Latest / Shiny New Object - a Marketing Podcast / Stop using AI for everything - Finding the best use cases with UserTesting CMO Johann Wrede
Transcript
- 0:00Johann Wrede: Trying to build something that's big and
- 0:02beautiful and fits together perfectly is overrated. Speed
- 0:05and agility, getting things out faster, succeeding and failing
- 0:08faster, but in smaller increments, is better.
- 0:10Speaker: Are you a fitness-loving foodie? Then Defy
- 0:18Snacks is for you. Defy offers real high-quality chocolate you
- 0:22love, combined with functional ingredients, giving you as much
- 0:24as 26 grammes of protein per bag, so you can be both a little
- 0:28bit bad and a little bit good at the same time. As a female-owned
- 0:31brand, Defy dedicates a portion of its profits to other
- 0:35women-owned businesses. Currently sold in the US, you
- 0:37can find us on Defy with an I snacks.com and follow us on
- 0:41social at Defy Snacks.
- 0:49Tom Ollerton: Hello, and welcome to The Shiny New Object Podcast.
- 0:52My name is Tom Ollerton. I'm the founder of Automated Creative,
- 0:56the fastest way to scale, test, and learn from your performance
- 0:59ads, and this is a podcast about the future of data driven
- 1:03marketing. We've been doing this podcast for about seven years,
- 1:06and it's brilliant. It's so much fun to speak to marketing
- 1:09leaders about their vision for this niche within our niche. And
- 1:12this week is no different. I'm on a call with Johann Wrede, who
- 1:16is CMO at UserTesting. Johann, welcome to the podcast. And for
- 1:20anyone who doesn't know who you are, could you give us a bit of
- 1:23an overview of your career?
- 1:25Johann Wrede: Oh, well, thanks, Tom. I'm first of all, I'm
- 1:27thrilled to be here. Thanks for having me. I'm Johann Wrede. I'm
- 1:30the CMO at UserTesting. We provide a solution that combines
- 1:35a live human network with AI-powered synthesis to help our
- 1:38customers understand their customers' perspectives, it's a
- 1:42really interesting business, and I kind of wound up here through
- 1:46a circuitous path. I started out as a coder and wound up in
- 1:51sales, and I spent 15 years at SAP, launched products, and then
- 1:57wound up in go-to-market, doing a bunch of different things, and
- 2:01eventually in marketing, and then I've been a multi-time CMO.
- 2:05This is my third CMO job.
- 2:07Tom Ollerton: And how did that happen? How did you be in, like,
- 2:09code a little comfortable little corner coding stuff, and come
- 2:13into the nebulous, ephemeral world of marketing? How did that
- 2:18crossover happen?
- 2:19Johann Wrede: These two things have a lot more in common than
- 2:21most people think, but it was actually kind of a low, a slow
- 2:24journey. It was a slow burn, you know. I went from writing code
- 2:27to being in pre-sales, being in sales, being in go-to-market
- 2:31product launches, and had the opportunity to dip my toe in the
- 2:35marketing pond, and discovered that it's really fun to be
- 2:39creative.
- 2:39Tom Ollerton: Hold on, so that's even more weird. So, you're a
- 2:41coder, and then you went into sales. I've never heard of that.
- 2:44Johann Wrede: Yeah, I've done, I've done most go-to-market jobs
- 2:47in my career in some way, shape or form.
- 2:49Tom Ollerton: But how did you get you coding one day? And then
- 2:51someone said, "Right, you know you're selling. How did that
- 2:53work? How did that transition happen?
- 2:55Johann Wrede: Wow, this is going back a million years. I worked
- 2:57for a British company called Dialogue. I was hired to rewrite
- 3:02a search algorithm. I did that. They thought it was going to be
- 3:04a full-time thing. Turned out to be nine weeks later, I was like,
- 3:08okay, I'm done now. What asked me to implement a CRM system? So
- 3:12I implemented Clarify Salesforce automation. This is like really
- 3:16old, old times, back in the Clarify versus Siebel days. And
- 3:20and then I got a call from a recruiter saying, hey, they're
- 3:23looking for a sales engineer position, and I said, what's
- 3:27that, and they said being a consultant, but you're helping
- 3:30sell the product, well, that could be fun, so I took the job
- 3:33and learned how to do solution selling, and then eventually
- 3:37made the transition into into sales, and then went on from
- 3:41there.
