Software Development for Media Brands with Jason Agnew from Big Bite
In this week’s episode, I had the pleasure of speaking with Jason Agnew, the founder of Big Bite, about the impact of technology on media. Jason shared his experiences working with major media brands and the challenges of larger and more complex projects. We discussed how having a larger team can make difficult tasks more manageable and delved into the differences between smaller and larger sites.
About Jason Agnew
Jason is the CTO and Co-founder Big Bite. Big Bite helps enterprise organisations to publish rich digital content faster and easier than ever before with WordPress.
Stewart Richie and Jason Agnew are the founders of two web software development agencies: Powered by Coffee and Big Bite respectively. They have known each other for a long time and have been discussing the impact of technology on media. Jason has been with Big Bite for 12 years and has worked with major media brands like DMG Media, Metro, IDG, Macworld, News Corporation UK and the Wall Street Journal. He notes that the journey has been like a rollercoaster with some great highs and some lows.
In this conversation, two people discuss how the success of their business has led to bigger and more challenging projects. The speaker recalls a project for Post Box for which they received 30 thousand dollars, which seemed like a lot at the time. However, they have since learned that for every big and scary project there is another, even bigger one further down the line. The speaker also notes that having a large team to help them with difficult problems makes it easier, as they are no longer dealing with the issues alone. Ultimately, the conversation highlights how success can lead to bigger projects, and how having a larger team can help to make difficult tasks more manageable.
Jason and Stewart discussed the differences between working with smaller sites and larger ones. Jason explained that the primary focus for larger sites is the backend, particularly writing content in the WordPress editor, Gutenberg. He noted that the team he works with transitioned from front end developers to React developers. He also mentioned that the team works with traditional WordPress sites as well as those with marketing brochure-style sites. The challenge is that the team needs to be able to work in both an editorial environment as well as with marketing brochure-style sites. In addition, they must optimize workflow to reduce the amount of time it takes to write an article. Jason concluded by noting that while they still hire PHP developers, most of the team is now doing day-to-day work in React.
The conversation between Stewart and Jason touches upon the opinionated people in the WordPress market and the divide between those who are eager to embrace React to build blocks and those who are resistant to the change. Jason acknowledges that the transition to Gutenberg may be difficult for smaller businesses, as they would need to take time away from making money to learn the new technology. He encourages people to give it a try, as it makes it easier in the long run to build a site. He mentions ACF as an example of a technology that was heavily used in the past, but still has some users.
00:00:00 Conversation between Stewart Richie and Jason Agnew on Big Bite’s Journey in the Media Industry
00:02:11 Conversation on Overcoming Challenges in Client Projects
00:03:35 Exploring the Differences Between Smaller and Larger Sites: A Conversation with Jason
00:07:34 “Exploring the Impact of React on WordPress Development”
00:09:09 “Exploring Scalability of Traditional WordPress Sites”
00:10:34 Exploring Cash Segments for Large-Scale Website Development
00:13:45 Exploring Decoupled Website Development: A Discussion with Jason
00:15:19 Headless CMS: When Does It Make Sense?
00:18:41 “Exploring the Challenges of Headless CMS in Media Businesses”
00:20:02 Exploring the Impact of AI on the Publishing Industry
00:21:41 Exploring the Use of AI in Content Generation and Automation
00:23:01 “Exploring the Potential of AI for Content Writing”
00:24:31 Exploring the Benefits of AI-Powered Mass Syndication and Co-Generation for Journalism
00:30:49 Discussion on the Impact of Code Generation on Software Development
00:32:08 Conversation on Software Requirements, Project Management, and Mentoring
00:33:34 Mentoring Program at Big: Training Junior Developers and Creating an Engineering Mentor Role
00:38:47 Conversation on Remote Working and Developing a New Product
00:42:12 Exploring the Benefits of WordPress Gutenberg Blocks and Block Querying for Multisite Content Management
00:43:53 Interview with Jason from Big Bite: Discussing Content Query and Engineering Mentorship
00:45:00 Conversation between Jason Agnew and Stewart Richie on Fair Processes for Contacting BigBite.net
[00:00:04] Stewart: Hi there. welcome to Scale, podcast for Modern media. I am Stewart Richie, the finder and lead developer at Par Coffee, web software development agency focusing on, media and publishers.
[00:00:16] scale is a podcast looking at high technology impacts media, how media is impacted by technology and kind of. In between
[00:00:25] Stewart: today we have Gsen Agni, co-founder and CTO of Big Bite. Big Bite are another, software and web development agency be a side of the uk. and we go way, way back. GS n tell everyone about yourself.
[00:00:43] do a better job of introducing you than I can on my own.
[00:00:46] Yeah. Well thanks for having me on. And as you say, we go way back Struck claims that, he taught me how to program, which isn’t true. I didn’t
[00:00:54] teach you how to program. I.
[00:00:56] Jason: Okay. Okay. Well, these days we, we maybe don’t, talk so much about the day to day programming, but more about business. but, as you’ve said, you know, our WordPress agency at heart, essentially we help people publish content better WordPress.
[00:01:10] So we focus a lot in the publishing space. that’s with companies like DMG Media, with the Metro idg, with macworld, and obviously a lot with News, corporate News UK with the Wall Street Journal. Post some of the time. So, quite a bit of experience around that and a lot of lessons learned so far.
[00:01:26] Stewart: it’s quite the clientless and I’m always very jealous every time we talk about it. I’d love, I’d love to give our listeners, some background and some context around how, how Big Bite got there. so the company Big Bite’s probably what, 10, 11 years old in that region?
[00:01:41] Jason: well, it was pretty much hitting 12 now. Yeah.
[00:01:43] Stewart: Yeah. so Samin, what’s, what’s that journey been like to go from, three or four of you kind of starting out together to working with some of the biggest media brands in the.
