WEBVTT

00:00:00.240 --> 00:00:02.859
Welcome to Artificially Intelligent Marketing,

00:00:03.100 --> 00:00:06.040
a weekly podcast where we stay on top of the

00:00:06.040 --> 00:00:09.320
latest trends, tips, and tools in the world of

00:00:09.320 --> 00:00:12.160
marketing AI, helping you get the best results

00:00:12.160 --> 00:00:15.300
from your marketing efforts. Now let's join our

00:00:15.300 --> 00:00:19.239
hosts, Paul Avery and Martin Broadhurst. Hello

00:00:19.239 --> 00:00:21.839
everyone, welcome back again to Artificially

00:00:21.839 --> 00:00:23.839
Intelligent Marketing. Paul Avery here, and I'm

00:00:23.839 --> 00:00:26.559
also joined by my good pal, Martin Broadhurst.

00:00:26.579 --> 00:00:29.120
Martin, how are we? oh i am alive and kicking

00:00:29.120 --> 00:00:32.439
this morning it's been half term last week so

00:00:32.439 --> 00:00:37.460
i was not in the office but i was exploring the

00:00:37.460 --> 00:00:40.579
science museum in london and the natural history

00:00:40.579 --> 00:00:43.859
museum and doing all that good touristy stuff

00:00:43.859 --> 00:00:47.619
with a the four -year -old in tow any robots

00:00:48.739 --> 00:00:52.700
There were robots at the Science Museum for sure.

00:00:52.799 --> 00:00:56.039
There was a large animatronic T -Rex at the Natural

00:00:56.039 --> 00:00:57.840
History Museum, although I'm not entirely sure

00:00:57.840 --> 00:01:02.240
that constitutes artificial intelligence. Goodness,

00:01:02.500 --> 00:01:05.920
I hope they don't put GPT -5 in that. And then

00:01:05.920 --> 00:01:10.019
it's like, oh, but it might be able to see, but...

00:01:10.250 --> 00:01:13.269
I learnt from Jurassic Park very reliably that

00:01:13.269 --> 00:01:15.670
it can only see you if you move. So you just,

00:01:15.730 --> 00:01:17.989
I'd love to see that. You just still stood in

00:01:17.989 --> 00:01:20.790
the museum now, Martin, like, we can't move.

00:01:21.230 --> 00:01:23.629
Just very still. I mean, that is an absolute

00:01:23.629 --> 00:01:27.569
nonsense myth perpetuated by Jurassic Park. Huge

00:01:27.569 --> 00:01:33.209
eyeballs and vision that was kind of, the eyes

00:01:33.209 --> 00:01:35.689
were at the front, right? So it's got proper

00:01:35.689 --> 00:01:39.370
depth perception. It was a very good vision.

00:01:40.519 --> 00:01:43.140
Are you suggesting that Hollywood movies can't

00:01:43.140 --> 00:01:44.939
be trusted for their factual accuracy? Because

00:01:44.939 --> 00:01:48.040
that would basically ruin my life if you suggest

00:01:48.040 --> 00:01:50.340
that's true. Look, I'm just saying that Jurassic

00:01:50.340 --> 00:01:53.439
Park, as great a movie as it is, and boy, do

00:01:53.439 --> 00:01:58.799
I love it, wasn't representative of the actual

00:01:58.799 --> 00:02:03.719
abilities of dinosaurs as we understand them.

00:02:04.000 --> 00:02:06.099
Sorry, listeners, we should cut this out. We're

00:02:06.099 --> 00:02:07.200
not going to. We're going to make you listen

00:02:07.200 --> 00:02:09.919
to this unnecessary stories about dinosaurs.

00:02:10.280 --> 00:02:12.599
But there's some stories that are very necessary

00:02:12.599 --> 00:02:14.719
about AI this week. And we're going to get stuck

00:02:14.719 --> 00:02:16.439
into everything that marketers need to know.

00:02:17.219 --> 00:02:19.439
Martin, kick us off with you've been trawling

00:02:19.439 --> 00:02:21.719
through earnings calls, haven't you? And found

00:02:21.719 --> 00:02:23.520
some interesting stuff. What have you been looking

00:02:23.520 --> 00:02:26.580
at? The past week, we've had earning calls from

00:02:26.580 --> 00:02:29.960
both Google and Meta. both of which highlight

00:02:29.960 --> 00:02:33.939
the importance of the ai investments within the

00:02:33.939 --> 00:02:37.620
actual advertising suites of both companies just

00:02:37.620 --> 00:02:44.919
starting with meta they have announced huge engagement

00:02:44.919 --> 00:02:50.520
improvements through using new ai tools and ai

00:02:50.520 --> 00:02:52.740
product offerings so they're saying that the

00:02:52.740 --> 00:02:55.819
ai recommendations led to five percent more time

00:02:55.819 --> 00:02:59.240
on facebook and ten percent on threads video

00:02:59.240 --> 00:03:02.780
time on instagram is up more than 30 percent

00:03:02.780 --> 00:03:06.879
last year and that's been driven by ai and facebook

00:03:06.879 --> 00:03:09.520
systems now surface twice as many reels published

00:03:09.520 --> 00:03:13.560
per day they've also launched this new tool called

00:03:13.560 --> 00:03:18.319
vibes where you can within the meta ai app you

00:03:18.319 --> 00:03:22.199
can basically create images and have them animated

00:03:22.199 --> 00:03:25.780
as well i think that's in partnership with mid

00:03:25.780 --> 00:03:29.400
journey where they're using some of the mid journey

00:03:29.400 --> 00:03:33.280
technology for a kind of aesthetics application

00:03:33.280 --> 00:03:38.090
within the image generation In terms of numbers

00:03:38.090 --> 00:03:40.689
of images created, it's quite massive. People

00:03:40.689 --> 00:03:43.629
have created over 20 billion images using Meta's

00:03:43.629 --> 00:03:47.590
products. So that's quite dramatic. But the big

00:03:47.590 --> 00:03:50.990
story really, as I mentioned, is the AI in advertising.

00:03:51.289 --> 00:03:54.430
So they're saying that the financial impact of

00:03:54.430 --> 00:03:58.949
AI is huge. There is an annual run rate of...

00:03:59.530 --> 00:04:02.550
end -to -end ai powered ad tools which has passed

00:04:02.550 --> 00:04:06.050
60 billion dollars so that's to say there is

00:04:06.050 --> 00:04:10.610
60 billion dollars worth of advertising on an

00:04:10.610 --> 00:04:14.729
annualized basis that is being affected impacted

00:04:14.729 --> 00:04:19.990
deployed call it what you may using the ai suite

00:04:19.990 --> 00:04:23.610
of tools within facebook it's super interesting

00:04:23.610 --> 00:04:26.149
because um obviously i think as marketers we

00:04:26.149 --> 00:04:29.019
spend a lot of our time thinking about generative

00:04:29.019 --> 00:04:32.379
ai and the stuff we can make but a big thing

00:04:32.379 --> 00:04:35.480
that many marketers are interested in but is

00:04:35.480 --> 00:04:40.120
taking a while to really solidify is targeting

00:04:40.120 --> 00:04:43.120
right we we want to send the right message to

00:04:43.120 --> 00:04:45.459
the right person at the right time and ai has

00:04:45.459 --> 00:04:48.500
a lot of applications to offer us there so it

00:04:48.500 --> 00:04:51.779
sounds like meta is really ramping up its ability

00:04:51.779 --> 00:04:54.339
to know what content it needs to show to what

00:04:54.339 --> 00:04:56.540
person to keep them on site which is an advertiser

00:04:56.540 --> 00:04:58.480
you're probably quite happy with because it probably

00:04:58.480 --> 00:05:01.079
means they're also surfacing the right ads to

00:05:01.079 --> 00:05:03.579
the right people more accurately and effectively

00:05:03.579 --> 00:05:07.519
than they did so in the past yeah very much so

00:05:07.519 --> 00:05:11.259
and it goes beyond uh the creative into all of

00:05:11.259 --> 00:05:13.100
those areas that you've touched so they've got

00:05:13.100 --> 00:05:17.959
this suite of tools called advantage plus which

00:05:17.959 --> 00:05:22.939
has audience targeting it has creatives it has

00:05:22.939 --> 00:05:26.240
placements which decides where the ads should

00:05:26.240 --> 00:05:29.160
appear so should they appear in feed in reels

00:05:29.160 --> 00:05:32.720
on instagram on facebook on threads etc but they've

00:05:32.720 --> 00:05:36.800
also got a destination targeting as well and

00:05:36.800 --> 00:05:40.720
budget optimization now you can deploy an ad

00:05:40.720 --> 00:05:45.410
campaign using the whole suite or you can just

00:05:45.410 --> 00:05:49.470
choose advantage plus audiences or creative for

00:05:49.470 --> 00:05:53.670
creative optimization now all of this kind of

00:05:53.670 --> 00:06:00.649
feeds into mark zuckerberg's vision for the advertising

00:06:00.649 --> 00:06:03.629
suite he actually says on the earnings call advertisers

00:06:03.629 --> 00:06:06.350
are increasingly just going to be able to give

00:06:06.350 --> 00:06:08.829
us a business objective and give us a credit

00:06:08.829 --> 00:06:11.350
card or bank account and have the ai system basically

00:06:11.350 --> 00:06:14.040
figure out everything that's necessary including

00:06:14.040 --> 00:06:16.680
generating video or different types of creative

00:06:16.680 --> 00:06:19.560
that might resonate with different people that

00:06:19.560 --> 00:06:22.699
are personalized in different ways, finding who

00:06:22.699 --> 00:06:24.740
the right customers are. So this is going to

00:06:24.740 --> 00:06:27.240
be interesting. I don't think it's so true now,

00:06:27.319 --> 00:06:30.060
but in the, I'm not going to say early days,

00:06:30.180 --> 00:06:34.519
but earlier, less sophisticated AI days, there

00:06:34.519 --> 00:06:37.600
was reticence from folks working in things like,

00:06:37.639 --> 00:06:41.240
you know, Google search ads to leverage AI, because

00:06:41.240 --> 00:06:44.759
I think there was this. feeling that ai from

00:06:44.759 --> 00:06:46.819
google's perspective their goal was to get you

00:06:46.819 --> 00:06:49.680
to spend more money and that if you turn over

00:06:49.680 --> 00:06:53.160
control of your campaigns to google you may or

00:06:53.160 --> 00:06:55.120
you may not get better results but one thing

00:06:55.120 --> 00:06:56.720
they felt you could guarantee is you would end

00:06:56.720 --> 00:07:00.279
up spending more money it'll be interesting to

00:07:00.279 --> 00:07:02.920
see if marketers continue to feel like that as

00:07:02.920 --> 00:07:06.839
these suites of tools emerge because you know

00:07:06.839 --> 00:07:09.720
do you really want to hand over complete control

00:07:09.720 --> 00:07:13.509
of your campaigns to your supplier basically

00:07:13.509 --> 00:07:15.949
he's like yeah i'll optimize your campaigns you

00:07:15.949 --> 00:07:18.990
will on average spend about five times more than

00:07:18.990 --> 00:07:20.610
you used to are you cool with that because i'll

00:07:20.610 --> 00:07:23.089
definitely get you better results and i guess

00:07:23.089 --> 00:07:26.029
there are some business models where you want

00:07:26.029 --> 00:07:27.649
to scale up and up and up but there are others

00:07:27.649 --> 00:07:29.509
where things like product inventory will become

00:07:29.509 --> 00:07:31.629
the limiting factor and all of this type of stuff

00:07:31.629 --> 00:07:35.829
so yeah i i can see the advantage for everyone

00:07:35.829 --> 00:07:38.550
i just i'm not sure i don't trust meta to take

00:07:38.550 --> 00:07:41.350
most of the spoils of making this better do you

00:07:41.350 --> 00:07:43.490
know what i mean Oh, absolutely. They're going

00:07:43.490 --> 00:07:45.629
to skim off as much as they can. And I think

00:07:45.629 --> 00:07:48.709
the experience of any advertisers using things

00:07:48.709 --> 00:07:50.910
like Performance Max in Google will tell you

00:07:50.910 --> 00:07:56.089
that there's a somewhat obfuscation of data that

00:07:56.089 --> 00:07:59.149
you can access. And when you dig into the kinds

00:07:59.149 --> 00:08:01.790
of things it's been bidding on and you see how

00:08:01.790 --> 00:08:04.050
much it's bid on some of the keywords, there

00:08:04.050 --> 00:08:09.410
is quite a lot of money being burnt on ad targeting