- 3:42Tom Ollerton: Brilliant, that is so unusual, trust me. So, see,
- 3:46you've done all the, as you said, you've done all of the
- 3:48positions, give or take, within go to market. So, what's been
- 3:52your biggest mistake in that? What's been that, like, that
- 3:56mess up where you're like, "Oh, I wish I hadn't done that, but
- 3:59you're glad you did now.
- 4:01Johann Wrede: Well, I've got a long list of mess-ups in my
- 4:03career, but I think the one that the one that stands out to me
- 4:06was in 2014 We were, I was already in a marketing role, and
- 4:12this was early in my career as a marketer, and we were building a
- 4:17campaign, and at that time I was at SAP, and we kind of built
- 4:21one, maybe two big campaigns per year for the business unit that
- 4:26I was in, and we decided to tie it to the Olympics because it
- 4:30was an Olympic year, and so we spent months building this
- 4:33really beautiful campaign, months and over a million
- 4:36dollars of invested into the build on this campaign, and it
- 4:42was probably the worst performing campaign I've ever
- 4:45created in my career. We ran long on the build. The audience
- 4:51was oversaturated with Olympics everything, so they could have
- 4:54cared less. The regions who were meant to be using their budget
- 4:58to run these campaigns. So starved of content while we were
- 5:02building, building, building, that they used their budget to
- 5:05go and build stuff instead of running the stuff that we were
- 5:08meant to be building, so then by the time we were done, they were
- 5:11like, we've already built stuff, and I think we made every
- 5:16mistake that you could make in that, in that process, and I
- 5:20learned so much about speed and agility, and understanding that
- 5:24it's not just serving the market, it's serving the
- 5:26different parts of the marketing organisation, and it, it taught
- 5:29me, taught me to be cautious with themes, right? Like, you
- 5:32think it's a good idea to attach to the big mega trend until
- 5:35everybody's saturated with the mega trend, you know, trying to
- 5:41build something that's big and beautiful and fits together
- 5:44perfectly is overrated. Feed and agility, getting things out
- 5:47faster, succeeding and failing faster, but in smaller
- 5:50increments, is better. And it helped me really design an
- 5:54operating model that I use to this day, and the marketing orgs
- 5:58that I lead, that is really centred around smaller
- 6:02campaigns, always iterating, moving more quickly, getting
- 6:05things into the marketplace. So taught me a lot, taught me
- 6:08really it taught me how to be a better marketer.
- 6:17Tom Ollerton: So my next question, as you know, is What
- 6:19is your best advice to become a better data driven marketer, but
- 6:24I kind of feel that, like, I want to hear what is that
- 6:26solution. What is that, that that playbook that you have,
- 6:29that you've taken on all these different organisations?
- 6:32Johann Wrede: Well, it's, you know, it's so the playbook is
- 6:35really around again smaller campaigns, lots of smaller
- 6:38campaigns, so not just running one thing and running it all
- 6:41year, but rather looking at the different themes and topics that
- 6:45matter to your buying audience, and then executing quickly,
- 6:48targeting smaller groups of people with with very targeted
- 6:52messaging around the pain points that they have, and doing it
- 6:56quickly, and iterating, figuring out what assets are performing,
- 6:58and retiring the ones that aren't, and don't, don't take
- 7:01forever to build something, and then, and then try to run it
- 7:04forever, move quickly, and so that's, and then build an
- 7:09organisation that's designed around that, so you have a build
- 7:11team that's iterating on the things that they're building,
- 7:14and the run team is constantly optimising channels and
- 7:17providing feedback to the build team, and creating a virtuous
- 7:20cycle, and again built around speed and agility and
- 7:24coordination between those, so that the run team isn't off
- 7:28building their own things, because the build team isn't
- 7:31feeding them what they need, and then the build team isn't
- 7:33building things that aren't needed because they're not
- 7:35getting the input and the feedback on what's working from
- 7:38all working
- 7:38Tom Ollerton: Well, that logic is very close to how we think
- 7:41automated creative, but you'd be amazed how few people even think
- 7:44like that, let alone be able to pull it off unfortunately.