[00:01:54] Jason: It’s, it’s been a wild ride, I guess, you know, we, we, we’ve talked before about this, but it’s like a rollercoaster, isn’t it? It’s, the, the good times are really good. and the bad times are really bad, so, you know, you’re always up and down in, in terms of how would it never expect us to get? To these sort of style of clients, you know, these big names that you would know.
[00:02:11] I remember when Post Box, which is a little, meal client you get on, the Mac. Some people probably don’t even know about it. And I remember running around the office like enjoyed and it was like 30 grand to Buildmo website. And I was like, this is so much money and this is such a big project and everyone will know their name.
[00:02:26] Not everyone knows their name. And it wasn’t that big of a project, but it was exciting for us. And I think what we’ve learned is. For every project that’s, it’s big and scary. There’s another one that’s bigger than that, later down the line. And you know, you have to do one to figure out the best way to do it for the next one.
[00:02:41] And you, you always get there. It’s just that sometimes it can be quite stressful. I think nowadays we have a, a big team and I think maybe it is a bit easier because you’re no longer on your own, which is something that you struggle with in the beginning, which is like, When you don’t know the answer to something, it’s just yourself and maybe one or two other people to ask.
[00:02:59] But now that we have a a large team, it’s a more of a consensus when we have difficult problems to solve.
[00:03:03] Stewart: So what. more, I suppose of a technical question cuz that’s at our hearts, that’s who we are. What, what do you see kind of as the differences, like having done kind of, you know, a lot of, I’m not gonna say smaller, but like the kind of stuff you were doing when you were kicking off a lot of market insights, you know, stuff for, postbox all the way up through to these, you know, Behemoths of sites that are doing hundreds of thousands of hits a month, hundred thousands of hits a day is very different way of working and developing in a whole different set of problems.
[00:03:35] what, what has your experience been of kind of that change between these kinda smaller sites into these huge things?
[00:03:40] Jason: I mean obviously the first thing is the front end nearly doesn’t matter anymore. I know that’s not true per se, but, and the front end of a new site is quite basic, really. in some ways you could argue a WordPress FEMA of the box would nearly satisfy the conditions of it. Obviously, it needs branded and it needs optimized, and a lot of these sites end up being decoupled anyway.
[00:03:57] So our primary focus is the backend, how you write content. So, you know, the team that started out were very much front end developers who have now transitioned into mainly React developers is the best way of thinking about it, because the amount of stuff that we’re doing in Gutenberg, which is the editor and WordPress now.
[00:04:14] So we talk about this a lot internally because we have still clients that are traditional WordPress sites and, our concern is that the ones that work with all the big publishers would struggle to move on to the solar projects and vice versa, because we’re diverging in some ways between having people who are excellent,working in an editorial environment.
[00:04:33] Now our people are working in the sort of, marketing brochures to style sites. So, but it’s, it’s a good, it’s a good transition because in some ways, Being able to optimize is really enjoyable to see someone’s workflow and say, I wonder how we can make that better. You know, if you watch someone do something and text ’em, a lot of times they, they record it and clicks.
[00:04:52] They’ll say, it texts 20 clicks to do that. And you’re able to sit there and think, well, I wonder if we can have that. And so some of it is a lot to do with like UX and working with that UX team to figure out what way we could change something and. That’s an enjoyable experience to see and get to the sides where someone says, well actually, if shaved off half the time it takes, write an article.
[00:05:11] whereas we used to build those brochure sites, there wasn’t really a goal per se. Maybe page speed or something would be a goal, but it was hard to quantify if you’d really made a difference.
[00:05:20] Stewart: Yeah, of course. I think that kind of distinction around like you’re not building a site, I suppose, but like a set of small products that enable. Their editorial teams and their newsroom teams, or however they kind of wanna call themselves to work faster and move faster and reducing the cost in, you know, human time, which has a direct impact on like, billable time to, you know, produce content, for, for these brands to go out, which obviously then either translates into ad spend, or, you know, connections through the rest of their. The rest of their funnel. Our last guess was, Leslie from Newsletter Glue and her experience is very much the cm, where they’re building a, a plugin to massively reduce the time it takes to produce, email newsletters by checking WordPress and building newsletters in that. So I just think it’s a really interesting, interesting space.
[00:06:11] but I suppose then you’re kind of getting that same thing. With your team almost as well. If they’re more and more specialists, you’re gonna end up presumably with folk who are very WordPress and PHP specific and React developers who are kind of doing that block work. Is that kind of where it’s going, I assume, or?
[00:06:30] Jason: I think there’s probably less and less people who, who would be considered a PHP developer in the business. That’s not to say that people can’t do phb. Everyone in the in the company can do PHP as a developer. So, I don’t wanna say that they can’t, but I think that people have had all transition to News React, and therefore everyone’s probably doing more and more of their day-to-day in React.
[00:06:53] It’s just that as WordPress is in php, at some point, if you’re writing a big plugin, You’re gonna walk into that situation where you’re gonna have to use PHP and think, quite a good example recently where we were working with a client and they have an internal team too. And it was pretty obvious that because they’d hired solely React developers who’d never used WordPress to build stuff, they were solving problems on client side that should have been on server side because they just didn’t know PHP and they wanted to avoid really getting into the weeds of WordPress.
[00:07:20] So we had to change a lot of that. And yeah, so I. I’d still hire a PHP developer right away because, you know, it’s the knowledge of over the years that they’ve learned that’s valuable. It’s not necessarily the language that they do.
[00:07:34] Stewart: Yeah, of course. as an aside, I think I’m, and you maybe seen a similar thing if you kinda look across the rest of the WordPress market where React is a real splitter between folk who are building sites where those who just refuse to touch it and are very angry that they’re kind of being pushed into a react world to learn about how to build blocks.