00:08:09.410 --> 00:08:12.300
and ad placements that... no human would have

00:08:12.300 --> 00:08:15.560
ever decided to have actually deployed. There

00:08:15.560 --> 00:08:18.240
is another interesting development on the meta

00:08:18.240 --> 00:08:21.800
AI piece, which is where they're getting their

00:08:21.800 --> 00:08:25.300
targeting options from. So starting from December,

00:08:25.420 --> 00:08:29.959
they are announcing that the company will use

00:08:29.959 --> 00:08:36.159
conversation data from the meta AI chat app to

00:08:36.159 --> 00:08:39.379
target ads. So that means any chats. that are

00:08:39.379 --> 00:08:44.220
conducted with meta ai on messenger on whatsapp

00:08:44.220 --> 00:08:46.779
or anything like that could soon inform the answer

00:08:46.779 --> 00:08:50.519
you see now this will obviously open up more

00:08:50.519 --> 00:08:54.340
insights into what people are looking for what

00:08:54.340 --> 00:08:58.500
kind of more stronger intent data shall we say

00:08:58.500 --> 00:09:03.549
but also privacy concerns I mean, talk about

00:09:03.549 --> 00:09:06.529
waving a red flag. Yeah, I think, you know, I

00:09:06.529 --> 00:09:09.470
think Meta's got like 1 billion monthly users

00:09:09.470 --> 00:09:13.529
of their AI, so they claim. I'd love to know

00:09:13.529 --> 00:09:16.769
what counts as a use of Meta AI, because certainly

00:09:16.769 --> 00:09:19.490
I don't chat with it on any platform, but I don't

00:09:19.490 --> 00:09:22.009
know if I search for something on Instagram or

00:09:22.009 --> 00:09:23.669
whatever, maybe that counts as an interaction.

00:09:24.009 --> 00:09:26.330
But ultimately, everybody's doing this, right?

00:09:27.110 --> 00:09:31.070
ChatGPT is... going to have a how do they create

00:09:31.070 --> 00:09:33.590
the memory function if they don't also give themselves

00:09:33.590 --> 00:09:36.629
the ability to share very personalized ads based

00:09:36.629 --> 00:09:39.330
on the same memory that you as a user are benefiting

00:09:39.330 --> 00:09:42.450
from obviously google's entire business model

00:09:42.450 --> 00:09:46.429
is understanding the searcher to show them better

00:09:46.429 --> 00:09:49.529
ads and you know more relevant ads over time

00:09:49.529 --> 00:09:52.909
it's almost like i feel like meta's the one that

00:09:52.909 --> 00:09:55.870
you know there's no smoking daggers we all know

00:09:55.870 --> 00:09:59.809
what their end game is um but anybody who with

00:09:59.809 --> 00:10:02.629
a product that's driven by ai is trying to basically

00:10:02.629 --> 00:10:05.529
optimize for our attention regardless and better

00:10:05.529 --> 00:10:08.809
understand us in order to to do that so yeah

00:10:08.809 --> 00:10:11.649
the cynic in me is like oh it makes me not want

00:10:11.649 --> 00:10:14.669
to use any meta products and then like i take

00:10:14.669 --> 00:10:17.049
a step back from that and i know i probably shouldn't

00:10:17.049 --> 00:10:19.190
use any digital products if i feel like that

00:10:19.190 --> 00:10:21.950
and then you know That's probably most of the

00:10:21.950 --> 00:10:24.049
things I use during my day. I've now got to start

00:10:24.049 --> 00:10:25.590
questioning whether or not I would use them or

00:10:25.590 --> 00:10:28.409
not. So maybe privacy is a myth. But I can see

00:10:28.409 --> 00:10:29.990
that people will definitely potentially be a

00:10:29.990 --> 00:10:32.129
bit freaked out by some of this. They will. But

00:10:32.129 --> 00:10:34.990
hats off to Meta for, you know, they're quite

00:10:34.990 --> 00:10:37.889
upfront about it. And I guess this is another

00:10:37.889 --> 00:10:40.190
way in which the incentive is there for them

00:10:40.190 --> 00:10:46.090
to develop a stronger foundational model. Because

00:10:46.090 --> 00:10:49.990
when people start using... meta ai when they

00:10:49.990 --> 00:10:53.090
have an experience using meta ai which is free

00:10:53.090 --> 00:10:58.289
and it's delivering results that are as good

00:10:58.289 --> 00:11:02.529
or comparable with the best chatbots elsewhere

00:11:02.529 --> 00:11:07.269
then low friction low cost is a path that we

00:11:07.269 --> 00:11:10.909
know consumers often take so if they can monetize

00:11:10.909 --> 00:11:14.870
that as you say there's no uh there's no hiding

00:11:14.870 --> 00:11:17.049
what meta's business model has always been about

00:11:17.049 --> 00:11:18.789
is advertising revenue at the end of the day

00:11:18.789 --> 00:11:21.750
so yeah there's a there's strong incentives for

00:11:21.750 --> 00:11:25.830
them to to create a super intelligent model to

00:11:25.830 --> 00:11:28.769
get you to chat to it a bit more yeah so obviously

00:11:28.769 --> 00:11:31.690
meta pretty busy but google had a pretty good

00:11:31.690 --> 00:11:34.909
quarter as well and uh supposedly driven quite

00:11:34.909 --> 00:11:37.509
a lot by ai what was going on over at google

00:11:37.509 --> 00:11:41.909
and their earnings call pretty good water indeed

00:11:43.140 --> 00:11:47.279
They posted their first ever billion dollar quarter.

00:11:48.279 --> 00:11:50.919
And the drive... 100 billion, 100 billion dollar

00:11:50.919 --> 00:11:53.559
quarter. Sorry, yeah. I mean, should we edit

00:11:53.559 --> 00:11:55.820
that out? No, we're not going to edit this out.

00:11:55.899 --> 00:11:57.460
Sorry, listeners, we're leaving it in. Because

00:11:57.460 --> 00:12:01.139
I have now just got Dr. Evil in my mind and mine's

00:12:01.139 --> 00:12:06.350
going... They made $1 billion. And then everyone

00:12:06.350 --> 00:12:08.629
with Dr. Eve was like, no, they're a trillion

00:12:08.629 --> 00:12:15.029
dollar company. $100 billion. $100 billion quarter,

00:12:15.190 --> 00:12:18.269
Martin. And what's driving this? Is it AI? Completely

00:12:18.269 --> 00:12:21.230
thrown me off there. Yes, it's AI, unsurprisingly,

00:12:21.330 --> 00:12:24.549
in its core ad products for both Search and YouTube.

00:12:25.519 --> 00:12:29.919
uh both seeing 15 year -on -year growth new features

00:12:29.919 --> 00:12:32.240
like ai overviews are encouraging users to make

00:12:32.240 --> 00:12:36.580
more complex queries and to stay on the platform

00:12:36.580 --> 00:12:39.500
a bit longer while automated campaign tools like

00:12:39.500 --> 00:12:43.580
the aforementioned performance max are delivering

00:12:43.580 --> 00:12:48.259
performance lifts as well so yeah all in all

00:12:48.259 --> 00:12:55.940
pretty good sounds like um the ai tools for youtube

00:12:55.940 --> 00:12:58.039
and video advertising is quite interesting so

00:12:58.039 --> 00:13:01.799
they're using ai to analyze video content to

00:13:01.799 --> 00:13:05.840
identify better better fit and ads and better

00:13:05.840 --> 00:13:10.360
fit creatives for certain videos based on the

00:13:10.360 --> 00:13:12.620
content of the video which i think is quite an

00:13:12.620 --> 00:13:16.580
interesting use case um but yeah they're saying

00:13:16.580 --> 00:13:21.240
um ai is the future of advertising but interestingly

00:13:21.240 --> 00:13:25.639
when you dig into the the numbers the ad network

00:13:25.639 --> 00:13:29.879
so the external network can you remember what

00:13:29.879 --> 00:13:32.059
that's called i can't the display network basically

00:13:32.059 --> 00:13:36.820
revenue from that is down and some people are

00:13:36.820 --> 00:13:40.460
saying well that's because ai mode and ai search

00:13:40.460 --> 00:13:45.139
is sending less traffic to publishers so if publishers

00:13:45.139 --> 00:13:48.559
are getting less traffic then google's display

00:13:48.559 --> 00:13:52.159
ads are getting shown less so it's kind of cannibalizing

00:13:52.159 --> 00:13:55.320
that bit of its market I'm sure there are markets

00:13:55.320 --> 00:13:58.440
where Google Display works really well, but it

00:13:58.440 --> 00:14:00.500
always used to make me smile. In the life sciences,

00:14:00.580 --> 00:14:03.200
it's so easy to set it up wrong and burn a load

00:14:03.200 --> 00:14:06.320
of cash sharing ads in places you don't want

00:14:06.320 --> 00:14:10.120
them. The classic is the amount of ads that get

00:14:10.120 --> 00:14:14.940
shown on kids' games on your mobile phone. You

00:14:14.940 --> 00:14:17.279
know, the free games where you play for 10 minutes

00:14:17.279 --> 00:14:21.120
and then you have to look at an ad for 15 seconds.

00:14:22.799 --> 00:14:25.720
So, yeah, I've got my own problems with Google

00:14:25.720 --> 00:14:27.759
Display anyway. Maybe people are just moving

00:14:27.759 --> 00:14:30.440
away from using it because it's not always that

00:14:30.440 --> 00:14:32.820
good. But it is interesting because we were just

00:14:32.820 --> 00:14:35.139
talking about, you know, who gets the spoils

00:14:35.139 --> 00:14:38.539
if ads work better. But there were a couple of

00:14:38.539 --> 00:14:40.779
case studies. Again, they can cherry pick these.

00:14:40.860 --> 00:14:42.860
I don't know how robust these are. But, you know,

00:14:42.899 --> 00:14:46.059
Kayak seeing that 10, 12 % increase in conversion

00:14:46.059 --> 00:14:49.840
value. SoFi achieving that 39 % improvement in

00:14:49.840 --> 00:14:53.059
conversion volume. That at least is a share of

00:14:53.059 --> 00:14:57.299
the spoils in terms of AI improving performance,

00:14:57.620 --> 00:14:59.519
which is pretty interesting. The other thing

00:14:59.519 --> 00:15:01.279
I saw that was pretty interesting, Martin, because

00:15:01.279 --> 00:15:04.379
we talked about this a lot off air, is reels

00:15:04.379 --> 00:15:06.740
and shorts. So YouTube shorts are now earning

00:15:06.740 --> 00:15:09.059
more revenue per watch hour than traditional

00:15:09.059 --> 00:15:13.399
in -stream ads, which is ever more evidence that

00:15:13.399 --> 00:15:18.379
people are migrating to short, bite -sized content.