- 7:48Johann Wrede: that's what you get when you put a coder with an
- 7:50engineering mindset marketing, but you know, to answer your, to
- 7:54actually answer your question about the better data driven
- 7:57marketer, I think you know the key really is investing the
- 8:02infrastructure for data you trust, and make sure you have
- 8:04one set of numbers that you align with with the rest that go
- 8:08to market, so sales, customer success. If you're, if you're
- 8:11looking at your customer base, and churn numbers, and you know,
- 8:16renewal numbers, and growth numbers, and white space
- 8:19numbers, and pipeline numbers, make sure they're all aligned
- 8:22with the rest of GTM, because if, if you don't have that and
- 8:26you don't have that trust, you wind up with, like, well, okay,
- 8:32what's our goal? Well, we have different numbers for the goal,
- 8:34and we have different numbers for what you produced, and now
- 8:36you have confusion and chaos, and so it's really important, if
- 8:39you want to be data driven, invest in the infrastructure to
- 8:43get the data and get it ratified and agreed to, so everybody's
- 8:47working off the same sheet.
- 8:48Tom Ollerton: So we're going to talk now about your shiny new
- 8:50object, which is one we've not had before, but one can be more
- 8:55shiny, can be more new, which is Claude code and or Codex. So
- 9:00yeah, and there's going to be people who like me who got that
- 9:03up to their elbows in this stuff, and there's some people
- 9:06going, "What's Claude?" So, can you help us explain what Claude
- 9:10Code and Codex are, and why you've chosen it as a shiny new
- 9:15object?
- 9:15Johann Wrede: So, Claude Code and Codex are basically agentic
- 9:19coding tools for engineers. It's the simplest explanation. Claude
- 9:23code is from Anthropic, Codex is from Open AI, paired with Chat
- 9:26GPT. So these are tools that developers typically use to
- 9:31write code to develop applications, versus what people
- 9:35might experience when they use Claude or Chat GPT, which more
- 9:39of the chat bot interface, ask it a question, get an answer,
- 9:42Tom Ollerton: and why have you chosen it as your shiny new
- 9:44object? So, because the shiny object, the future of the age of
- 9:48a marketing for you is Claude Code and Codex. Why? Why have
- 9:51you chosen those two?
- 9:52Johann Wrede: And maybe it's because of my, my engineering
- 9:55coding background, but I find that these tools, I find that
- 9:58there's a lot of overlap between marketing and engineering, like
- 10:02I think marketing is effectively engineering with a different
- 10:05output. They're both, they both require skill, structured
- 10:10thinking. There's an art and a science to it. There's a
- 10:13tremendous amount of overlap between the disciplines, and so
- 10:16I really like the way these tools allow us as marketers to
- 10:22approach solving marketing problems, and right now
- 10:25everybody's rushing into agentic tools for marketing, and I think
- 10:29it's, it's really critical to think about a structured
- 10:33approach and building a new architecture in marketing tech
- 10:37stack, and and so I'm using it, and then the reason I'm so
- 10:42obsessed with this as my shiny new object is that I'm obsessed
- 10:46with finding all of the small challenges that marketers are
- 10:50using spreadsheets or other manual processes for, and then
- 10:56solving, solving some of the challenges in a way that doesn't
- 11:00just make everything, hey, use AI for it. So I'm building
- 11:03deterministic tools for repeatable processes, like UTM
- 11:08code generation, for example. Highly deterministic, you don't
- 11:11need to have an LLM to do it. There's a formula, you just have
- 11:14to surface it to a market or ask them to fill in some fields, and
- 11:18then the UTM can be generated, and you want to generate it the
- 11:21same way every time, and so building a generator is
- 11:25incredibly easy in an hour, and so then putting an MCP server on
- 11:31top of that, so that an agent can call it to generate that ATM
- 11:37code, and then having these wrapped LLM tools as a second
- 11:43layer, where you can create more structured generation, so that
- 11:48if you need something that is versus deterministic, like a UTM
- 11:53code, if everybody just goes to their Claude or their Chat GPT
- 11:59or their Gemini, context is different, like if you and I
- 12:03gave ours the same prompt, same, we would get different results.