[00:07:56] and those who are fully, fully embracing it as like, well, this is just how it’s going to be. have you, have you seen, is that kind of my paranoia about the space?
[00:08:06] Jason: No, I think there’s, there’s a lot of opinionated people in every industry and, and there’s a lot in the word press space as well. I supervise a lot actually cuz you know, although I know React, I don’t do that much of it. And, I guess in some ways if I was still a small business and I was. You know, coding day to day.
[00:08:24] Then the, the change to Gutenberg maybe would’ve had more of an effect on the overall business because it would’ve been a reskill of me, but, so maybe, maybe the smaller you are, the bigger of an effect it’s actually had, because how can you take time out of making money to keep the business going to learn?
[00:08:41] So I, I sympathize with ’em and, but I would also encourage people to think again, because actually once it starts, you’ll realize that it makes a lot of sense, the approach with, with blocks, it actually makes it. Easier in the long run to to build a site out because you’re just building a kit and then the site builds itself essentially.
[00:08:58] So, it’s everything that back in the ACF days, I say back in the a cf, I know people still don’t use it, but back when we heavily used it, and it seemed like there was no other way of doing it. That’s what you were trying to achieve, but you just couldn’t do it properly.
[00:09:09] Stewart: Yeah, I agree. I think so. coming back from WordPress, cuz that’s just an implementation detail of all of this, I’m really interested to hear that you kind of almost see the, the front end as, not irrelevant, but kind of the easy part when we’ve kind of worked at sites that were at. Like some of the stuff we’ve really had problems with has kind of like been some of the PHP stuff of like making sure it’s like scalable and it’s gonna work under like, all the, all the load that can kind of come from these sites.
[00:09:39] you know, particularly if you’ve got something where there’s a membership system and you kind of can’t cash per page. and so that’s a really interesting split. I think in some of the stuff that people were doing that I kind of see. Yeah, I dunno what my question was there. I just, it’s an interesting observation.
[00:09:54] Jason: I, I just, I, I know what you’re saying. Right. but will the site scale? it is always an important question. I don’t wanna take away that it is, it is difficult sometimes to get a site scale, but I, I, I really believe that if you, if you keep it very traditional, WordPress, And you’ve, you’ve got the right monitoring and, and the right host with you on this.
[00:10:13] So, I, I think you can make, traditional WordPress site very scalable and, but obviously cash plays a big part in that. And so if there’s situations where you can’t cash stuff, then you sometimes need to think out of the box. So you might be saying, This tiny component maybe needs to actually use a third party service, to hit.
[00:10:34] So for example, maybe you have something that’s very interactive on the front end and you’re like, well, it would be good to cash most stuff, but these requests that this is gonna make the save of the data, maybe we need to fire off to some like fire base or a little node app or similar application where it’s okay if that goes down because the overall site’s still functioning.
[00:10:49] and I think cast segments is probably the key to getting. Different scenarios to be cashable, in that concept. So definitely some that I didn’t do, do so well in the very beginning, but now we’d use it all the time.
[00:11:01] Stewart: cash segments, was that what you said?
[00:11:03] Jason: Yeah.
[00:11:04] Stewart: Okay. I, it’s not free as I’ve heard. what’s a cash segment?
[00:11:08] Jason: Well, if you think about. when people visit a site, there’s, there’s different types of people you could consider a logged in person, and then a logged out person is two different types of users. So you can generate cash based on that. So you can have a logged in session for cash users and a logged and logged out user to have a separate cash, meaning that, the site’s technically still cashed after you’ve, you’ve got them in the door.
[00:11:29] so, and, and it works even better if you’ve got situations. You have different users from different areas. So like the site might have all directions for those specific users. They’re from a different country. And now you can, you can essentially have as many cash segments as you want. And there’s also the ability to cash individual elements in a page.
[00:11:45] So, you know, cashing part of the page, not the entire page and letting part of it render as well. So there, there’s just a lot of flexibility in it. and I think most managed hosts in WordPress space now are pretty well set up for it. And you.
[00:11:57] Stewart: Yeah. Okay. That makes sense. I think I just, in my head, we’ve done this by like stuffing cash keys with lots of variables to like make sure that we’re like not cashing loads and loads of things or have really easy ways to invalidate it as I users see it change. But that makes a lot more sense. I just hadn’t heard it called, Yeah, cash segments before.
[00:12:18] but yeah, and I think like you’re absolutely right, getting down into that level of like, what I would call, what do we call it? Fragment caching of like, here’s just this little bit of this page. You know, dump that into Redis or APC or something like that and just pull it back, pull it back as needed.
[00:12:34] but would you say those are kind of then like the bigger things when it comes to development work for these kinda large. sites, regardless of cms, you know, good caching fast, more focus on the actual content and the front end is relatively straightforward.
[00:12:49] Jason: Yeah, I don’t wanna say it’s like, you know, these big sites do go down and, and people, you. I guess the, and equally when they do go down, they lose a lot of money. like a serious amount of money. You know, it can be in the millions if a site goes down during like a, a royal wedding or queen funeral or something like that.
[00:13:04] I guess you’ll find from our side that when we enter twerk, one of these clients, they already have a front end. it’s something that they’ve been battling with for, for years to get it right and they see that it is right. so it is less of a concern from that side. Granted, I. they all fall into the trap of like, because it’s been such hard work to get that front end to work, they’re scared to change it.
[00:13:26] So sometimes the technology behind it isn’t the simplest to integrate it with WordPress to get the content out there. And I think that’s actually a trap that they’ve fallen into because I’d understand if I had been the same, if I spent years been stressed and worrying about the site going down every night, that if someone knew arrived and said we could do it differently, that I’d be like, we’re not changing.