00:15:19.279 --> 00:15:22.159
we've seen success little insider secret for

00:15:22.159 --> 00:15:25.860
the for the podcast people um with our own reels

00:15:25.860 --> 00:15:27.700
running them on youtube shorts being able to

00:15:27.700 --> 00:15:29.899
get much greater penetrance than we could with

00:15:29.899 --> 00:15:32.940
you know the full length episode google obviously

00:15:32.940 --> 00:15:35.620
working overtime to try and you know show these

00:15:35.620 --> 00:15:38.820
short form content to people get them engaged

00:15:38.820 --> 00:15:42.120
in it etc so it's going to be interesting to

00:15:42.120 --> 00:15:45.570
see that continue And maybe that is a great place

00:15:45.570 --> 00:15:47.649
then to be putting your ads as well. And I'm

00:15:47.649 --> 00:15:50.730
sure PMAX is doing that automatically. So it

00:15:50.730 --> 00:15:52.870
creates quite a complex ecosystem, right? They've

00:15:52.870 --> 00:15:56.190
got their own chips. They've got their own AI

00:15:56.190 --> 00:15:58.909
models. They've got the reach that comes with

00:15:58.909 --> 00:16:01.690
Google search and YouTube and all of these platforms.

00:16:02.929 --> 00:16:06.029
It's hard to, I mean, for me, Google would really

00:16:06.029 --> 00:16:10.070
have to mess this whole AI thing up not to be

00:16:10.070 --> 00:16:13.100
the winner in the end, surely. Yeah, and their

00:16:13.100 --> 00:16:15.480
capital investments are increasing dramatically

00:16:15.480 --> 00:16:17.879
as well. So in this earnings report, they talked

00:16:17.879 --> 00:16:23.620
about $95 billion of capital expenditure forward

00:16:23.620 --> 00:16:26.940
looking. So I think that might have been this

00:16:26.940 --> 00:16:28.779
year and they're looking to continue that as

00:16:28.779 --> 00:16:31.960
well. So huge investments across the board, but

00:16:31.960 --> 00:16:34.000
they're not alone in that. I think you can see

00:16:34.000 --> 00:16:37.179
the same thing from, well, obviously from OpenAI,

00:16:37.340 --> 00:16:39.639
but also from all the other big labs as well.

00:16:40.019 --> 00:16:43.299
Rightio. Let's move on to our next story then,

00:16:43.360 --> 00:16:45.320
Martin. What have you got for us next? Okay,

00:16:45.360 --> 00:16:48.379
so some big acquisition news from my favorite

00:16:48.379 --> 00:16:52.200
CRM provider. I should say, full disclosure,

00:16:52.299 --> 00:16:54.639
I am a HubSpot Platinum partner, so I'm bound

00:16:54.639 --> 00:16:58.519
to plug them wherever I can. But yes, HubSpot

00:16:58.519 --> 00:17:03.059
have announced the acquisition of XFunnel, which

00:17:03.059 --> 00:17:06.140
is a platform dedicated to what it calls... answer

00:17:06.140 --> 00:17:09.440
engine optimization aeo you might see it geo

00:17:09.440 --> 00:17:11.880
generative engine optimization call it what you

00:17:11.880 --> 00:17:14.920
will but it's basically how visible are you in

00:17:14.920 --> 00:17:20.359
search queries within ai chat assistance so this

00:17:20.359 --> 00:17:23.000
acquisition is pretty much a direct acknowledgement

00:17:23.000 --> 00:17:26.750
that more and more people are turning to ai first

00:17:26.750 --> 00:17:29.710
rather than search engines and brands need to

00:17:29.710 --> 00:17:31.509
make sure that they're part of that conversation

00:17:31.509 --> 00:17:34.990
hubspot says ai drone leads already convert three

00:17:34.990 --> 00:17:37.910
times better than traditional search and this

00:17:37.910 --> 00:17:43.069
is a bet on that accelerating as a trend back

00:17:43.069 --> 00:17:48.329
to aeo yeah i it's one of those i still think

00:17:48.329 --> 00:17:53.920
i understand why everybody's betting that people

00:17:53.920 --> 00:17:56.319
start interacting with chatbots more than google

00:17:56.319 --> 00:18:00.480
search and and people remember that a lot of

00:18:00.480 --> 00:18:03.500
people i think feel they miss the boat on seo

00:18:03.500 --> 00:18:06.460
like if i was the first to market for seo can

00:18:06.460 --> 00:18:08.519
you imagine all the traffic i would have got

00:18:08.519 --> 00:18:10.500
and how much money i would have made and oh does

00:18:10.500 --> 00:18:12.460
everybody remember when google ads was really

00:18:12.460 --> 00:18:14.700
cheap not me because i didn't use it early enough

00:18:14.700 --> 00:18:16.460
and by the time i figured out to use it it was

00:18:16.460 --> 00:18:18.880
too expensive for everyone so i think there's

00:18:18.880 --> 00:18:21.059
this weird fomo that's driving all of this which

00:18:21.059 --> 00:18:25.490
is not to say So interacting with chatbots, having

00:18:25.490 --> 00:18:28.009
your brand present in chatbots is strategically

00:18:28.009 --> 00:18:30.490
valuable. It is probably where everybody's going

00:18:30.490 --> 00:18:32.470
to go. Don't know what that time period is going

00:18:32.470 --> 00:18:36.190
to be. But using a chatbot to get information

00:18:36.190 --> 00:18:38.650
is quite often not the same as Google searching

00:18:38.650 --> 00:18:41.490
for stuff. I do think there'll be a time where

00:18:41.490 --> 00:18:44.549
we're like, chat GPT, I need to buy X, Y, Z.

00:18:44.589 --> 00:18:46.289
Can you go do the research and give me a breakdown

00:18:46.289 --> 00:18:50.009
of which one I should buy? But I just don't know

00:18:50.009 --> 00:18:52.630
how quick that's going to come around. And so

00:18:52.630 --> 00:18:56.170
this massive push into AEO, I know it's important.

00:18:56.349 --> 00:18:58.390
I do think people should be thinking about it,

00:18:58.450 --> 00:18:59.990
especially if you've got quite a sophisticated

00:18:59.990 --> 00:19:02.309
marketing stack already and you do SEO and you

00:19:02.309 --> 00:19:05.829
do PPC, like you can probably start to get ahead

00:19:05.829 --> 00:19:07.549
early here if you're doing the right things.

00:19:07.730 --> 00:19:10.289
But look, obviously HubSpot think it's important

00:19:10.289 --> 00:19:11.390
because otherwise it wouldn't have made this

00:19:11.390 --> 00:19:13.690
acquisition. Yeah, and I'm very much inclined

00:19:13.690 --> 00:19:16.109
to agree with you. We've discussed previously,

00:19:16.130 --> 00:19:20.430
people are saying, how do you optimize for search

00:19:20.430 --> 00:19:25.509
in AI? And I think this is such a nascent technology

00:19:25.509 --> 00:19:28.950
and channel for people that the tactics at work

00:19:28.950 --> 00:19:32.230
today will probably change in six months. I think

00:19:32.230 --> 00:19:35.069
I saw an interesting study that looked at how

00:19:35.069 --> 00:19:38.789
prominent Reddit was in the search results of

00:19:38.789 --> 00:19:43.549
OpenAI. And now within six months of that first

00:19:43.549 --> 00:19:46.089
piece of research coming out, Reddit has dramatically

00:19:46.089 --> 00:19:49.309
fallen off in terms of the visibility within

00:19:49.309 --> 00:19:51.859
search results. So clearly there's some... tweaking

00:19:51.859 --> 00:19:55.140
and ongoing optimization of search results and

00:19:55.140 --> 00:19:59.500
sources in the back end but generally speaking

00:19:59.500 --> 00:20:03.400
if you are trying to optimize for ai i don't

00:20:03.400 --> 00:20:07.720
think you can do yourself any harm by focusing

00:20:07.720 --> 00:20:11.319
on just having good brand building campaigns

00:20:11.319 --> 00:20:16.299
the classics of good pr will stand you in good

00:20:16.299 --> 00:20:19.839
stead when it comes to getting found in both

00:20:19.839 --> 00:20:23.480
search whether that's traditional search or AI

00:20:23.480 --> 00:20:26.839
search. Yeah, I think that's, you've hit the

00:20:26.839 --> 00:20:29.400
nail on the head. I think about interacting with

00:20:29.400 --> 00:20:31.319
the chatbot on if I'm going to buy something

00:20:31.319 --> 00:20:33.519
or if I'm gathering information on something

00:20:33.519 --> 00:20:37.900
as having an intern and that that intern is going

00:20:37.900 --> 00:20:40.140
to try and gather information from lots of sources

00:20:40.140 --> 00:20:42.279
and then weigh up what is important and what

00:20:42.279 --> 00:20:45.500
isn't. And ultimately having... You know, that

00:20:45.500 --> 00:20:47.680
high brand awareness, that brand equity, that

00:20:47.680 --> 00:20:49.460
people speak about your brand favourably online

00:20:49.460 --> 00:20:53.440
in lots of different places. So it's very likely

00:20:53.440 --> 00:20:56.619
that if you've been good at marketing and you

00:20:56.619 --> 00:20:59.059
continue to be good at marketing, you're probably

00:20:59.059 --> 00:21:02.000
going to do quite well when it comes to the alien

00:21:02.000 --> 00:21:05.740
intern AI going off and doing some research and

00:21:05.740 --> 00:21:07.480
coming back and saying, I've looked into it.

00:21:08.250 --> 00:21:11.470
this product here has five star reviews is you

00:21:11.470 --> 00:21:14.549
know is mentioned in lots of case studies etc

00:21:14.549 --> 00:21:18.970
etc so yeah i think i suspect where most people

00:21:18.970 --> 00:21:21.490
will go hunting is for where the arbitra the

00:21:21.490 --> 00:21:24.130
early arbitrages are like you said if if reddit

00:21:24.130 --> 00:21:27.009
was being disproportionately used to drive results

00:21:27.829 --> 00:21:29.730
and you go all out on Reddit and try and make

00:21:29.730 --> 00:21:31.849
sure your brand dominates in Reddit, then probably

00:21:31.849 --> 00:21:34.230
for a month or two, maybe you are doing much

00:21:34.230 --> 00:21:37.049
better. And I suspect that is the bit that everybody's

00:21:37.049 --> 00:21:39.769
hunting to try and capture. But like you said,

00:21:39.829 --> 00:21:43.710
if it moves that quickly, you may invest a load

00:21:43.710 --> 00:21:46.529
of time and energy optimising for something that's

00:21:46.529 --> 00:21:49.220
a short -term hack. that doesn't actually then

00:21:49.220 --> 00:21:52.039
deliver long -term value do you get the value

00:21:52.039 --> 00:21:54.500
for money of trying to you know push too hard

00:21:54.500 --> 00:21:58.180
or should you white hat style invest in all the

00:21:58.180 --> 00:22:00.299
good marketing practices that you either get

00:22:00.299 --> 00:22:02.680
good results over the long term and will probably

00:22:02.680 --> 00:22:04.559
get you good results when it comes to being mentioned

00:22:04.559 --> 00:22:07.200
in ai answers yeah i think if you've been in

00:22:07.200 --> 00:22:12.099
the seo game for the last well forever uh you've

00:22:12.099 --> 00:22:19.380
seen this story before That being said, it's

00:22:19.380 --> 00:22:21.460
quite nice when you have one single platform.

