Are you eager to boost your website's performance on search engines like Google but unsure where to start or what truly makes a difference in SEO?
Then "SEO Is Not That Hard" hosted by Edd Dawson, a seasoned expert with over 20 years of experience in building and successfully ranking websites, is for you.
Edd shares actionable tips, proven strategies, and valuable insights to help you improve your Google rankings and create better websites for your users.
Whether you're a beginner or a seasoned SEO professional, this podcast offers something for everyone. Join us as we simplify SEO and give you the knowledge and skills to achieve your online goals with confidence.
Not strictly an SEO episode this time, but a real-world example of what AI tools can now do. A friend from the village pub sent Edd an eleven-minute training video full of personal data, plus four plain-English rules for what needed blurring, and a note saying he suspected the AI would struggle.
Edd walks through how Claude Code tackled it: reading the text on nearly 20,000 frames, writing a program to apply the rules, checking its own work with a debug render, and verifying the result.
The takeaway: jobs that once looked too fiddly, too specialist or too time-consuming are now within reach.
SEO Is Not That Hard is hosted by Edd Dawson and brought to you by KeywordsPeopleUse.com
To get a personal no-obligation demo of how KeywordsPeopleUse could help you boost your SEO and get a 7 day FREE trial of our Standard Plan book a demo with me now
Hello, welcome back to SEO Is Not That Hard. It's me here, Ed Dawson, as always. And today I'm going to talk about, well, this isn't really an SEO thing, but it's really a power of AI thing, so it's an interesting little anecdote to share, I think.
So regular listeners will know that I live in quite a rural part of the UK in a little village, and we've got a village pub, which is kind of like the hub of the village, and it's this really old coaching pack horse inn. real fires in winter, that kind of place. You know, the, the, the archetypal British village pub. Anyway, I go in most Wednesdays with my dog, Brodie, and, uh, another one of my neighbors, a local villager called Peter, comes in with his dog, Poppy. And Poppy and Brodie, they really get on. They love each other. So we have a meet up on a Wednesday for a chat and a beer, and the dogs have some pork scratchings
Speaker 2
Now, Peter works in tech too. He's got a business which is a hardware business, um, but obviously every hardware business has got software attached to it, and he has customers who use his hardware, and they have online web-based systems which allow them to, you know, utilize, manage, and work with this hardware And I've been telling him lots about AI recently and how we're using AI in our business, and how it's a complete game changer in many respects, especially along the lines of doing those things that you wouldn't have done before or solving problems that would have been very time-consuming, even if they're quite simple in themselves So there's been a couple of occasions where I get an email or a text from Peter in the week and says, "I've got a problem. Maybe your AI can solve it for me." And there was one that I did a while ago that was just cleaning up some data in some spreadsheets, which is really quite a simple thing for AI to do. And he was amazed that literally, I think he sent them me, and I sent them back within two or three minutes completely cleaned up, and it was a job that if he'd done manually would've taken him hours. And even if he'd written some code to do it himself, again, that would've taken time to run through and test, whereas the AI could just do it really, really quickly
I got a WhatsApp from Peter just yesterday, and it said, Okay, let's see what AI can do. What I have is a rather basic little training video to show customers the difference between our existing machine load of software and a new version coming out soon. But parts of the video contains people's personal data, so I need certain sections blurring out. Surprisingly, there are no easy tools out there despite a lot of searching. So I thought if I write a spec that says which parts of the picture to blur and when, maybe Claude could do it." So I said to him, You know, I've actually never used AI for this kind of video manipulation before. But let's give it a try and we can always ask, and we can see what we get out of it." So he sent the, um, video over. It's about a hundred megs, megabyte video, so that's about eleven, ten, ten or 11 minutes long it was. And, um, then he sent me an email with it- That just had some basic rules in it. Things like, um, what he wanted blurring, what columns he wanted blurring, what kind of data he wanted blurring, like personal details, um, the names of customers who were using the software, 'cause he was using one of their customers' interfaces to, like, show, um, the data and how it h- and how it's changed, how the new system has changed, and a few others. And he finished it by saying, "I have a feeling it might struggle, but let's input the video and these rules and see what happens." So that all came through last night. Then this morning I thought, "Right, let's give it a go." Now the rules it sent were, one, always blur two specific words wherever they appear. Two, from a specific timestamp about six minutes in, blur everything underneath five named column headers and keep blurring those columns even when the page scrolls and the header disappears off the top of the screen. Three, don't blur a column whose header is the plural of one of those words. Four, don't blur the pinned column if the word sent is directly under the header. So I gave the video and those rules to Cloud Code and asked, "Can you do this or do you know a tool that can?" Now the answer was interesting.