[00:13:45] Stewart: yeah, of course. I suppose then when you say like they’ve got their front end that they’ve kind of already, you know, worked with and they’re very b with and kind of got it to the point where they want, Another kind of thing. You mentioned a couple, a little back while back was decoupled. so for anyone who’s kind of not, isn’t aware of kind of what that is, can you give us a bit of a, a breakdown on what, what you mean by that?
[00:14:06] Jason: Yeah, so, you know, people got decoupled or headless, but essentially all it is is that, you’ve got your backend where you write your content, you’ve got your front end when someone reads it, and it’s just the two are separate. They live on separate systems, so they have to communicate with each other. A good example, which is like always decoupled, is the fact that like an app on your phone, It’s not the website where it’s written on, it uses some sort of API to get that data.
[00:14:27] And so it’s a similar approach, but on the website and a lot of times the reason they go down this route is, is that, it actually does make it a bit easier to scale because you have a lot more options for what you can do on the front end. and I think probably noticing quite a big trend now of people exploring different ways to do this and there is different frameworks coming out as well.
[00:14:47] the P engine have one for example.
[00:14:49] Stewart: Great.
[00:14:50] Jason: And internally, we’ve been experimenting quite a lot with options and in different ways of doing some of this stuff as well.
[00:14:57] Stewart: Great. so total, total buzzword, like the headless stuff get asked all, all the time, about it. Very rarely by our clients, so I think. One of the things I’ve noticed is that like there’s definitely a point in a company’s life cycle where headless starts to make sense. And up to that point, it’s a terrible idea.
[00:15:19] do you, do you agree or do you think it’s good for everyone?
[00:15:22] Jason: Oh, no, I totally agree. the best way, but I, what I think about it is, is there’s an extra cost to doing it. So, and if you’re not big enough, you’re more likely gonna cause yourself more problems thinking and ending from it. And, and I think for anyone who doesn’t really, from a technical side, doesn’t really understand what the problem is, think of the very simple idea of doing a release.
[00:15:41] So you release some new code or a new feature, you have to ship it to two different places and you have to time it right. And it sounds, sounds easy, but even from working in the, mobile app space before, it can be just be frustrating to manage. and I would stick with that. WordPress can go pretty far on its own if you do it right.
[00:15:59] So I would always encourage people to take that point of view. Until it’s no longer able to do that, and then that’s when you go down the decoupled route.
[00:16:06] Stewart: Yeah. any more kind of insider thought on when someone should go decoupled entirely?
[00:16:13] Jason: Well, I think. one, one good scenario might be if, you’re still maintaining multiple publishing systems. So if you’re writing content in two different places, and, this happens in organizations which do print, they’ll have a print system and then they will likely be using the print system to write digital content still, and then they may.
[00:16:30] Bring in, say an agency like one of us and say, write what? Bring more put into WordPress into the mix. Well the problem WordPress serving that content is, it makes it trickier because how do you get the print content in there and out to the front end, like it’s digital format, cuz they’ll still use both at the same time for a while.
[00:16:46] cuz you can’t just switch, turn a switch on and have everyone move over one day. So having it decoupled ax a lot simpler to run two systems at the same time. As for that, I think it’s just, it is just a case of there’s no, no hard and fast rule. Unfortunately, I’m not gonna say there’s just like a number that you know when to change it.
[00:17:02] and if I ever find the perfect number, I’ll let you know.
[00:17:05] Stewart: please do. from my ideas, it’s like similar. It’s around a lot of complexity. It’s like if you just have one site, With one front end and one brand, like there’s no point. You’re just adding so many potential problems. but if you have one, one WordPress setup that you wanna like pump multiple sites with and you kind of don’t want to go down the multi-site route for various reasons, like awesome.
[00:18:02] but I think it’s, it’s an interest. It’s really an interesting one. Or if you’re using a lot of third party services, like if you’ve got a paywall or membership system, the WordPress site doesn’t really need to know about that, but the front end probably does. So I’m kind of coming around on a lot of, a lot of that space being much more where that should live.
[00:18:23] Jason: Yep, I totally agree. I think other thing to remember though, is that we need to go down this route. Some of the simple features no longer work in WordPress, so you can’t preview an article out the box anymore. so there, there is that initial headache of rebuilding some of the most basic features of CMS to get them to work.
[00:18:41] And equally, WordPress’ concept of a published article doesn’t really make sense in a headless system anymore because, A lot of times it you, when you publish an article, you need it to go somewhere, and then go to the front end. But WordPress doesn’t detach the concept of shipping an update to an article and saving an article.
[00:18:58] They’re actually the same thing. and so, it does actually take a bit of work to build in, a more common sense approach for handling, issues. Perhaps the content doesn’t make it out the door. This is if you’re not using, say, the WordPress REST API to serve your site, which I feel like if you’re gonna go, decoupled, you probably should have someone in, in, in between the two, because otherwise you’re still burdened by, the restrictions of WordPress in itself that you maybe, you know, led you down that path in the first place.
[00:19:25] Stewart: Sure. So you wanna publish to like a caching layer in between and then have the, your headless app pull from, from that caching layer rather than, yeah, that makes sense. Yeah. So I think then it’s, well to say that like, Technology and development really impacts on kind of as many pieces of the media businesses as possible.
[00:19:44] Like our, our experience is very much in that front end of like, here’s what the content looks like, here’s some of the user experience of like the membership systems and subscriptions and stuff are in place. and you guys almost come in entirely from the opposite side of like, we build your content management system and we build your editorial workflows and stuff.
[00:20:02] And between, between those kind of two approaches, it. Really is impacting kind of the whole business, in profound ways. I think that’s really to push on to say we’re seeing a lot of change kind of going on at the moment. as we’re recording this, like there’s a lot going on around, large language models coming out, general approaches to ai.
[00:20:22] So it’s clearly a time of like a huge change, both for the whole society, but for the publishing industry in part. it seems like, are you, how do you feel about these? How do you feel about the future, of publishing a media in this new AI world? Are you concerned about it? Are you excited for the challenges of it?