00:22:21.559 --> 00:22:23.920
Obviously, we're HubSpot users as well. Our website

00:22:23.920 --> 00:22:28.680
is built on HubSpot. And it's nice when you have

00:22:28.680 --> 00:22:30.359
one tool to rule them all. It certainly does

00:22:30.359 --> 00:22:32.380
make your lives a bit easier. So I'm all for

00:22:32.380 --> 00:22:34.440
HubSpot adding more and more capabilities that

00:22:34.440 --> 00:22:36.839
just make it easier for us to do holistic marketing

00:22:36.839 --> 00:22:39.900
and track performance and results. Yeah, I think

00:22:39.900 --> 00:22:42.880
having this baked in, because the product that

00:22:42.880 --> 00:22:48.079
they've got in terms of its capabilities x funnel

00:22:48.079 --> 00:22:52.319
will search across all of the major search platforms

00:22:52.319 --> 00:22:54.119
so it will look on claude it will look on chat

00:22:54.119 --> 00:22:57.960
gpt it will look on grok etc etc so it's got

00:22:57.960 --> 00:23:00.680
some really nice dashboards and reports i assume

00:23:00.680 --> 00:23:02.519
they're going to bring in these core elements

00:23:02.519 --> 00:23:05.960
into the to the product but i expect as much

00:23:05.960 --> 00:23:08.480
as anything given that the product itself isn't

00:23:08.480 --> 00:23:12.180
that old i dare say this is as much a talent

00:23:12.180 --> 00:23:15.539
acquisition as anything else um but yeah i think

00:23:16.630 --> 00:23:19.890
HubSpot with their shift towards what they're

00:23:19.890 --> 00:23:24.069
calling loop marketing, which is a kind of progression

00:23:24.069 --> 00:23:28.190
away from inbound marketing of attract, convert,

00:23:28.309 --> 00:23:31.089
close, delight, or whatever that ended up becoming

00:23:31.089 --> 00:23:34.789
attract, engage, delight. I think it was with

00:23:34.789 --> 00:23:36.849
the flywheel. They're now talking about loop

00:23:36.849 --> 00:23:41.170
marketing and this very much feeds in into that.

00:23:41.390 --> 00:23:44.190
Now, whether you think loop marketing as a methodology

00:23:44.190 --> 00:23:47.190
is good or not. We can have that conversation

00:23:47.190 --> 00:23:52.430
over on LinkedIn, but maybe one for another day,

00:23:52.470 --> 00:23:55.390
Paul. No, we're having it here. There's two things.

00:23:56.190 --> 00:23:58.269
You've triggered me now. The first thing is,

00:23:58.349 --> 00:23:59.829
and I don't want this to be confirmation bias,

00:24:00.029 --> 00:24:01.829
but I saw a really interesting post on LinkedIn

00:24:01.829 --> 00:24:06.650
this week of someone highlighting the amount

00:24:06.650 --> 00:24:08.809
of traffic for a bunch of websites they managed

00:24:08.809 --> 00:24:11.309
that was coming from ChatGPT and other AI sources

00:24:11.309 --> 00:24:13.950
have basically bottomed out. So obviously it

00:24:13.950 --> 00:24:16.109
started at naught. went up to like two, three,

00:24:16.230 --> 00:24:19.789
5 % as a source and then didn't move over the

00:24:19.789 --> 00:24:22.910
last few months. And that chat GPT was by far

00:24:22.910 --> 00:24:25.089
the biggest referrer of traffic compared to clawed

00:24:25.089 --> 00:24:27.829
perplexity and stuff. Take of that what you will.

00:24:28.490 --> 00:24:30.470
Obviously my confirmation bias was like, yeah,

00:24:30.529 --> 00:24:32.829
see, see, it may not be a massive source yet

00:24:32.829 --> 00:24:36.009
because they answer, they provide you with answers

00:24:36.009 --> 00:24:38.730
in the chat. They don't give you links to go

00:24:38.730 --> 00:24:40.930
click, which of course doesn't mean that people

00:24:40.930 --> 00:24:43.420
don't go Google stuff and then you can't. track

00:24:43.420 --> 00:24:45.680
that through and you know i take it that that's

00:24:45.680 --> 00:24:48.819
a bit of a a dark art to try and figure out that

00:24:48.819 --> 00:24:52.339
that traffic and the second thing is hubspot

00:24:52.339 --> 00:24:56.079
you gotta love them they did so well coining

00:24:56.079 --> 00:24:59.640
a methodology inbound marketing basically you

00:24:59.640 --> 00:25:02.140
know create a new way of marketing that's so

00:25:02.140 --> 00:25:04.259
tightly associated with their commercial brand

00:25:04.259 --> 00:25:07.180
genius right of course it was genius however

00:25:07.180 --> 00:25:09.819
they're now like Do you know how we were really

00:25:09.819 --> 00:25:12.180
successful? We came up with this cool methodology.

00:25:12.420 --> 00:25:14.400
We got everyone to do it and then we basically

00:25:14.400 --> 00:25:17.440
completely tied it to our brand. Now we need

00:25:17.440 --> 00:25:20.099
a new methodology so that we can tie it to our

00:25:20.099 --> 00:25:25.019
brand. Enter Loop Marketing. Oh, what's old is

00:25:25.019 --> 00:25:28.279
new and give it a name and all this other stuff.

00:25:28.400 --> 00:25:30.140
So I'm a bit cynical on that stuff. I know you

00:25:30.140 --> 00:25:31.640
are as well, Martin. You must tell us your real

00:25:31.640 --> 00:25:34.359
feelings about Loop. Can you do it? I just do

00:25:34.359 --> 00:25:39.509
not buy it at all. First of all, I think my biggest

00:25:39.509 --> 00:25:44.650
issue is with it. The previous iterations of

00:25:44.650 --> 00:25:49.650
attract, engage, delight, and attract, convert,

00:25:49.849 --> 00:25:52.049
close, delight, what have you, were all related

00:25:52.049 --> 00:25:55.190
and aligned to different areas of the customer

00:25:55.190 --> 00:25:58.210
journey. You were moving someone through stages

00:25:58.210 --> 00:26:00.190
of they buy, whether it was the awareness stage,

00:26:00.410 --> 00:26:03.089
the consideration stage, the decision making

00:26:03.089 --> 00:26:07.640
stage, you were aligning that. The loop marketing

00:26:07.640 --> 00:26:10.799
piece moves away from that. So we don't get into

00:26:10.799 --> 00:26:15.299
the whole customer. It's not customer focused.

00:26:15.420 --> 00:26:19.240
It's how you, it's an internal methodology. It's

00:26:19.240 --> 00:26:23.460
how you execute marketing internally. So you

00:26:23.460 --> 00:26:28.099
find your brand voice, you create and amplify,

00:26:28.140 --> 00:26:32.700
and then you iterate and evolve. I haven't even

00:26:32.700 --> 00:26:35.450
got the whole thing right there. such as my disinterest

00:26:35.450 --> 00:26:37.869
in it but it's about how you internally execute

00:26:37.869 --> 00:26:42.369
and it's so far away so far removed from the

00:26:42.369 --> 00:26:45.809
customer which used to be the central point of

00:26:45.809 --> 00:26:49.450
the whole method that i find it a little bit

00:26:49.450 --> 00:26:51.809
i find it a bit jarring if i'm honest it feels

00:26:51.809 --> 00:26:55.950
like such a shift in terms of hubspot's tone

00:26:55.950 --> 00:26:57.910
that's super interesting when you put it that

00:26:57.910 --> 00:26:59.809
way because i think one of the things that made

00:26:59.809 --> 00:27:03.829
hubspot so successful was identifying how important

00:27:03.829 --> 00:27:05.930
it was to put the customer at the center of everything

00:27:05.930 --> 00:27:11.049
that you do and how to push brands more towards

00:27:11.049 --> 00:27:14.569
doing that so then to pivot into a model which

00:27:14.569 --> 00:27:17.369
doesn't do that i mean how long before they put

00:27:17.369 --> 00:27:20.230
the ai at the center of everything that'll be

00:27:20.230 --> 00:27:22.150
the next model it's probably four or five years

00:27:22.150 --> 00:27:24.250
away and it's when you have to optimize for ai

00:27:24.250 --> 00:27:29.250
agents instead of humans it's coming it's absolutely

00:27:29.250 --> 00:27:31.710
it's just around the corner i can i think i can

00:27:32.029 --> 00:27:33.829
Smell it in the air. Do you know what, I was

00:27:33.829 --> 00:27:35.849
doing some work this week just thinking about

00:27:35.849 --> 00:27:39.250
scenario planning for, you know, if you're a

00:27:39.250 --> 00:27:41.410
marketer over the next five years that you run

00:27:41.410 --> 00:27:42.829
a marketing team, you're a marketing director,

00:27:42.990 --> 00:27:46.369
let's say, what should you do if AI never gets

00:27:46.369 --> 00:27:49.250
any better? Right through to what should you

00:27:49.250 --> 00:27:54.369
do if... Companies become swarms of AI agents

00:27:54.369 --> 00:27:57.250
where there's like five human employees and a

00:27:57.250 --> 00:28:00.130
thousand AI agent employees. And this is mostly

00:28:00.130 --> 00:28:02.329
driven by the book that we've been having a bit

00:28:02.329 --> 00:28:03.789
of a reading club, haven't we, Mark? We've been

00:28:03.789 --> 00:28:07.309
reading The Last Economy by Imad Mustaq, which

00:28:07.309 --> 00:28:10.230
is super interesting and it kind of drove my

00:28:10.230 --> 00:28:15.450
thinking here. He paints a picture where it's

00:28:15.450 --> 00:28:18.980
not unreasonable to expect AI agents. running

00:28:18.980 --> 00:28:24.059
most of commercial organizations um so i've just

00:28:24.059 --> 00:28:25.920
been i've been thinking about a lot of this stuff

00:28:25.920 --> 00:28:31.680
and it's there's such a wide variety of possibilities

00:28:31.680 --> 00:28:34.200
actually on the technology front because it's

00:28:34.200 --> 00:28:37.039
moving so quickly that the biggest bottleneck

00:28:37.039 --> 00:28:40.839
is probably going to be human change and adoption

00:28:40.839 --> 00:28:42.400
we talked about this a little bit on the last

00:28:42.400 --> 00:28:45.420
podcast when we were talking about training but

00:28:45.420 --> 00:28:49.190
i was just wondering What sort of futures are

00:28:49.190 --> 00:28:54.029
you preparing for? Like in your mind, what does

00:28:54.029 --> 00:28:56.369
marketing as a function look like over the next

00:28:56.369 --> 00:29:00.170
couple of years? This is a question that people

00:29:00.170 --> 00:29:04.309
ask in the workshops that I deliver quite a lot.

