From where I'm sitting, the interesting signal is simpler.
129
00:14:13.654 --> 00:14:18.957
Adobe is putting AI where professional creatives already work, and people appear to be paying for it.
130
00:14:19.197 --> 00:14:22.398
And then routers and ITN announced something this week that I love.
131
00:14:22.598 --> 00:14:26.280
ITN has more than 70 years of broadcast journalism sitting in its archive.
132
00:14:26.840 --> 00:14:39.666
Nelson Mandela being released, the Berlin Wall falling, royal weddings, political upheaval, wards, culture, millions of moments where somebody pointed an actual camera at something that actually happened.
133
00:14:40.107 --> 00:14:46.350
Well, starting in January, Reuters is going to make that archive substantially easier to search and license through Reuters Connect.
134
00:14:46.930 --> 00:14:48.671
AI will handle a lot of the boring work.
135
00:14:48.791 --> 00:14:50.572
This is one of my favorite uses of AI.
136
00:14:50.912 --> 00:14:53.534
Nobody needed to generate a fake Berlin Wall falling.
137
00:14:53.654 --> 00:14:55.015
We have the fucking footage.
138
00:14:55.475 --> 00:14:57.136
The problem is finding it.
139
00:14:57.616 --> 00:15:07.022
Somewhere inside of gigantic media archives around the world are extraordinary images, interviews, and moments that barely exist commercially because nobody can locate them officially.
140
00:15:07.202 --> 00:15:11.785
Suddenly, something that spent 50 years sitting on tape becomes economically alive.
141
00:15:12.306 --> 00:15:14.027
That's not replacing human creativity.
142
00:15:14.367 --> 00:15:17.049
That's giving old human creativity a better search engine.
143
00:15:17.329 --> 00:15:18.830
So more of this, please.
144
00:15:19.886 --> 00:15:24.690
And in other news, AI is looking for a pretty substantial credit rating.
145
00:15:25.270 --> 00:15:32.276
OpenAI and Anthropic are talking to ratings agencies about obtaining investment-grade credit ratings after their IPOs.
146
00:15:32.396 --> 00:15:37.320
Investment-grade status would give the companies cheaper access to the enormous corporate bond market.
147
00:15:37.460 --> 00:15:41.584
It's another step in AI's transformation from software story to industrial story.
148
00:15:42.144 --> 00:15:46.308
But neither OpenAI nor Anthropic currently generates positive free cash flow.
149
00:15:46.748 --> 00:15:52.473
Yet they're discussing the kinds of financing structures we associate with giant industrial businesses.
150
00:15:53.133 --> 00:15:54.094
Safe businesses.
151
00:15:54.534 --> 00:15:55.715
Profitable businesses.
152
00:15:55.916 --> 00:16:00.660
And that access only comes usually with a certain amount of safety and provability.
153
00:16:01.300 --> 00:16:03.162
And the ratings agencies are what give it.
154
00:16:03.763 --> 00:16:05.926
So they're asking for it ahead of time.
155
00:16:06.126 --> 00:16:09.610
And this is something we should all be reasonably scared about.
156
00:16:09.910 --> 00:16:12.513
We don't know if they're going to get it, but they're asking for it.
157
00:16:12.653 --> 00:16:16.778
So they're basically asking pension funds to bet on the AI bubble.
158
00:16:16.979 --> 00:16:19.141
And that brings us to my next question is –
159
00:16:20.182 --> 00:16:21.603
Is AI going to kill us all?
160
00:16:22.263 --> 00:16:24.105
The answer is not no.
161
00:16:24.625 --> 00:16:26.226
It's a resounding possibly.
162
00:16:26.866 --> 00:16:37.353
Anthropic researcher Jacob Coxon resigned this week and publicly warned that people building frontier AI genuinely believe the technology could pose an existential threat to humanity.
163
00:16:37.913 --> 00:16:47.480
His main post went properly nuclear online, passing 100 million views, and Axios reported that he left about two months before his anthropic equity would have vested.