[00:20:42] Are you, do you think it’s an existential threat to it? Where, where are you on this
[00:20:48] Jason: well, I’ll leave that to wi, I’ll leave, I’ll leave it to wiser people about the industry itself, but, I think AI and. Machine learning, it is maybe a bit of a buzzword at the minute. cause everyone’s very excited about chat gdp. and that’s, it’s, it’s right to be excited at the minute because it is, it is pretty impressive what, what’s being done by open AI and other firms.
[00:21:10] Stewart: on an easier level, I’m, I’m, I am excited in terms of some of the stuff that we can build with it. we, before all this,I had sort of kicked off and everyone got excited. we were working on a metadata tagging service, with one of our clients. And, we don’t par the sort of machine learning end of it.
[00:21:25] Jason: You know, there’s another third part that helps us do that. and we do all the interfacing, connecting it through and stuff. but essentially that the concept’s pretty simple. You know, you, you write an article and it suggests the tags that should go with it. and we’ve expanded that to entity link, which is Lincoln upwards in articles.
[00:21:41] So, You know, use a hot topic like Brexit, I guess that comes up so much these days. if you’re writing something like about that, it might link to the most recent Brexit article or a summary or if you’re writing about, say, like an election, it might be linked to the party, a recent article about the party or, you know, it’s just trying to link back to content better.
[00:21:59] the same as related articles trying to be, more intelligent about linking back to an articles that’s done well so that you continue that cycle of driving people to them. That was like initially what we’ve been doing with it. But, I think now that there’s even more support out there, I have a lot of other ideas of what I wanna do.
[00:22:13] Stewart: Great. so are you, are you seeing a lot of your. Clients of folk you’re talking to, like starting to within the editorial side, like use AI more for kind of content generation or has it been very much in the support and automation of things like seo, things like tagging, things like the recommendation engines.
[00:22:36] Have you very, like, has it all been kind of supporting work or has it moved into more of the content side?
[00:22:41] Jason: I think it depends on the publication. So, some publications I guess, would, would argue obviously the, the right quality content. They do a lot of investigation pieces, and for them, I think the idea of letting AI write the article would probably be crazy. so they, they see it as, I always, I always sort of summarize it as a bit like clippy, you know, you used to have in, in Word, you know, that little like, helper.
[00:23:01] I think that’s what AI’s best for. It’s like, oh, you’ve, you’ve made a spell mistake there. Or, oh, you, you know, this doesn’t fit our style guides. you know, it’s, it’s those kind of things. Or oops, don’t use that word. We know that. Or that person’s name cuz we know that, is an issue illegal. and, and trying to guide them in writing their articles better.
[00:23:16] Stewart: I hadn’t, that’s, that’s a new one of like kind of almost being able to train a model on your. Your organization’s particular slander and defamation history, to be like, don’t, don’t talk about this first. That’s, I hadn’t thought of that one. That’s cool.
[00:23:30] Jason: We’ve, we’ve built, we’ve built it as a prototype a few years ago. it hasn’t made a hint of production just yet because I think. Maybe the tech, the sort of machine learning AI side of things wasn’t quite up to scratch enough, but that’s actually changed of late. So I feel like some of this stuff’s gonna get re rebooted over the next year.
[00:23:50] Stewart: Yeah. How cool. but that’s, that’s good to hear that there are like orgs kind of at your scale that are like, you know, we write our content and it is a human being doing important work, important like journalism, like capital J, and then like, we don’t want AI impacting upon that. Like, let’s use to support what we’re, we’re doing rather than replacing replacing the people.
[00:24:14] Jason: I do know of one particular, I’m not mentioning obviously the brand, but I do know if one, that does use it to write content in some respects. And the way it works is, the best thing about it is it’s more of a, a mass syndication piece. think about like the Weller, for example, like. It’s a pretty basic thing.
[00:24:31] You can get the data on the Weller, so you could say like, oh, there’s a storm coming, or, you know, it’s a sunny day or whatever. So the idea would be that for a company that has local papers, which is maybe not as common these days, but that you write a template for just Weller reporting, and then you have the data pass in through AI and you basically print a ton of articles for each local area around the Weller.
[00:24:52] Stewart: All right.
[00:24:53] Jason: And you can do the scene for like roadworks road closures, just like. You can feel like filler pieces that they can put on the site still generate traffic. so it is being used for those.
[00:25:03] Stewart: Yeah, I suppose like the open secret is that sports content has been generated this way for a long time.
[00:25:10] Jason: Yeah, I guess, I guess, you know, I mean sports desk might disagree cuz sounds like quite high pressure job to be in some, in some, like how fast you have to get the content out. But you are right in terms of like, A sports article, a lot of times they’ll have, pictures from the match. They’ll have, the league table.
[00:25:26] the score, it is probably one of the more templated concepts, I guess is the best way of thinking about it.
[00:25:32] Stewart: Yeah, but I think because it fits, that’s the model of like, it’s very data driven and there are facts. There are like indisputable, like this, you know, point was scored at this time by this person that can be broken down into like that kind of factual. Play by play of what happened. I’m not saying that there are agis writing, there are new agis, there are new LLMs writing, you know, analysis pieces that like people, you know, from columnists that specialize in that sport.
[00:26:01] That’s definitely like, comes under the capital J journalism. But yeah.
[00:26:05] Jason: I guess probably not yet, but at some point there will be, I guess is, is maybe why people do get concerned. But, you know, four years ago, we worked on the sports concept a little bit in terms of,we didn’t wanna call template because Tim Platon sounds rigid. So we built something called scaffolding, and the idea was that you could scaffold together the common elements of an. And, and essentially have variables in the article. And so when you need to do the match report, typically a match report has to go out within 10 minutes after the match. So it is a little stressful for them to get out on time. So the idea was that can we do stuff like put the team names in, fire the league table in get scoring?