- 12:07So building LLM tools allows us to inject more context and
- 12:13structure into the way that the prompts work, and allows us to
- 12:16get more consistent generative output, and then on top of that,
- 12:20we apply agentic AI that can call these other tools, and now
- 12:25what we get to is more automated and repeatable workflow, and so
- 12:31using these tools to build deterministic and wrapped LLM
- 12:40tools, as well as creating agents to them, all those things
- 12:45that has, that is, is what, why I'm so excited about these
- 12:49tools.
- 12:55Tom Ollerton: This episode of The Shiny New Object Podcast is
- 12:59brought to you in partnership with Mad Fest, whether it's live
- 13:01in London or streamed online to the global marketing community,
- 13:05you can always expect a distinctive and daring blend of
- 13:08fast-paced content, startup innovation pitches, and
- 13:11unconventional entertainment from Mad Fest events. You'll
- 13:14find me causing trouble on stage, recording live versions
- 13:17of this podcast, and sharing a beer with the nicest and most
- 13:20influential people in marketing, check it out at www dot madfest
- 13:25london.com
- 13:32So I'm gonna have to push you for a more plain English
- 13:35description of what you've just said, a few of most of it. I'll
- 13:39be honest, not completely, you know, I'm old enough to, but
- 13:42someone's, someone's coming new to this, right? Then what a UTM
- 13:46is, they don't know what deterministic means, and they're
- 13:49going, "Look, I'm the marketing manager at this chewing gum
- 13:53company, I'm looking at some chewing gum, right, like Johann
- 13:57is clearly miles ahead of everyone. But how, what, what do
- 14:01you mean? How do I start? Why is this sort of immediate benefit
- 14:03to me, the marketer who's got a boss breathing down the net,
- 14:06going, why aren't you AI-ing everything and getting rid of
- 14:09the agency?
- 14:10Johann Wrede: Yeah, so if we start from that premise, right,
- 14:13why aren't you AI-ing everything and getting rid of the agency?
- 14:16Typically, what we're talking about there is agentic AI, so
- 14:20effectively having bots that are running around doing things that
- 14:24other humans might have done in the past, those bots are only as
- 14:29good as the as the tools that they have access to and the data
- 14:33that they have access to, and so things like a UTM code is used
- 14:38in digital advertising to track where a click came from, and it
- 14:41has a certain format that's structured always the same, but
- 14:45you know that when you talk to Claude or you talk to Chat GPT,
- 14:48you always get slightly different outputs. So there are
- 14:51systems that you need that are highly repeatable, right? So
- 14:55rather than doing that in a spreadsheet like most people do,
- 14:58if it's in a spreadsheet, it's. Hard for an agent and automation
- 15:01to call that spreadsheet and say I need one of these codes, so
- 15:05you have to write a little app that generates that for you, and
- 15:09then you can put an API, basically it's called an MCP
- 15:13server, allows an agent to call that app as a tool and then go
- 15:19get that UTM code as part of a workflow process. Wrapped LLM
- 15:24tools that I was talking about are, and we see those - there
- 15:28are lots of staff products out there where it has AI, so that's
- 15:34a wrapped LLM where we take functionality and we put it on
- 15:37top of a tool like Chat GPT or Gemini or one of these ll large
- 15:43language model AI generative AI tools, and what that allows us
- 15:48to do is it creates repeatability, because the user,
- 15:55the marketer, will you'll write things differently than I'll
- 15:57write things differently, and even if we wrote the things the
- 15:59same way, copy and paste the prompt in will get outputs
- 16:03because our when we are using our own tool like GPT or Claude,
- 16:08it learns who we are and it has different ideas about what it
- 16:12should give us, and so being able to build these applications
- 16:17that create more context consistency will allow for more
- 16:24consistent output, and again allow an agent to call those
- 16:30tools to get a consistent response back, and you need
- 16:34those building blocks first. If you just go and slap an agent on
- 16:37top and say, 'Hey, I want you to automate all these things,
- 16:40what's going to happen is, you're going to have chaos.