00:29:04.730 --> 00:29:10.250
And I really struggle to have an answer because

00:29:10.250 --> 00:29:14.549
it's so difficult to know. The scenarios vary

00:29:14.549 --> 00:29:18.920
so wildly. ai agents become a thing in the way

00:29:18.920 --> 00:29:23.319
that they are described autonomous goal -oriented

00:29:23.319 --> 00:29:28.259
on 24 7 able to take action if that world transpires

00:29:28.259 --> 00:29:31.920
in say three to five years that's a dramatically

00:29:31.920 --> 00:29:35.940
different world from one in which they don't

00:29:35.940 --> 00:29:38.059
really happen and actually when we refer to ai

00:29:38.059 --> 00:29:41.559
agents what we're really talking about is ai

00:29:41.559 --> 00:29:47.519
agentic workflows that are 80 built by humans

00:29:47.519 --> 00:29:50.599
with a few steps where agents get involved along

00:29:50.599 --> 00:29:54.240
the way and so they're wildly different and i

00:29:54.240 --> 00:29:58.440
think it also depends entirely on how that agentic

00:29:58.440 --> 00:30:02.720
capability manifests at the consumer level as

00:30:02.720 --> 00:30:08.160
well we've seen google at the developer conference

00:30:08.160 --> 00:30:12.440
recently sharing ai mode with the ability to

00:30:12.440 --> 00:30:16.019
make purchases so you can do your search you

00:30:16.019 --> 00:30:19.099
can research a handbag it will bring up a list

00:30:19.099 --> 00:30:21.900
of handbags and then you can you can say buy

00:30:21.900 --> 00:30:26.180
that and they're they're creating these ai commerce

00:30:26.180 --> 00:30:29.500
platforms that you can integrate with and paypal

00:30:29.500 --> 00:30:32.319
is integrating with with chat gpt as well so

00:30:32.319 --> 00:30:37.819
in terms of what we can say is happening i think

00:30:37.819 --> 00:30:40.359
marketers are going to need to make sure that

00:30:40.359 --> 00:30:44.930
whatever they're routes to market are they have

00:30:44.930 --> 00:30:50.730
got those optimized for ai commerce in the next

00:30:50.730 --> 00:30:53.309
12 months that needs needs to happen you need

00:30:53.309 --> 00:30:57.309
to enable your route to market to be able to

00:30:57.309 --> 00:31:01.670
sell through an ai agent at a basic level as

00:31:01.670 --> 00:31:05.109
just described beyond that where people are autonomously

00:31:05.109 --> 00:31:08.650
sending out agents to run 24 hours a day seven

00:31:08.650 --> 00:31:13.150
days a week that feels so far away When you try

00:31:13.150 --> 00:31:18.250
and use an agent over a long time horizon, you

00:31:18.250 --> 00:31:20.470
just get so many errors and frustrations. It

00:31:20.470 --> 00:31:23.549
just feels like another world. If that actually

00:31:23.549 --> 00:31:27.089
can happen, then I'm not the person to answer

00:31:27.089 --> 00:31:29.990
that question because at that point we are reshaping

00:31:29.990 --> 00:31:32.289
the future. I think that's a really good answer.

00:31:32.289 --> 00:31:34.109
Seeing as I put you on the spot, that's a cracking

00:31:34.109 --> 00:31:37.710
answer. Shall I take you through the scenarios

00:31:37.710 --> 00:31:39.930
that I mapped out? I would love to hear them.

00:31:40.349 --> 00:31:42.609
Alrighty. So I think the first thing with this

00:31:42.609 --> 00:31:45.470
is I was doing a brainstorm here with AI, which

00:31:45.470 --> 00:31:47.950
I guess feels a bit meta and potentially somewhat

00:31:47.950 --> 00:31:50.609
biased. But it was it was usually ideas I had

00:31:50.609 --> 00:31:52.609
that I wanted AI to critique. And it did a really

00:31:52.609 --> 00:31:54.609
good job. And if you if you don't do that, dear

00:31:54.609 --> 00:31:57.609
listener, very often, it is amazing. If you analyze

00:31:57.609 --> 00:31:59.329
a particular situation or you come up with some

00:31:59.329 --> 00:32:03.410
ideas, trying. encourage the sycophantic AI not

00:32:03.410 --> 00:32:05.230
to be a sycophant and to actually critically

00:32:05.230 --> 00:32:08.170
assess your ideas and find the holes. Because

00:32:08.170 --> 00:32:10.349
then you can plug them or change your idea or

00:32:10.349 --> 00:32:11.829
what have you. So while I was going through this

00:32:11.829 --> 00:32:15.009
process, I was mindful that the timelines I was

00:32:15.009 --> 00:32:17.829
working on were three to five years. But I do

00:32:17.829 --> 00:32:21.109
think this, we will reach one of the later scenarios,

00:32:21.150 --> 00:32:24.390
but it just might take 10 years instead of three,

00:32:24.509 --> 00:32:26.950
five. And so I think as you listen to this, I

00:32:26.950 --> 00:32:28.349
want you to keep that in mind. And then I want

00:32:28.349 --> 00:32:31.410
you to be the human critique agent. and you can

00:32:31.410 --> 00:32:33.890
see if you can pull these apart so scenario number

00:32:33.890 --> 00:32:36.730
one is basically business as usual it's where

00:32:36.730 --> 00:32:39.029
we're at now for whatever reason hallucinations

00:32:39.029 --> 00:32:41.990
are still rife agents can't be trusted to do

00:32:41.990 --> 00:32:44.289
anything because they just make too many compounding

00:32:44.289 --> 00:32:47.269
errors and we're left adopting the tech that

00:32:47.269 --> 00:32:51.509
we have now and it does help improve speed and

00:32:51.509 --> 00:32:54.440
efficiency in areas by allowing us to create

00:32:54.440 --> 00:32:57.759
first drafts of content, start to do some smart

00:32:57.759 --> 00:33:00.359
analysis of different channels and how well they're

00:33:00.359 --> 00:33:03.559
performing so that humans can optimize strategies

00:33:03.559 --> 00:33:06.680
and place different bets and etc. But you're

00:33:06.680 --> 00:33:09.720
getting not much more than the efficiency gains

00:33:09.720 --> 00:33:12.539
that you're getting now. And you're doing this

00:33:12.539 --> 00:33:17.700
also in a sea of fears around AIs going rogue

00:33:17.700 --> 00:33:19.420
and making mistakes. I mean, we've already seen

00:33:19.420 --> 00:33:22.259
those examples, right? chatbots providing information

00:33:22.259 --> 00:33:25.460
that they shouldn't and and um and companies

00:33:25.460 --> 00:33:28.339
being held accountable for what ai chatbots say

00:33:28.339 --> 00:33:31.440
right just those types of compliance fears um

00:33:31.440 --> 00:33:35.460
which probably means how you're marketing now

00:33:35.460 --> 00:33:39.140
is how you market for quite a while i don't think

00:33:39.140 --> 00:33:40.839
that's going to be the scenario but that's kind

00:33:40.839 --> 00:33:47.220
of scenario one scenario two is ai is more broadly

00:33:47.220 --> 00:33:51.059
used And it starts to augment campaigns and it's

00:33:51.059 --> 00:33:54.559
more trustworthy for delivering on certain aspects

00:33:54.559 --> 00:33:58.700
of work. So there might be small niche agents

00:33:58.700 --> 00:34:01.279
that can do stuff that you can rely on without

00:34:01.279 --> 00:34:04.619
having to micromanage them. And this is interesting

00:34:04.619 --> 00:34:06.799
because even though the humans are still the

00:34:06.799 --> 00:34:09.280
strategists, maybe this conductor role that we've

00:34:09.280 --> 00:34:13.519
talked about in the past, the ability to do much

00:34:13.519 --> 00:34:16.139
more at scale becomes significantly increased

00:34:16.139 --> 00:34:19.300
when. the first draft of content needs barely

00:34:19.300 --> 00:34:22.820
any edits or you can set an ai agent to go in

00:34:22.820 --> 00:34:27.820
you know maybe p max max max max is able to do

00:34:27.820 --> 00:34:29.340
a little bit what you said about meta earlier

00:34:29.340 --> 00:34:31.280
like give us a goal give us a credit card and

00:34:31.280 --> 00:34:34.539
we'll optimize for that um then i think you start

00:34:34.539 --> 00:34:39.800
to see a real speed increase um i want to believe

00:34:39.800 --> 00:34:42.900
that marketing teams stay the same size and do

00:34:42.900 --> 00:34:45.079
five times as much work and drive that type of

00:34:45.079 --> 00:34:46.989
commercial growth but of course That might not

00:34:46.989 --> 00:34:48.750
be the case. I do think that will be, though.

00:34:49.730 --> 00:34:52.050
I think under these conditions, as an individual,

00:34:52.289 --> 00:34:54.889
this is where you're like, I need to make sure

00:34:54.889 --> 00:34:59.030
that I am learning how to leverage these systems

00:34:59.030 --> 00:35:06.329
to scale myself. But there's still error. There's

00:35:06.329 --> 00:35:08.210
no set and forget. It's human in the loop in

00:35:08.210 --> 00:35:10.269
a lot of different areas. And it's just those

00:35:10.269 --> 00:35:12.670
niche things. And I can imagine this is where

00:35:12.670 --> 00:35:16.909
having a HubSpot that has 15 different ai products

00:35:16.909 --> 00:35:20.809
that allow you to build great email campaigns

00:35:20.809 --> 00:35:24.630
or um automate your social media publishing but

00:35:24.630 --> 00:35:26.429
with a human review step before anything goes

00:35:26.429 --> 00:35:28.070
live and stuff like that what i think will be

00:35:28.070 --> 00:35:32.510
quite critical um because i think you'll be living

00:35:32.510 --> 00:35:34.769
in the marketing tools you've already got and

00:35:34.769 --> 00:35:37.070
getting better and faster because of bits that

00:35:37.070 --> 00:35:39.429
they do for you that you would have done yourself

00:35:39.429 --> 00:35:42.010
um you could argue that we're here already but

00:35:42.010 --> 00:35:44.440
i don't think we are like you can't You can't

00:35:44.440 --> 00:35:46.260
tell HubSpot I want to run an email marketing

00:35:46.260 --> 00:35:48.320
campaign and it builds five emails and you just

00:35:48.320 --> 00:35:50.119
like run your eyes over them. We just we're not

00:35:50.119 --> 00:35:54.280
there yet. In scenario three, it's kind of the

00:35:54.280 --> 00:35:58.159
same thing, but the scale up is being driven

00:35:58.159 --> 00:36:01.739
by personalization. So you told us a bit earlier

00:36:01.739 --> 00:36:04.840
about what Meta's being able to do with capturing

00:36:04.840 --> 00:36:07.579
more and more data about folks. And it's the

00:36:07.579 --> 00:36:12.309
ability for AI to segment audiences. in fine

00:36:12.309 --> 00:36:17.389
detail based on behavioral signals but we then

00:36:17.389 --> 00:36:20.469
have to use ai to create content at the scale

00:36:20.469 --> 00:36:22.590
to meet those people where they're at because

00:36:22.590 --> 00:36:24.550
it's almost like ai gives us ability to market

00:36:24.550 --> 00:36:26.789
to smaller groups but we don't want a hundred

00:36:26.789 --> 00:36:28.730
people trying to do that that wouldn't be very

00:36:28.730 --> 00:36:30.909
efficient or effective so ai does a lot of the

00:36:30.909 --> 00:36:34.789
work and again to be able to market to go from

00:36:34.789 --> 00:36:37.670
five segments to 500 you're going to be trusting

00:36:38.400 --> 00:36:42.159
the ai to run sam campaigns without human oversight

00:36:42.159 --> 00:36:44.500
right we're starting to get to the point where

00:36:44.500 --> 00:36:46.619
i feel like that's plausible in the next couple

00:36:46.619 --> 00:36:48.780
of years i think the biggest thing to watch out

00:36:48.780 --> 00:36:54.159
for is how well ai driven advertising really

00:36:54.159 --> 00:36:56.900
can leverage those signals so this meta how this

00:36:56.900 --> 00:36:58.780
develops with meta is going to be really interesting

00:36:58.780 --> 00:37:03.699
um and also the pushback from humans right personally

00:37:03.699 --> 00:37:07.809
i would rather have ads that are much more personalised

00:37:07.809 --> 00:37:09.929
to me, then I know I'm going to see that because

00:37:09.929 --> 00:37:11.309
that's how these platforms make their money.