[00:26:41] and so it, it plays into what you’re saying there and the next step would. Can something pick the images quicker for them? Can they write the captions and stuff? So, and I think actually that is probably a bit easier to do now. So it’ll be interesting to see where we get in the next two years with that stuff.
[00:26:55] Stewart: That’s cool. Yeah, that’s really good. Good pl good idea. within the whole LM thing, have you. Have you looked much at experimenting with these LMS for co-generation at all?
[00:27:07] Jason: Funny enough, this week we, we, it finally enabled, a co-pilot. I get’s co-pilot,
[00:27:12] Stewart: really?
[00:27:12] Jason: coding assistant, on some of the more senior devs in the organization to trial on it. I’ve been wanting to do it for a while. I guess the concern is around how does it affect junior members staff? Is actually gonna stop them learning.
[00:27:26] and so we’ll try it. And I, I kind of feel like we may keep it to senior people because although we’ll save time, I do think it’s important to make all those mistakes from the very beginning. but I’m excited for myself because, like you, I suffer a bit, it’s dyslexia and can, a lot of typos and misspellings in my code and I feel like.
[00:27:45] I sometimes don’t notice it right away and it can take a while. For me, the track of down, especially, cause I don’t, when I read, I don’t always spot what I’ve done wrong, that someone like a, an assistant like that would really streamline where I’ve made those mistake.
[00:27:55] Stewart: Yeah, that’s cool. But I wonder if. That particular example is, should that not be solved by like the IDE or your development environment being like, Hey, this variable doesn’t exist. It’s undefined, or, or the opposite way of like you’ve defined it somewhere and then it’s never used. I just suppose it just depends on, on your set and we’re getting in the weeds there with that, but that’s what I’ve handled, that
[00:28:19] Jason: That’s true. I guess maybe, Because I’m from a messier mindset in life. Maybe I don’t take the effort
[00:28:26] Stewart: Yeah. Type systems exist to help me with spelling, not for actual, like getting the right type back from a function anymore.
[00:28:34] Jason: well, like I think this is the thing isn’t, it’s like I said to the developers, I said, oh, we’re a like, you know, copilot. And he’s like, all right. He’s like, cool, well what’s it do? And I was like explaining it. And he is like, well, I’ve got snippets, right.
[00:28:45] Stewart: Mm-hmm.
[00:28:46] Jason: And I was like, yeah, but it’s gonna be better than snippets.
[00:28:49] He’s like, I don’t know. And I was like, he’s, he’s very organized. So he’s one of those people that has all these snippets and, and like he, he was like, I’ll not see the new in as much because I’m so organized. And it’s true. It might not offer as much value for something like that, but for someone like me who probably doesn’t do all the setup that he should do, anxiety, then it’s, the box just makes my life easier.
[00:29:08] Stewart: Yeah. I think with things like that, the like, There. I think two really useful things of like, I just need this to do the same thing over and over again. So like writing loops and things like that that are just processing the cm, like getting your maps and your reduced and whatever. It’s doing the same thing over and over again.
[00:29:25] Like it’s invaluable, but also going the other way of like, I’ve written this and I need to explain it to someone else in the future, including me. Can you give me a comment back for like doc block that like helps with. Understanding in the future when this is all dropped outta my bread.
[00:29:41] Jason: Well, you know, Even the best developers sometimes don’t comment their code enough. I think that is one value of having, co-pilot and or tools like this involved because for it to generate the code, you have to have a pretty good comment written for it to know what you want. So in some ways it, it really encourages the right full featured comments in order for it to save you time.
[00:30:01] And I think that that actually might be one of the best bits about it because I think we may end up with a lot better ACC code. Well, The stuff it provides, is that useful or not? It doesn’t matter. At least that would be a benefit.
[00:30:12] Stewart: Yeah. Yeah, absolutely. outside of co-pilot, which is a product, you know, from GitHub targeted at developers to, you know, help them do what they wanna do better. what about tools back into the chart, G p t open, um, GB four and stuff like that? I’m trying to, that are targeted much more at your. The day-to-day internet user trying to get by, and we’re seeing a lot of people do code generation in it. Are you worried at all about being a developer in a world where it is so trivial in some ways to generate code in these ways?
[00:30:49] Jason: I’m not worried. I think YouTube. Let’s anyone figure out how to do inning, I guess like diy, right? I could, I could look at YouTube, find how to do stuff, and so sometimes I’ll do DIY myself cuz it can do it. But it hasn’t put tradespeople out of demand because they’re still high demand aren’t, and it’s still something that people need.
[00:31:07] So I think it may mean that people are more empowered to do simple tasks. Maybe, they’ll make small edits to sites and feel more confident because they have someone guiding them. But I think at our level, Changing so all.
[00:31:21] Stewart: Yeah. Fair. Cool. the way I, I was thinking about it the other day is like, repeatability in, in software is really important. You give the CM input, you need the CM output. If you’re building an app with chat, e p d at the moment, you’re gonna get different things every time that are kind of gonna introduce subtle different changes and differe.
[00:31:44] Modifications. And then how do you maintain that as requirements change? annoyingly I think all of these problems could be overcome, but it comes down to, like you were saying, like I didn’t replace all these like trades of people when you could DIY with the development stuff. I’m like, have you ever met a client that could accurately articulate the requirements for software?
[00:32:08] So in my world, I think I’m looking. Developers using things like chat, PT to, you know, augment the work they’re doing, but they still exist as that important translation layer between what is stated and what is actually needed. Does that make sense?
[00:32:23] Jason: That makes sense. I mean, where would we be without project managers and product owners? we need people like that to help get out the details of what. What the person actually wants. and experience developers feedback and say, I feel like this would be an issue here. And we have testers. And you know, I guess as you said, all those roles aren’t taken into consideration when it’s just one person chatting to, to an AI to get a piece of code.