It said yes, but not by asking the AI to edit a video. He said-- it said the job is really write a small program that follows those rules, so that's what it did. And here's how it worked. First of all, it split the video into individual frames, nearly twenty thousand of them, and it ran text recognition on every one, uh, using the recognizer that's built into my Mac. That gave you every word on the screen and exactly where it sits. Secondly, it then wrote a script that applies the rules frame by frame. So it found the two names is the easy part. The columns it found it was harder. It has to find the header row, work out where each column starts and ends, and then blur everything below. Then when the user scrolls and the headers vanish, it has to remember where to keep those column-- where those columns were and keep blurring them. And it has to notice when the screen changes to a different page, so it doesn't cl-keep blurring stale positions. Thirdly, it then re-rendered the video with the blurs applied and the original audio kept.
The part I find most useful to watch was the checking. So rather than blur straight away, it rendered a debug version with colored outlines showing what it intended to blur, and then it checked those against its rules. And from that it found several problem, uh, several problems such as columns that were coming out too wide, a column that briefly dropped out whenever the mouse cursor sat on top of its header because the header text couldn't be read, uh, pop-up dialogues covering half the table, stray blur blockses that, that were stuck at the bottom edge of the screen. It fixed. It found all of those, fixed each one, and re-rendered until the outlines were right, and only then did it produce the real thing And after it did the first video and I watched, I watched it, I just said to it, you know, "Are you confident whether that all the things that are meant to be hidden are actually hidden?" So instead of just saying yes, it actually ran the text recognition on the finished blurred video, sampling the frames, uh, looking for all the words that it was meant to and all the things it was meant to block out. It checked, checked the whole video through, no problems found, so it reported that back to me.
Speaker 5
So that, yeah, brought up the output. Yeah, the full 11-minute video, audio intact, with all the names blurred out everywhere, all the columns, sensitive columns blurred, um, and everything sorted.
Start to finish, I think it probably took Claude maybe about 40 minutes. Um, which for saying how much work, if you consider every single frame rendered, then marked what it was going to do, checked, double-checked, then blurred, then checked again, that's quite incredible really how quickly that was done. And if you'd sat down and tried to do that manually, even with specialist video editing tools, it would have taken, I'm sure, qui- a lot longer for a person to actually get through and do that. Um, yeah, and for saying how my friend Peter thought this might be something that was gonna be quite challenging for it, it actually just nailed it completely. So this, as I said, it isn't actually an SEO-related topic or podcast today, but this is the kind of small wins that using AI tools like Claude Code can really take your work to the next level. Because it just means you can potentially now create content that you would have found too consuming or too complex to do before, or required too specialist skills that you don't have or couldn't afford It means that you can accelerate your work and accelerate the ability for you to create an output work well beyond a point that you could have done previously. And it means you can think potentially more ambitiously about what k-kind of content and the kind of projects that you could complete and produce than you could previously. So I just thought I'd give it a share. Peter's not the first person in the pub that I've managed to get onto AI and explain how things that they thought would've been impossibly complicated or impossibly expensive or impossibly time-consuming, um, previously are now completely possible. And, um, yeah, the bonus of all this is I sent the video to Peter. He's been through it. Yeah, Peter's completely happy with it, and he said all the beers tonight are on him. So yeah, we did an interesting little project for him, and yeah, I'm gonna get a few beers for free out of it, so I'm not gonna complain at that.
So I hope you found this episode interesting. If you've got any questions, comments, always do get in, in touch, and yeah, look forward to seeing you in the next one. Take care till next time.