164
00:16:47.981 --> 00:16:54.246
Walking away from meaningful equity doesn't prove the fear is correct, but it makes the warning harder to dismiss as cheap attention-seeking.
165
00:16:54.966 --> 00:16:58.091
Other people at Anthropic have expressed similarly alarming views.
166
00:16:58.271 --> 00:17:05.982
Alignment researcher Evan Hubbinger has put the probability of AI causing human extinction within the next decade at greater than 10%.
167
00:17:06.643 --> 00:17:07.785
Which do I say, 10%!
168
00:17:09.773 --> 00:17:14.977
I don't, I won't get on a plane if somebody tells me there's a 10% chance there isn't Diet Coke on the plane.
169
00:17:15.558 --> 00:17:20.201
So I don't want to wave this away, but these are very smart people working very close to the technology.
170
00:17:20.241 --> 00:17:24.585
And when people with that level of access say they're genuinely frightened, we should listen to them.
171
00:17:25.085 --> 00:17:27.449
But historical context also matters.
172
00:17:27.769 --> 00:17:38.905
When OpenAI announced GPT-2 in 2029, it withheld the full model because of concerns around malicious use, then released increasingly capable versions before eventually releasing the whole thing.
173
00:17:39.105 --> 00:17:40.668
GPT-2 today...
174
00:17:40.948 --> 00:17:43.431
feels about as threatening as an ambitious toaster.
175
00:17:43.672 --> 00:17:45.514
None of this means today's warnings are fake.
176
00:17:45.714 --> 00:17:55.567
And to be honest, the thing that gave me more pause was that Dario Amadei wrote on his blog that we should pause or slow down the rate of progress of these models.
177
00:17:56.028 --> 00:17:56.869
His fear was...
178
00:17:57.710 --> 00:18:03.192
A really highly capable new agent could create bot swarms that could literally take down the internet.
179
00:18:03.272 --> 00:18:04.792
And where would we be without the internet?
180
00:18:04.992 --> 00:18:12.174
What really gave me pause was that Sam Altman and Elon Musk all agreed with him.
181
00:18:12.954 --> 00:18:15.595
Elon Musk tweeted, Dario's right.
182
00:18:15.955 --> 00:18:22.937
The notion of modern CEOs in tech all calling to slow down...
183
00:18:24.265 --> 00:18:32.895
is either incredibly self-serving because they're trying to save on the R&D costs, or it is actually a harbinger of something.
184
00:18:33.135 --> 00:18:33.956
I don't know what.
185
00:18:34.749 --> 00:18:38.272
But we can't slow down too much because China's doing this too.
186
00:18:38.533 --> 00:18:51.445
But when three of the biggest guys in tech, two of whom we know are pretty cavalier, are all calling for a slowdown in lines with what Bernie Sanders is suggesting, that's a moment we should all look at.
187
00:18:51.665 --> 00:18:54.127
And this is a perfect moment for government to step in.
188
00:18:54.227 --> 00:18:54.868
So I don't know.
189
00:18:54.888 --> 00:18:56.650
I don't know what to do about that news.
190
00:18:57.310 --> 00:19:06.078
I think it's great, but these CEOs, especially when they go public, they're not really in control of things like safety and governance anymore.
191
00:19:06.319 --> 00:19:11.463
The responsibility to deliver shareholder value becomes their absolute North Star.
192
00:19:12.004 --> 00:19:14.186
So this does seem like a job for the government.
193
00:19:14.306 --> 00:19:16.688
And if the people running the companies at least accept that—
194
00:19:17.469 --> 00:19:19.690
that might be a really good step in the right direction.
195
00:19:19.790 --> 00:19:29.436
But it is interesting that after dooming and glooming every single time there's a model release, almost as a marketing strategy, it's hard to know when to take these guys seriously.
196
00:19:29.996 --> 00:19:38.060
Because after so many years of this model maybe too powerful to release, I think the public is entitled to increasingly specific answers.