00:37:11.389 --> 00:37:15.690
So I'd rather they were personalised. But not

00:37:15.690 --> 00:37:17.409
everybody feels like that. And you will have

00:37:17.409 --> 00:37:20.409
those scenarios. Do you, I don't know if this

00:37:20.409 --> 00:37:21.969
is an apocryphal tale at this point in time.

00:37:22.050 --> 00:37:25.909
Do you know that story about the supermarket

00:37:25.909 --> 00:37:30.070
club card style point system where somebody,

00:37:30.250 --> 00:37:32.409
yeah, I see you nodding. Tell me, do you know

00:37:32.409 --> 00:37:35.599
this story? yeah so it was it was from America

00:37:35.599 --> 00:37:37.880
wasn't it was like Target or one of those was

00:37:37.880 --> 00:37:42.780
supposed to have sent somebody the um the the

00:37:42.780 --> 00:37:45.940
pregnancy related catalog was it through the

00:37:45.940 --> 00:37:49.199
post and it was supposed to be that their buying

00:37:49.199 --> 00:37:52.760
purchases indicated that they were pregnant before

00:37:52.760 --> 00:37:56.420
they even knew they were pregnant right right

00:37:56.420 --> 00:37:59.760
and I think also they lived in a household with

00:37:59.760 --> 00:38:03.400
their wider family and therefore the other members

00:38:03.400 --> 00:38:05.579
of their family knew they were pregnant as well

00:38:05.579 --> 00:38:08.320
because this thing lands on the on the table

00:38:08.320 --> 00:38:10.619
and triggers that conversation that they didn't

00:38:10.619 --> 00:38:12.679
really want so there's obviously going to be

00:38:12.679 --> 00:38:16.340
a number of times where this doesn't work well

00:38:16.340 --> 00:38:18.059
and they're going to be really big news and i

00:38:18.059 --> 00:38:20.159
do think that that will somewhat hold some of

00:38:20.159 --> 00:38:24.260
this stuff back scenario four this is where we

00:38:24.260 --> 00:38:27.510
start to like overreach a little bit um And it

00:38:27.510 --> 00:38:29.210
becomes hard to imagine. But like I said at the

00:38:29.210 --> 00:38:30.909
beginning, I don't think that means it doesn't

00:38:30.909 --> 00:38:34.190
happen. I just think it maybe takes longer, which

00:38:34.190 --> 00:38:39.250
is AI can automate marketing workflows at scale

00:38:39.250 --> 00:38:46.010
on a near one to one level. And we go from like

00:38:46.010 --> 00:38:49.849
an AI conductor. that's managing like a copywriting

00:38:49.849 --> 00:38:53.929
agent and an ad management agent to like an AI

00:38:53.929 --> 00:38:57.050
business director that has a marketing agent

00:38:57.050 --> 00:39:01.230
that manages all those other agents. And marketing

00:39:01.230 --> 00:39:03.570
is borderline a function that's being driven

00:39:03.570 --> 00:39:08.510
by AI. AI is capturing data on customers, create

00:39:08.510 --> 00:39:11.050
a hypothesis about products that should be developed

00:39:11.050 --> 00:39:15.190
and influence the whole business stack, which

00:39:15.190 --> 00:39:18.500
becomes much harder. to imagine i think and then

00:39:18.500 --> 00:39:21.400
the last scenario is the one that's the hardest

00:39:21.400 --> 00:39:23.699
to imagine as you alluded to a moment ago mine

00:39:23.699 --> 00:39:27.880
which is kind of along the email stack lines

00:39:27.880 --> 00:39:31.719
for the uh you know for the next economy where

00:39:31.719 --> 00:39:34.320
agents are the entire business and there there

00:39:34.320 --> 00:39:38.099
may not even be a director of the agents the

00:39:38.099 --> 00:39:43.820
director is an agent themselves and as humans

00:39:43.820 --> 00:39:46.699
we might just be lucky to be the shareholders

00:39:46.699 --> 00:39:51.039
or whatever that are maybe giving the business

00:39:51.039 --> 00:39:53.239
overall goals about the markets we want to be

00:39:53.239 --> 00:39:54.679
in and what have you. But the agents are doing

00:39:54.679 --> 00:39:57.260
all the work. This, if you, I went down a rabbit

00:39:57.260 --> 00:40:00.659
hole with this one, mine, and you can start to

00:40:00.659 --> 00:40:04.139
imagine these weird agentic AI swarms that recruit

00:40:04.139 --> 00:40:06.179
other agents. It's almost like a gig economy

00:40:06.179 --> 00:40:09.420
for agents. Where the business itself is not

00:40:09.420 --> 00:40:12.179
like, oh, well, there were 100 agents and now

00:40:12.179 --> 00:40:14.880
there are 1 ,000 agents. It's like, no, there's

00:40:14.880 --> 00:40:17.960
probably just like one agent. And it recruits

00:40:17.960 --> 00:40:20.880
all these other agents when it needs them for

00:40:20.880 --> 00:40:23.599
certain aspects, whether that might be, you know,

00:40:23.599 --> 00:40:26.340
product strategy, manufacturing, because maybe

00:40:26.340 --> 00:40:28.719
this is being driven by robotics. And then you

00:40:28.719 --> 00:40:31.900
have an entire economy almost entirely driven

00:40:31.900 --> 00:40:35.300
by AI and robots. At which point as a marketer,

00:40:36.280 --> 00:40:39.500
you're probably not required let's be honest

00:40:39.500 --> 00:40:41.480
you probably weren't required in scenario four

00:40:41.480 --> 00:40:45.159
but by scenario five um you're hopefully not

00:40:45.159 --> 00:40:46.920
only you're not it's not a four -day week or

00:40:46.920 --> 00:40:49.400
three -day week it might be a 0 .5 day week and

00:40:49.400 --> 00:40:51.280
hopefully we have got this wonderful abundant

00:40:51.280 --> 00:40:55.840
future where we can play golf or knit or go to

00:40:55.840 --> 00:40:58.739
the beach or play call of duty whatever it is

00:40:58.739 --> 00:41:00.199
you want to do with your life and you don't actually

00:41:00.199 --> 00:41:02.280
necessarily have to worry so much about working

00:41:02.280 --> 00:41:05.760
for money to live but um I've dumped a lot of

00:41:05.760 --> 00:41:08.360
information on you there, Martin. Thoughts? Well,

00:41:08.420 --> 00:41:13.719
Scenario 5 is borderline singularity territory,

00:41:13.920 --> 00:41:16.679
isn't it? If marketing as a function is affected

00:41:16.679 --> 00:41:19.780
to that degree, so is pretty much every other

00:41:19.780 --> 00:41:23.199
business function at that point. And I think

00:41:23.199 --> 00:41:27.400
that is where some of the AI labs envisage this

00:41:27.400 --> 00:41:31.119
going. And instead of you having a payroll to

00:41:31.119 --> 00:41:33.929
your employees, you just... have all of your

00:41:33.929 --> 00:41:37.329
AI agents powered by board or whatever it may

00:41:37.329 --> 00:41:40.670
be. I really like the idea of them just kind

00:41:40.670 --> 00:41:43.010
of dipping in and out of other agents, so coordinating

00:41:43.010 --> 00:41:46.289
agents, not just your own internal agents, but

00:41:46.289 --> 00:41:48.650
pulling on external ones. And it's interesting

00:41:48.650 --> 00:41:50.230
if you think about what that would mean as a

00:41:50.230 --> 00:41:54.210
pivot for, say, software providers. Let's say

00:41:54.210 --> 00:41:59.389
you're, I don't know, MailChimp, and you have

00:41:59.389 --> 00:42:05.030
an... email marketing agent that company's internal

00:42:05.030 --> 00:42:07.530
marketing director agent can just kind of tap

00:42:07.530 --> 00:42:10.170
into and go, right, we've got this database,

00:42:10.389 --> 00:42:14.250
go create a campaign for us. Here's our creative

00:42:14.250 --> 00:42:19.570
from our Canva or Adobe agent. Execute that at

00:42:19.570 --> 00:42:25.510
scale across this segment. Yeah, I'd consider

00:42:25.510 --> 00:42:27.760
it. considered operating like that but i guess

00:42:27.760 --> 00:42:31.400
in some respects it's like mcp and api credit

00:42:31.400 --> 00:42:33.400
should be dipping in and out of the services

00:42:33.400 --> 00:42:36.719
as and when you need them i think the scenario

00:42:36.719 --> 00:42:40.360
that you describe so scenario four just just

00:42:40.360 --> 00:42:42.800
remind me what that one scenario four is the

00:42:42.800 --> 00:42:47.400
one where um we are effectively mostly driven

00:42:47.400 --> 00:42:50.260
by agents there's like a business director who's

00:42:50.260 --> 00:42:52.340
a human who's like okay we want to enter this

00:42:52.340 --> 00:42:55.309
market But the marketing director is an agent

00:42:55.309 --> 00:42:58.710
that directs a bunch of other agents. I think

00:42:58.710 --> 00:43:00.429
in some ways you can think of these scenarios

00:43:00.429 --> 00:43:04.449
as less human coordination of niche agents and

00:43:04.449 --> 00:43:07.230
the agents becoming the coordinators until they

00:43:07.230 --> 00:43:10.309
are the entire economy, basically. So in scenario

00:43:10.309 --> 00:43:14.030
four, there is still a CEO. There may even be

00:43:14.030 --> 00:43:15.989
some sort of human leadership team. But a lot

00:43:15.989 --> 00:43:19.030
of the planning strategy and execution is being

00:43:19.030 --> 00:43:21.210
done by the agents based on the goals set by

00:43:21.210 --> 00:43:25.610
the humans. I think that's certainly feasible

00:43:25.610 --> 00:43:29.789
looking a few years out, particularly when businesses

00:43:29.789 --> 00:43:32.550
start to reimagine their entire workflows and

00:43:32.550 --> 00:43:36.489
start thinking AI agent workflow first. If you

00:43:36.489 --> 00:43:38.750
look at the positioning with, say, Zapier at

00:43:38.750 --> 00:43:41.429
the moment, they're really talking about AI orchestration

00:43:41.429 --> 00:43:43.449
is the big thing they're talking about. How do

00:43:43.449 --> 00:43:47.260
you get AIs to orchestrate? other ai's using

00:43:47.260 --> 00:43:50.039
either predetermined workflows that maybe have

00:43:50.039 --> 00:43:53.760
an ai step in them or agentic workflows where

00:43:53.760 --> 00:43:56.800
a big chunk of the work is done by ai but then

00:43:56.800 --> 00:44:00.940
the following steps or the trigger might be initiated