[00:32:48] maybe one benefit will be that it starts a generation of self caught top coders. You know, like imagine we, you know, we were reading books in, secondary school, to learn the code and I guess like if that existed, maybe we would be chatting online on the evenings and, you know, maybe that would’ve taught us a lot of different stuff.
[00:33:06] Maybe taught us better than we could have ever done from a book. So that would be awesome for our industry, I guess. Better to have more.
[00:33:12] Stewart: Yeah. more people. I think that brings us nicely into some that I’m really interested in kind of from your kind of approach within Big Bite is, mentoring, some more kind of insight. Inside Big. Tell us about the mentoring, program with Big By how you bringing in juniors or creating juniors, I suppose as as it comes up.
[00:33:34] Jason: Yeah, so you know, I guess, for context, when I finished uni, I didn’t really know what to do with myself, so, I was just working in CX in, in town and kind of giving up on programming and did try and get a job, as a programmer. but they made me sit in a little room and do a test and like, Being the slack and abusing a piece of paper.
[00:33:53] It’s not how I could, it’s just not. And I just couldn’t, couldn’t land a job outta that. And I kind of given up, but I stumbled my way into essentially, working for free by emailing companies and stuff. So, and that eventually led certain big fights. So I’ve always had a passion for trying to find people like me who, maybe, you know, not to be bigheaded, do have talent, but just maybe not as obvious. the standard recruitment process. So, we’ve tried to do lots of things like higher apprenticeships and uni grads and whatnot. And then a few years ago, I had this idea, I was a little bit optimistic. I decided to find six people. obviously we’re based in northeast, so we were gonna pay in 21 grand a year each, which was decent enough to live off, off there, and enough to, for someone to change roles, so they could come across from a different job.
[00:34:38] And we essentially trained them to be developers over six months. And, and I was teaching them and then, they went onto projects, and it was all right. You know, we, we had, we still have, four of them today. and another one is working elsewhere. so only one person didn’t actually end up being a developer by the end of it,
[00:34:55] Stewart: an incredible success. R like two-thirds are still in there. Like another person still working in the industry and only one, only one out of six kind of didn’t make it like that’s an incredible success actually. Just taking a moment to think on that. Sorry, I interrupted on you.
[00:35:13] Jason: it, it does, it is, it is when you look at the like that, but I feel the problem was it took a lot longer to get them to the level that I expect it. And, and I, you know, I don’t wanna blame anyone apart from myself, but it, the Covid didn’t help cause we had to do it during lockdown and it wasn’t a great time for everyone.
[00:35:33] And like these people learning, it just changed job roles. They were locked in their house. And it was tricky for everyone to kind of figure out what to do next while we were teaching. but we got there in the end. But I, I find a new appreciation for teachers in general, like people who actually go out and teach day to day and how they could transfer that knowledge because it was pretty clear at the end of it.
[00:35:53] I might be able to teach myself, but I can’t necessarily teach other people that easily. and the success of those people being here today isn’t my success. It was the success of the team because once I had finished with them, they had a long way to go and our team got up and it was quite a burden on a lot of people.
[00:36:09] And so we didn’t redo it because this was gonna be a yearly process where we do this training academy every year, but the team were quite worn out by it. And, and that I said we’d have to rethink how we do it. We’ve just launched a job role, last week, and it’s essentially an engineering mentor, and the idea is that we basically find someone who can teach.
[00:36:27] Uh, who then can run these academies properly, and know how to deal with questions and ensure the people are learning. and also when they finally move on to projects for them to manage questions that come back. Cuz a lot of times what happens is when you get a junior involved, they sometimes ask the same question to different people.
[00:36:43] So they don’t actually learn. They, they, they get answering in two weeks later, they ask another person and no one knows that they keep asking the same question. So no one’s actually saying, well, hold on. Why don’t you understand this? And so we want a dedicated person to always be there for these people.
[00:36:55] so it’ll either be a teacher who has loads of teaching experience, who’s interested in technology and has some background, hobby, programmer maybe, and we’ll teach them how to code. Or it might be an experienced developer who’s decided actually, They have quite a passion for teaching. so hopefully we find someone really interested and we can get this up and running soon.
[00:37:13] Stewart: And so that’s, that’s a difference from the last time. It was, it was you that was running the, the teaching day to day for the, the last cohort.
[00:37:21] Jason: Yeah,
[00:37:22] Stewart: Yeah.
[00:37:22] it was my idea, so I thought I’ll, I thought my idea. I’ll take the, I’ll take it and see where it goes. Great.
[00:37:29] Jason: Overconfidence was the problem. I just assumed it was easy and it’s not easy.
[00:37:35] Stewart: Nope. It’s not easy at all. but that’s awesome. Right. So, and is this someone in the northeast that you’re looking for, I imagine cuz most of your teams all, be a citizen in that area?
[00:37:46] Jason: Yeah. So we’re, we’re not fully remote. Like a lot of agencies might be. I guess we’d be hybrid these days. You know, the idea is at least if you can come in the office one day a week, it, it gives us a bit of a, it’s hard to explain, but, you know, people might call it culture, but it just keeps the team in sync a little bit easier.
[00:38:01] We often talk about if we didn’t have an office, what it would be like. But I think it’s nice, and I especially think with juniors that they need to be in the office. I think we learned that during covid, like it’s much easier to spot when something’s going wrong or they’re feeling a bit upset about something.
[00:38:15] And, in, in that effect, actually I really enjoy it from home these days, but I’ve realized that I’m doing a disservice to some of the younger members of the team by not being in the office. So I’m trying to get myself going back three days a week soon so that I can try and offer my.