- 16:43Everything's going to look different. You're never going to
- 16:44get the same results twice. It, if you have different people
- 16:49that are using these agents, they're going to get different
- 16:51results. And then you wind up with disagreement in your
- 16:54organisation, because people are like, well, I used the tool, the
- 16:57agent ran it, and it did this, what's the right one, right? And
- 17:03so, so creating using these coding tools to, to do less just
- 17:10trusting AI and building these structured elements, these
- 17:13structured building blocks, and then letting that agent, that
- 17:17agentic AI, call those structured building blocks gets
- 17:21you a much more consistent, repeatable.
- 17:23Tom Ollerton: Yeah, so we've had the extreme user definition and
- 17:26the newbie definition. Beautiful. Thank you. So, so the
- 17:30issue that I'm having is repeatability, right? And when
- 17:33you have like a fixed data set, like this is our UCM structure,
- 17:38for example, and you have the agent governing how governing
- 17:43it, it's a fixed thing. This is how this, our UTM structure,
- 17:47like I need these input output, right? Whereas what I'm the
- 17:52issue I'm finding, and I spotted this when I've been on Reddit,
- 17:55so I love all the AI Claude chats on Reddit. It's literally
- 17:59terrifying. Nothing, nothing will keep you awake at night
- 18:03longer than thinking about your career once you've read some of
- 18:06those posts, but it's great. What this person was talking
- 18:09about is how more work shy Claude has become, like how less
- 18:14willing it is to check its own homework, and so if you're like
- 18:17even with their fixed data set, so like you know, I have a
- 18:20transcript of this call, and I was like, well, what were your
- 18:23own points? Oh, yeah, you know, we talked about LLMs, or
- 18:26whatever. And then you go, yeah, but did he? And then Claude will
- 18:28go, actually, no, I didn't check. And you're like, mate
- 18:31what am I paying you for? What am I paying you for. So, how are
- 18:34you getting around that inconsistency?
- 18:36Johann Wrede: I think you know, for starters, it's a lot of it
- 18:40is prompting, so ask it to check its work, build, build loops,
- 18:44Tom Ollerton: of course. So I've done that, you see, but, but
- 18:47Johann Wrede: use different tools, so ask Codex, like I will
- 18:50ask one AI to check another AI's work. Also, I mean, let's be
- 18:55honest, right now it's a little bit of the wild west out there,
- 18:58and, and you know, you mentioned Reddit, if you go out and you
- 19:01read about these tools, they are evolving, they are leapfrogging.