00:44:00.940 --> 00:44:07.019
by a human um it requires rethinking the structure

00:44:07.019 --> 00:44:09.519
of the organization doesn't it it's a complete

00:44:09.519 --> 00:44:12.639
re -evaluation of the workload and the tasks

00:44:12.639 --> 00:44:16.670
done from start to finish Easier done from a

00:44:16.670 --> 00:44:20.449
startup. If you're a marketing agency today and

00:44:20.449 --> 00:44:24.389
you're thinking, right, we want to scale up and

00:44:24.389 --> 00:44:27.309
we're currently one person, five people, 10 people,

00:44:27.389 --> 00:44:30.190
whatever it may be. If you're a small agency,

00:44:30.489 --> 00:44:34.530
you can think about redesigning your workflows

00:44:34.530 --> 00:44:38.110
from the ground up. Much more difficult if you're

00:44:38.110 --> 00:44:42.250
a 500 person manufacturing business with long

00:44:42.250 --> 00:44:45.719
established workflows. And then you've got the

00:44:45.719 --> 00:44:48.480
threat then of other people in your industry

00:44:48.480 --> 00:44:51.179
going, well, actually we know how they manufacture

00:44:51.179 --> 00:44:54.599
their product. If we can do all of the, let's

00:44:54.599 --> 00:44:57.159
say the back office stuff, we can do the procurement,

00:44:57.179 --> 00:45:00.480
the finance, the marketing, the sales. We'll

00:45:00.480 --> 00:45:03.159
do that with agents, but we'll do the manufacturing

00:45:03.159 --> 00:45:07.559
as is. You've got a competitor who can do what

00:45:07.559 --> 00:45:09.940
you can do with 500 people, but they can do it

00:45:09.940 --> 00:45:13.559
with say 100. And that changes the game. I'm

00:45:13.559 --> 00:45:15.139
so glad you said that because there's two things

00:45:15.139 --> 00:45:17.139
that spring to mind. The first one is once you

00:45:17.139 --> 00:45:19.659
get up to scenario three and four, probably four.

00:45:21.079 --> 00:45:24.400
You are looking at AI first companies that eat

00:45:24.400 --> 00:45:26.960
the lunch of existing companies because retooling.

00:45:27.239 --> 00:45:29.280
I've gone through those things like that phase

00:45:29.280 --> 00:45:31.860
transitions easy. Oh, so it starts out and the

00:45:31.860 --> 00:45:33.900
agents do a bit of the work and then it's like

00:45:33.900 --> 00:45:37.360
by the end they do all of the work and. that's

00:45:37.360 --> 00:45:39.780
that's maybe fine if there's like 10 people in

00:45:39.780 --> 00:45:42.059
a company and they have to like adapt and they

00:45:42.059 --> 00:45:44.860
get a decent timeline to do it what if you're

00:45:44.860 --> 00:45:46.800
a thousand person company like that you know

00:45:46.800 --> 00:45:50.559
humans are likely to resist this for for obvious

00:45:50.559 --> 00:45:53.500
reasons plus the systems and processes of these

00:45:53.500 --> 00:45:56.380
businesses will have and their and their data

00:45:56.380 --> 00:45:59.079
and their tech stacks that are not augmented

00:45:59.079 --> 00:46:02.199
for this it's so much easier to make that leap

00:46:02.199 --> 00:46:05.809
if you just build it from scratch that way And

00:46:05.809 --> 00:46:08.670
in theory, those organizations would be so much

00:46:08.670 --> 00:46:12.369
more efficient and effective. They would eat

00:46:12.369 --> 00:46:15.809
the lunch of the big old businesses easily, right?

00:46:15.869 --> 00:46:18.570
They'd be like, oh, yeah, so we are able to achieve,

00:46:18.710 --> 00:46:22.429
you know, 10x revenue per employee as you can,

00:46:22.550 --> 00:46:26.510
as an example. So like, how quick does it take

00:46:26.510 --> 00:46:30.010
to out -compete those businesses under those

00:46:30.010 --> 00:46:33.820
conditions? Very, very easy to do. um one would

00:46:33.820 --> 00:46:36.679
assume the other thing that this forgets is the

00:46:36.679 --> 00:46:39.460
customers because if you're assuming that your

00:46:39.460 --> 00:46:44.599
customers are humans then you you need we talk

00:46:44.599 --> 00:46:46.920
the way we frame this is the efficiency gains

00:46:46.920 --> 00:46:50.000
etc and i and i do think you know hyper personalization

00:46:50.000 --> 00:46:52.219
of products and ads is obviously hopefully going

00:46:52.219 --> 00:46:54.719
to be better for the for the customers but there

00:46:54.719 --> 00:46:56.179
are a number of different things that can massively

00:46:56.179 --> 00:47:00.400
shatter these scenarios where humans value human

00:47:01.099 --> 00:47:04.079
products and output so much they're willing to

00:47:04.079 --> 00:47:07.780
pay much more for things that humans crafted

00:47:07.780 --> 00:47:10.559
and created it's like actually i don't i don't

00:47:10.559 --> 00:47:12.900
want to buy something that was all robotics and

00:47:12.900 --> 00:47:15.860
ai i want to work with a human like i want to

00:47:15.860 --> 00:47:17.739
speak to a human on the phone those things could

00:47:17.739 --> 00:47:20.619
emerge and completely disrupt all of this and

00:47:20.619 --> 00:47:23.800
so if the customers don't get a better experience

00:47:24.800 --> 00:47:27.340
this won't be as adopted as it sounds like it

00:47:27.340 --> 00:47:29.760
obviously should be because ultimately if it's

00:47:29.760 --> 00:47:32.119
still human spending the money then it's going

00:47:32.119 --> 00:47:34.780
to be all about customer experience so i think

00:47:34.780 --> 00:47:36.500
that's that's probably quite an interesting factor

00:47:36.500 --> 00:47:40.199
the other thing i dug a bit deeper into is how

00:47:40.199 --> 00:47:43.840
do we know where we are and what sort of things

00:47:43.840 --> 00:47:46.860
should we look for if we want to be go if we

00:47:46.860 --> 00:47:49.539
want to say oh scenario two's in the often should

00:47:49.539 --> 00:47:52.920
probably start planning ahead with that in mind

00:47:52.920 --> 00:47:55.420
so i had quite a good conversation with chat

00:47:55.420 --> 00:47:57.400
gbt about that as well and we started sort of

00:47:57.400 --> 00:48:02.500
digging into you know where are we um i'm not

00:48:02.500 --> 00:48:04.260
going to describe now i think we all know where

00:48:04.260 --> 00:48:06.940
we are right now but if we will know if things

00:48:06.940 --> 00:48:10.440
are stalling if there's more indications that

00:48:10.440 --> 00:48:14.900
this is a ai bubble and um you know earnings

00:48:14.900 --> 00:48:18.480
of ai companies don't accelerate and money stops

00:48:18.480 --> 00:48:21.780
going into the sector maybe regulation like the

00:48:21.780 --> 00:48:26.559
eu ai act just makes it too hard to start personalizing

00:48:26.559 --> 00:48:28.960
at scale and stuff like that would be quite useful

00:48:28.960 --> 00:48:33.579
um but we'll be you know we will be moving into

00:48:33.579 --> 00:48:38.260
scenario two if we start to see niche agents

00:48:38.260 --> 00:48:42.500
that can actually deliver on a particular task

00:48:42.500 --> 00:48:45.849
without any hand -holding like If HubSpot introduces

00:48:45.849 --> 00:48:51.550
a social media posting agent where all you do

00:48:51.550 --> 00:48:54.369
is review the copy before it goes out, we're

00:48:54.369 --> 00:48:58.929
starting to drift into scenario one. If we start

00:48:58.929 --> 00:49:02.550
to see Meta's pilots start to be successful,

00:49:02.670 --> 00:49:05.869
if we see those case studies that show we're

00:49:05.869 --> 00:49:10.079
able to gather leverage. and advertise based

00:49:10.079 --> 00:49:12.920
on the data in these platforms and the humans

00:49:12.920 --> 00:49:18.460
don't cry, no, you've destroyed my privacy. We're

00:49:18.460 --> 00:49:22.360
starting to drift into, you know, scenario three

00:49:22.360 --> 00:49:25.440
at that point. And from there, it's just about

00:49:25.440 --> 00:49:29.119
how much better the agents get and what roles

00:49:29.119 --> 00:49:32.539
the humans continue to play. It's interesting

00:49:32.539 --> 00:49:35.900
you used the example of the HubSpot social...

00:49:36.539 --> 00:49:38.840
agent that is actually an agent that they've

00:49:38.840 --> 00:49:41.719
got in the offering at the moment a social posting

00:49:41.719 --> 00:49:46.500
agent which will create the content for you based

00:49:46.500 --> 00:49:50.400
on some keywords that's quite limited but we're

00:49:50.400 --> 00:49:52.260
kind of at day one of this technology i think

00:49:52.260 --> 00:49:56.579
literally this agent was launched in beta like

00:49:56.579 --> 00:49:59.760
three months ago maybe not even that long so

00:49:59.760 --> 00:50:07.329
really early days and at the limited use cases

00:50:07.329 --> 00:50:10.369
but you can absolutely see as they become more

00:50:10.369 --> 00:50:14.909
deeply integrated into a platform like hubspot

00:50:14.909 --> 00:50:17.489
is you're all in one powwow so it has your social

00:50:17.489 --> 00:50:20.650
posting it has all of your customer data where

00:50:20.650 --> 00:50:25.969
you have fully integrated systems agents in that

00:50:25.969 --> 00:50:28.849
environment should thrive and that should give

00:50:28.849 --> 00:50:33.289
companies like hubspot or salesforce and other

00:50:33.289 --> 00:50:38.699
platforms a huge advantage because when they

00:50:38.699 --> 00:50:42.239
can both go off and do a small task like oh research

00:50:42.239 --> 00:50:44.880
this topic and create some social posts but then

00:50:44.880 --> 00:50:48.880
also manage the approval workflow the actual

00:50:48.880 --> 00:50:52.360
schedule scheduling optimization and reporting

00:50:52.360 --> 00:50:56.500
of all of that workflow then things start to

00:50:56.500 --> 00:51:00.420
get um yeah very stage two as you you would say

00:51:00.420 --> 00:51:03.659
yeah i think that's the key and um if you're

00:51:03.659 --> 00:51:06.010
listening to this You've either been thinking

00:51:06.010 --> 00:51:07.670
about this a lot or read some of this stuff already

00:51:07.670 --> 00:51:09.210
and you're like, okay, that sounds interesting.