197
00:19:38.521 --> 00:19:39.581
What can it actually do?
198
00:19:39.681 --> 00:19:40.822
What happened in testing?
199
00:19:40.922 --> 00:19:41.582
What failed?
200
00:19:41.943 --> 00:19:43.003
What safeguards worked?
201
00:19:43.103 --> 00:19:44.164
What safeguard didn't?
202
00:19:44.744 --> 00:19:45.825
What safeguards didn't?
203
00:19:46.583 --> 00:19:51.687
I mean, show us, because trust us, it's terrifying is not a safety framework.
204
00:19:51.767 --> 00:19:55.169
I think that what connects these stories is that the contracts are changing.
205
00:19:55.610 --> 00:19:57.051
And that's why this feels weird.
206
00:19:57.731 --> 00:19:59.913
Technology doesn't just change what we can do.
207
00:20:00.033 --> 00:20:07.499
It changes the little agreements we make about how the world is supposed to work, the social and corporate contracts we have with one another.
208
00:20:08.019 --> 00:20:08.939
Advertising had one.
209
00:20:09.479 --> 00:20:11.040
Agencies did agency things.
210
00:20:11.440 --> 00:20:13.081
Production companies made the films.
211
00:20:13.521 --> 00:20:19.642
Great directors and producers went to production companies because that was where the craft, the mentorship, and the opportunity lived.
212
00:20:20.043 --> 00:20:26.064
But the next generation is coming out of universities right now, and most of them are not protecting an old career model.
213
00:20:26.504 --> 00:20:27.585
They want to make things.
214
00:20:27.865 --> 00:20:28.705
They want to get better.
215
00:20:28.745 --> 00:20:31.346
They want to build a body of work, and they want to prove themselves.
216
00:20:31.586 --> 00:20:31.906
Right.
217
00:20:31.966 --> 00:20:32.807
They want a shot.
218
00:20:33.147 --> 00:20:41.433
They want reps. And if those opportunities increasingly sit inside agencies or on social media, then that's where they'll go.
219
00:20:42.014 --> 00:20:42.874
They'll grow there.
220
00:20:42.934 --> 00:20:44.576
Their skill sets will develop there.
221
00:20:45.116 --> 00:20:47.158
And over time, the center of gravity shifts.
222
00:20:47.818 --> 00:20:49.699
Talent flows towards opportunity.
223
00:20:49.979 --> 00:20:51.859
That's why this moment feels so strange.
224
00:20:52.099 --> 00:20:54.180
Dial-up became the internet over decades.
225
00:20:54.520 --> 00:20:56.960
Social media rewrote the culture much faster.
226
00:20:57.380 --> 00:21:04.862
Generative AI feels like it's renegotiating authorship, truth, expertise, and work within seven years.
227
00:21:05.282 --> 00:21:08.083
And we're still arguing about the social media revolution.
228
00:21:08.283 --> 00:21:10.984
Technology is moving faster than the conventions around it.
229
00:21:11.564 --> 00:21:13.425
And eventually we'll write new rules.
230
00:21:13.485 --> 00:21:16.287
Then somebody will invent something else and fuck them up again.
231
00:21:16.327 --> 00:21:19.049
But I think we have to get faster at writing the new rules.
232
00:21:19.369 --> 00:21:24.252
Because we've all bought into a society where corporations have nearly unchecked power.
233
00:21:24.312 --> 00:21:26.153
We can't be surprised when they use it.
234
00:21:26.493 --> 00:21:29.815
So on that shiny note, thank you for listening this week.
235
00:21:29.955 --> 00:21:34.098
Tomorrow's interview episode will be with Toby Walsh of Made by Humans and Imagine This Studio.
236
00:21:34.378 --> 00:21:38.660
Doing some amazing AI stuff in the production company space.
237
00:21:38.881 --> 00:21:40.842
And crucially not doing it inside an agency.
238
00:21:41.302 --> 00:21:44.367
So anyway, thank you for listening, and we'll see you next week.