- 19:05It was all about ChatGPT, and then at the beginning of this
- 19:08year, Claude ate the universe, and now all of a sudden
- 19:11Anthropic - there's some news about Anthropic - is trying to
- 19:15manage their costs, so they're tinkering with the formula
- 19:18behind the scenes, right? And so the amount of thinking time,
- 19:22because if you, you know, if you remember that, that these
- 19:26companies are their businesses, and they, they have costs on the
- 19:30back end that they're trying to manage, the every millisecond of
- 19:34extra compute winds up piling up into a lot of additional cost,
- 19:40and so you know, I think that's where you see all the recent
- 19:43Reddit threads and Instagram reels about, has Claude lost its
- 19:47mind? I think a lot of that is attributed to some tinkering
- 19:51they're doing on the back end, but that's why you know I
- 19:54believe it's important to subscribe to all of these tools
- 19:57right now and make your workflows independant of the
- 20:01tool, so that you can pull, pull it out, replace it with a
- 20:04different tool, see if you get a better result, but that's also
- 20:08why I'm a big believer in building applications that don't
- 20:12require AI in order to do things that are highly repeatable and
- 20:16structured. I think one of the big mistakes that I see right
- 20:20now, marketers are using Chat GPT or Claude, or whatever, for
- 20:24everything. They just go, they go there, and they're like, "Oh,
- 20:26I want this and analyse this data, do all these things. I'm
- 20:29like, "Well, you don't need to analyse the data, you, what
- 20:31you're actually asking for is a formula. Like, help me build a
- 20:34formula, and then put the formula in your Excel
- 20:36spreadsheet and pull it down the column and get your answer.
- 20:39Don't ask it to calculate the answer and populate the
- 20:42spreadsheet, right, and I think that's where we have to get
- 20:46better at identifying the use cases and what types of AI uses
- 20:51appropriate and when, and that brings me back to why Claude
- 20:55code versus Claude, right? I want to build things that are
- 20:59reusable and appropriate, as opposed to just asking the LLM
- 21:03to do everything, saves me on tokens, and it also improves the
- 21:08quality of the output, regardless of these tools. But
- 21:12you can, you can always ask Codex to check Claude's Claude
- 21:16codes output, debug, do code reviews. You can do the same
- 21:19thing, Chat GPT. Hey, I got this from Claude. Can you give it a
- 21:24review and give me feedback? Get a second opinion. Ask both tools
- 21:28to review the same transcript and look at the two highlights
- 21:31that they, that they put together. And then, you know,
- 21:34human in the loop, right? Review it.
- 21:37Tom Ollerton: So, you and I could go in so many different
- 21:38directions here, and we've run out of time. So, got one
- 21:40question left for you, so I can't get away from the argument
- 21:44that AI is being completely subsidised by the big players at
- 21:48the minute, right? All of that compute, so no one's making any
- 21:51money, everyone's losing money, everyone's losing money faster
- 21:54than anyone's ever lost money.
- 21:55Johann Wrede: It reminds me the dot-com days.
- 21:59Tom Ollerton: I mean, I came into this industry just after
- 22:01that, but it's like, I mean, random numbers. Someone said
- 22:04that on Anthropic's most expensive product, they still
- 22:07lose 20% or something, like in ChatGPT, are burning -
- 22:11everyone's burning money. Someone said ChatGPT is a
- 22:14website, so it's a money pit with a website built on top,
- 22:17right? So, you, you, and I need your cloud code, and I'm using
- 22:21it all the way across my work and the business, and it's fun,
- 22:25if not frustrating, but it's pretty close to free. It's not
- 22:28far off free, right? But no one's making any money,
- 22:31someone's gonna have to make some money, so the price is
- 22:33gonna go up, right? Or it's gonna get worse, right? Less
- 22:38compute power, go through your usage in a heartbeat, so I'm no
- 22:43economist, but that day is coming. What happens then?