00:51:09.329 --> 00:51:12.329
But, you know, we'll see how it plays out. Others

00:51:12.329 --> 00:51:13.869
will be like, we've just scared the bejeebus

00:51:13.869 --> 00:51:17.949
out of them. You know, let's not forget the data

00:51:17.949 --> 00:51:21.110
centers that are being built right now are to

00:51:21.110 --> 00:51:24.940
serve AI inference. that's agent swarm level

00:51:24.940 --> 00:51:27.460
basically right that's where all these big build

00:51:27.460 --> 00:51:30.280
outs are we're all going to want way more ai

00:51:30.280 --> 00:51:32.980
access than we currently use and you might be

00:51:32.980 --> 00:51:35.179
thinking to yourself well i use chachi pt quite

00:51:35.179 --> 00:51:37.139
a lot every day already why do we need these

00:51:37.849 --> 00:51:40.170
data's centers and it's because of this right

00:51:40.170 --> 00:51:42.670
so that kind of shows you what they're they're

00:51:42.670 --> 00:51:44.909
betting on there's a guy you and i follow on

00:51:44.909 --> 00:51:47.789
the linkedins and on the whatsapps called ruv

00:51:47.789 --> 00:51:52.869
martin who deploys agent swarms for coding requirements

00:51:52.869 --> 00:51:55.849
and has shipped ridiculous amounts of working

00:51:55.849 --> 00:51:59.989
code over the last 12 months by building development

00:51:59.989 --> 00:52:04.929
agent swarms that collaborate together um so

00:52:04.929 --> 00:52:08.269
these swarm things this is not like science fiction

00:52:08.269 --> 00:52:11.550
idea that martin and i just found on a reddit

00:52:11.550 --> 00:52:13.989
forum somewhere like people are building these

00:52:13.989 --> 00:52:16.130
not for marketing but for other use cases and

00:52:16.130 --> 00:52:19.630
so you know it's not unrealistic to think that

00:52:19.630 --> 00:52:23.610
they will emerge so if we've bamboozled you and

00:52:23.610 --> 00:52:25.670
hopefully we haven't but there are some things

00:52:25.670 --> 00:52:29.230
that you can do and should do regardless of which

00:52:29.230 --> 00:52:33.309
scenario comes to pass um some of the easy things

00:52:33.309 --> 00:52:37.219
that you can do is absolutely if you're an individual

00:52:37.219 --> 00:52:41.760
investing in your own ai training and becoming

00:52:41.760 --> 00:52:44.619
you know an expert in these things is critical

00:52:44.619 --> 00:52:46.820
and if you run a team invest in their training

00:52:46.820 --> 00:52:50.420
because the the old adage it is old now i think

00:52:50.420 --> 00:52:53.320
but you know um it's not marketers aren't replaced

00:52:53.320 --> 00:52:55.760
by ai but they are replaced by marketers that

00:52:55.760 --> 00:53:00.960
use ai um and so that's critical i think getting

00:53:00.960 --> 00:53:05.449
your customer data in order so There is definitely

00:53:05.449 --> 00:53:08.090
a world where privacy clampdown makes this hard

00:53:08.090 --> 00:53:11.130
and that you can only rely on the first party

00:53:11.130 --> 00:53:13.849
data that you have about customers. So how can

00:53:13.849 --> 00:53:19.510
you secure opt -ins, gather data in a legal and

00:53:19.510 --> 00:53:22.590
fair way and structure and store it in a way

00:53:22.590 --> 00:53:25.849
that it's easy for AI models to draw upon as

00:53:25.849 --> 00:53:28.329
inspiration and insight to help drive some of

00:53:28.329 --> 00:53:31.050
these smart personalizations and automations

00:53:31.050 --> 00:53:33.840
that we predict will happen in the future. In

00:53:33.840 --> 00:53:35.519
line with that, you probably do need some sort

00:53:35.519 --> 00:53:41.840
of AI ethics and usage charter, both to give

00:53:41.840 --> 00:53:44.579
your customers trust in how you use AI and data,

00:53:44.699 --> 00:53:46.820
but also to give guidance to your employees about

00:53:46.820 --> 00:53:50.699
how they can use AI and data. And then starting

00:53:50.699 --> 00:53:52.820
to build out some of those human in the loop

00:53:52.820 --> 00:53:55.300
QA processes. Most people are starting to do

00:53:55.300 --> 00:53:57.280
this. I think now this we talked previously,

00:53:57.420 --> 00:54:00.320
Martin, about. work slop and how ai is brilliant

00:54:00.320 --> 00:54:02.340
at writing stuff that makes no sense and how

00:54:02.340 --> 00:54:04.039
important it is to have humans in the loop but

00:54:04.039 --> 00:54:06.980
looking at how you can start to smart automate

00:54:06.980 --> 00:54:09.440
aspects of processes with human touch points

00:54:09.440 --> 00:54:12.199
you do this really beautifully martin and a number

00:54:12.199 --> 00:54:14.300
of the automation tools that you've been using

00:54:14.300 --> 00:54:17.079
but having those processes in hand will allow

00:54:17.079 --> 00:54:19.539
you to maintain the quality that you need but

00:54:19.539 --> 00:54:23.239
also Figure out where you can deploy an AI agent

00:54:23.239 --> 00:54:26.340
as a particular NIS agent emerges. It's like,

00:54:26.400 --> 00:54:27.880
oh, that bit doesn't have to be a human anymore.

00:54:28.159 --> 00:54:32.840
It can be an agent instead. And then just be

00:54:32.840 --> 00:54:35.079
testing tools or listening to podcasts like this

00:54:35.079 --> 00:54:38.420
where we test the tools to figure out what those

00:54:38.420 --> 00:54:41.760
capabilities look like as they actually emerge.

00:54:43.440 --> 00:54:46.840
Things like look out for things like the AI coordinator.

00:54:47.159 --> 00:54:50.949
If an agent appears. that can coordinate other

00:54:50.949 --> 00:54:54.130
agents that is going to be a major trigger point

00:54:54.130 --> 00:54:55.949
that everybody needs to keep an eye on and the

00:54:55.949 --> 00:54:59.469
early version of that will probably be not dissimilar

00:54:59.469 --> 00:55:01.650
to one of the use cases martin told us all about

00:55:01.650 --> 00:55:05.150
where he was using manus to construct online

00:55:05.150 --> 00:55:09.250
proposals where manus came up with the idea based

00:55:09.250 --> 00:55:11.889
on some information martin gave it and built

00:55:11.889 --> 00:55:15.010
this dynamic online experience it will probably

00:55:15.010 --> 00:55:17.809
be these computer use agents that do some of

00:55:17.809 --> 00:55:21.099
the first coordination of tasks as that gets

00:55:21.099 --> 00:55:24.300
better expect to see true agents that can deploy

00:55:24.300 --> 00:55:26.980
these things so um they're the things to watch

00:55:26.980 --> 00:55:31.159
out for martin any thoughts on that big takeaway

00:55:31.159 --> 00:55:34.360
for me is this is going to look very different

00:55:34.360 --> 00:55:39.059
depending on the size of your organization and

00:55:39.059 --> 00:55:42.800
the markets that you serve we talked about ai

00:55:42.800 --> 00:55:45.599
managed campaigns that does hyper personalization

00:55:45.599 --> 00:55:52.139
at scale great if you're a huge b2c consumer

00:55:52.139 --> 00:55:55.139
brand with deep pockets then there's a huge amount

00:55:55.139 --> 00:55:57.900
that you can go and you know facebook's massive

00:55:57.900 --> 00:56:02.260
data set allows it to to test and orchestrate

00:56:02.260 --> 00:56:06.119
ai campaigns at scale and you can kind of really

00:56:06.119 --> 00:56:09.800
benefit from from that now and in the future

00:56:09.800 --> 00:56:11.519
if you've got your own first party data it will

00:56:11.519 --> 00:56:14.079
serve you well that's going to look very different

00:56:14.079 --> 00:56:19.119
if you're a niche b2b manufacturing and engineering

00:56:19.119 --> 00:56:21.559
company that realistically is speaking to an

00:56:21.559 --> 00:56:25.340
audience of say 50 other companies in the world

00:56:25.340 --> 00:56:30.940
so how you think about this is is going to change

00:56:30.940 --> 00:56:34.940
i think if you are on that more niche b2b side

00:56:34.940 --> 00:56:39.219
the episode that we had previously with harrison

00:56:39.219 --> 00:56:43.250
talking about the role of ai in sales is probably

00:56:43.250 --> 00:56:46.050
a really good starting point for you to go back

00:56:46.050 --> 00:56:47.630
and listen to if you haven't heard that already

00:56:47.630 --> 00:56:50.429
um but yeah i think it's going to look very different

00:56:50.429 --> 00:56:55.230
for for different size organizations um as this

00:56:55.230 --> 00:56:58.289
this whole thing develops i think that's a such

00:56:58.289 --> 00:57:01.110
an important point and if you're listening to

00:57:01.110 --> 00:57:04.630
this why not you know spin up claude chat gpt

00:57:04.630 --> 00:57:08.269
give it the context about your business And ask

00:57:08.269 --> 00:57:10.130
it to go through the process that I went through.

00:57:10.170 --> 00:57:12.409
Right. And try and model out scenarios that are

00:57:12.409 --> 00:57:15.369
going to be much more specific to your context.

00:57:15.449 --> 00:57:17.889
That's probably the easiest way to get started.

00:57:17.929 --> 00:57:21.190
You can even pull the transcript of this episode,

00:57:21.530 --> 00:57:23.650
pull the bits that will be relevant and give

00:57:23.650 --> 00:57:25.929
that as part of the context as well. Give it,

00:57:25.929 --> 00:57:27.550
you know, give it some information about your

00:57:27.550 --> 00:57:29.050
business. Give it some information about the

00:57:29.050 --> 00:57:31.789
scenarios that we've been discussing and have

00:57:31.789 --> 00:57:35.050
some fun pulling it apart, making it more appropriate

00:57:35.050 --> 00:57:40.460
for you. You should never, you know, you shouldn't

00:57:40.460 --> 00:57:42.380
take anything we've said here as red. This is

00:57:42.380 --> 00:57:44.460
just us guessing of how these things could different

00:57:44.460 --> 00:57:48.059
play out in the future. And certainly ChatGPT

00:57:48.059 --> 00:57:50.059
can't predict the future either. But it might

00:57:50.059 --> 00:57:53.639
give you ideas where your intuitive sense of

00:57:53.639 --> 00:57:55.559
your customers, your market, your business was,

00:57:55.619 --> 00:57:57.579
oh, I've already seen a few things going this

00:57:57.579 --> 00:58:01.219
way. How can I capitalize on that? It can help

00:58:01.219 --> 00:58:04.929
you build the types of if this, then that. scenario

00:58:04.929 --> 00:58:07.349
plans that most small businesses just don't have

00:58:07.349 --> 00:58:11.230
the time to build right you know martin and i

00:58:11.230 --> 00:58:13.989
have both run small businesses we know that you're

00:58:13.989 --> 00:58:16.570
busy flying the plane and building the plane

00:58:16.570 --> 00:58:19.070
and and going through these types of processes

00:58:19.070 --> 00:58:22.309
with an ai chatbot is you can do it in an hour

00:58:22.309 --> 00:58:25.010
instead of having to you know take yourself out

00:58:25.010 --> 00:58:26.929
your business for a week to really think deeply

00:58:26.929 --> 00:58:29.409
on it as long as you keep that critical analysis

00:58:29.409 --> 00:58:32.070
part of your brain very much engaged and you're

00:58:32.070 --> 00:58:35.090
questioning everything that you think and that

00:58:35.090 --> 00:58:37.650
the ai suggests and really like having both of

00:58:37.650 --> 00:58:39.789
you criticize and analyze things to try and get

00:58:39.789 --> 00:58:43.449
as close to robust ideas and experiments you

00:58:43.449 --> 00:58:46.489
can run and stuff like that so all right bud

00:58:46.489 --> 00:58:48.590
good to see you as always i'll speak to you soon

00:58:48.590 --> 00:58:53.130
thank you for listening to artificially intelligent

00:58:53.130 --> 00:58:56.469
marketing to stay on top of the latest trends

00:58:56.469 --> 00:59:00.690
tips and tools in the world of marketing ai be

00:59:00.690 --> 00:59:03.630
sure to subscribe We look forward to seeing you

00:59:03.630 --> 00:59:04.630
again next week.