- 22:46Johann Wrede: Well, I mean, I think this is it. So, first of
- 22:49all, if we look back, so many industries, so many businesses
- 22:54have started that way, right? There's the free period, where
- 22:57the VC guys provide a lot of money, you have everything is
- 23:00free, and then over time we start to pay for services, and
- 23:04then we see, you know, it's all about the land grab, grab the
- 23:08market share, and then once you have the market share, you can
- 23:11raise your prices, you can take cost out, right, and we're
- 23:14starting to see that, we're starting to see the model
- 23:16providers making choices about what you can use, you know,
- 23:20their LLMs with, and how you're going to pay for that. What,
- 23:23what do you pay API fees for? What's included in your
- 23:26subscription? They're going to work to make it profitable over
- 23:30time. Right now, they're in that land grab phase, where they're
- 23:32investing to own market share. So, I think we need to be
- 23:36prepared for the fact that right now people are trying to take
- 23:40cost out of their business by using AI. That cost is going to
- 23:44come back. I also think there is a future for especially larger
- 23:49enterprises hosting their own models and building those models
- 23:52up, so using base models like Gemma or Quen or Deep Seek,
- 23:56taking those base models and hosting and managing their own
- 24:01infrastructure as part of that, so they can make those tuning
- 24:04decisions. I don't think that's going to be universal, because
- 24:07it requires a lot of resources in a large IT organisation, but
- 24:11I think businesses will figure out how to adapt to a change in
- 24:15cost structure around AI. But for right now, I think it's
- 24:19about us learning these tools, figuring out how to use them,
- 24:22and keeping an eye on the fact that you know there may be some
- 24:27cost takeout now as you automate things and think about reducing
- 24:31headcount, but those costs are going to come back as the as the
- 24:35profitability requirements change on on the frontier model
- 24:39providers.
- 24:46Tom Ollerton: So you've argued that point far better than I
- 24:49have. No one in marketing, no one wants to listen. It is so
- 24:53fun to do job now with these tools, right? You have an idea,
- 24:57you make the thing, right? It's like that's never. Assisted in
- 25:00me 20 years in this industry, so to be able to go, I want a thing
- 25:03that does this thing. Oh, here's the thing, amazing, literally
- 25:06amazing.
- 25:07Johann Wrede: The big question is, should you build it, though?
- 25:09Like, that's that's the question we're trying to help people
- 25:12answer.
- 25:12Tom Ollerton: Well, I said, well, not, not, not my brain is
- 25:16just, well, what I could do is, I could build an app to do,
- 25:19like, what if I did this thing for my daughter, and it's, but
- 25:23really, so no. Do I actually need any of these things, which
- 25:27is,
- 25:27Johann Wrede: and that's the question we're trying to help
- 25:29companies answer, is like, do your customers actually want
- 25:32these things? You know, that's that's the big question on
- 25:35everybody's mind right now, because with AI, you can build
- 25:38anything. So, the question is, what should you build? What,
- 25:40what is valuable to your customer?
- 25:42Tom Ollerton: Right, we need to do another seven of these
- 25:44podcasts. Johann, thank you so much. If someone wants to get in
- 25:48touch with you, what's the best way to do that? And what makes a
- 25:50message that you'll respond to?
- 25:52Johann Wrede: Find me on LinkedIn and message me with
- 25:54anything but a sales pitch, anything but a sales pitch,
- 25:56please. I get so many of those.
- 25:58Tom Ollerton: Yes, indeed. Yeah, because someone's built a
- 26:02someone's built a bot to automate the production spaces.
- 26:05Johann Wrede: Yeah, exactly.
- 26:08Tom Ollerton: Johann, thanks so much for your time,
- 26:09Johann Wrede: Tom. This was brilliant. Thanks for having me.
- 26:18Tom Ollerton: Hi, just before you go, I'd really appreciate it
- 26:21if you could take the time to write a review of the Shiny New
- 26:25Object podcast on Apple Podcasts or iTunes, whatever it's called
- 26:29these days, or whichever podcast provider you use. We're at Indie
- 26:33Podcast, so it would go a long way for us if you could just
- 26:36share the word and give us a bit of a support on those channels.
- 26:39That'll just be fantastic. If you haven't got time, that's
- 26:41also cool, and yeah, if you could tell your colleagues about
- 26:44the podcast, and also, if possible, don't forget to
- 26:48subscribe, and I'd love to hear your feedback, if you'd like to
- 26:51speak on the podcast, I'll be a guest, or you think I'm asking
- 26:54the wrong questions, anything, I'd be super interested to hear
- 26:57what you think, so please email me at Tom at Automated
- 27:02creative.net that's T O M at.. I'm not going to bother spelling
- 27:07it anyway. You'll work it out. Thanks so much.