[00:38:27] Stewart: Yeah, absolutely. I, I completely agree. we have always struggled with like junior members of staff because we don’t, as a remote company, don’t have the eyes on them to be able to like, support them and kind of pick up, pick up those cues. I get asked about work experience all the time, and I’m like, I can’t in good conscious.
[00:38:47] Bring someone into your work experience. Cause we just couldn’t like look after them in the way that they would need to like support them as they get like set up. Because it’s not like web dev five, 10 years ago where it was install map and kind of get going. It’s hugely more complex now. but yeah, I definitely
[00:39:04] Jason: would, I would jump in there and say that, we have said no to work experience quite a lot in the beginning. And there was one guy, who wouldn’t take no as an answer. So we gave up and we let him come in, and then we hired him. and he was like 16 when we hired him. and we had to get his mom and dad in, because I was quite concerned. If it didn’t work out for him, we would be affecting his education. cause that’s like, it’s quite important to get your A levels. so we made an agreement for him to continue his A levels to which very quickly he decided he didn’t wanna do that anymore cuz he is enjoying work so much. So, but I guess I agree it’s hard and we’re trying to think if we could do more work experience stuff, but,we, we just need to, as you said, you need to make sure you’re giving them the support that they need so it’s worth.
[00:39:46] Stewart: Absolutely. and then I imagine the, the cohort you’re looking for to kind of go through would kind of need to be up and around, the Middlesborough T side, northeast sort of geography to Mela work.
[00:39:57] Jason: I think for the, for the next set for sure. Just because until I can prove that we can run it. and if that engineering mentor says to me, Why don’t we just do it remote, next time, because then you can have a wider pool of people then We’ll, we’ll give it a go. I have no idea where we’ll be in a few years time with stuff.
[00:40:13] So, the, the way things are changing, maybe we’ll end up fully remote, who knows? But for now I think we need to make sure we can do it properly before we trial. Something like that.
[00:40:21] Stewart: Yeah, of course. science. Incredible. So all the best of luck with finding, finding an engineering teacher. I’m certainly keeping an ear out for you, and a cohort to go with it, but that sounds awesome. anything else that’s on your mind? I’ve kind of got through everything I wanted to talk through.
[00:40:41] What’s, what’s, what’s interesting you these days?
[00:40:44] Jason: Yeah. So like, I guess the, the, the more excitement development for us is doing our own product, for like the proper first time. I think the reason it’s exciting is just that it wasn’t my idea.
[00:40:55] So there’s been loads of times we’ve tried to build products and I guess I’ve like dictated my idea, which means that we. Well, we’re just, I’m like the, I’m like the worst client, so, because no one understands what I want. whereas this time it was a, a member of the team went, Hey, we had this conversation about our project working on, and we were like, wouldn’t it be good if you could do this?
[00:41:12] And we were like, that would be a good idea if you could do this. And he was like, I’m gonna go build it. And so we supported, him and building it by reducing his time on client work and. It’s fully built, it’s been built for a while actually. surprisingly the bit that took the longest, which is actually sticking the marketing site together cuz
[00:41:26] Stewart: Oh
[00:41:27] Jason: like fitting in between stuff and then people are like, oh, I gotta do some WooCommerce.
[00:41:31] And I was like, all right, can we get the terms done for legals? And so we’re at that point now where,It’s pretty much ready to go live. So the site’s up at the minute, it’s got a few little tweaks to make, but it should be live when this is aired. So it’s called Content Query. So it’s content query.com.
[00:41:46] and hopefully there should be a nice little demo video on there. By the time you visit it,
[00:41:50] Stewart: do you want to tell people what it is or are we gonna, do you wanna leave it vague and let them find out?
[00:41:55] Jason: I’ll explain it in a really per, in a really per way because. know, it’s not even my idea, so I will tell it, but someone will probably go. That was a per way of explaining it. But in short, the best thing about it is obviously WordPress, Gutenberg these days has blocks and,they’re news all over the place, so you’ll have your image blocks and whatnot.
[00:42:12] The problem is, is as you build blocks, you have to change them. And when you change them, sometimes it can break old content, if you’re not careful. So it kind of becomes this bit of a maintenance headache. The more and more a site gets old. Essentially the idea was born. That wouldn’t be great if you could find where blocks were being used and which versions were being used.
[00:42:27] So it gives you an easy way when you’re making a change to a block, go find some more content to make sure your change isn’t gonna affect it. and so marketing teams can sanity check deployments that are going out. They can be like, oh, why don’t we just check a couple pieces of content that use this block and they can find them.
[00:42:42] It is all our features around, find out what plug-ins are being used across which multisite and. You know, find out which media is being used, different places as well. So, it, it, it’s basically querying your content for different reasons. So it’s to make it a little bit easier. And the site that it kind of encouraged us to do this was, we have a site with, um, EI Subsites.
[00:43:01] so it’s a big multisite.
[00:43:02] Stewart: Yeah.
[00:43:03] Jason: And the team often run into this is a problem. It’s been going for quite a few years now. The thing’s massive. and so we built it based off that kind of mindset. And so, it’s free to use unless you’re a certain size of company, which then you’ll need a license for.
[00:43:16] So hopefully that means plenty of people can just use it and enjoy it. And then any real big companies can contribute towards us being able to.
[00:43:23] Yeah, that sounds awesome. just looking at it, would love to see, short coats included in the things it’s able to look at, for those kind of legacy sites that maybe not quite mute it to the block editor yet. I will, I will pass that onto the team. We had to stop making features because it was becoming uncontrollable, the amount of features that we’re getting at it. So at one point we just put our foot down and we said, let’s just stop changing this plugin. It’s gotta go out the door, because until it’s being, people have their hands, we don’t really know.
[00:43:53] If it’s something people want, but, hopefully it is, it’d be nice for us to have a, a product so that the team had something interesting to work on outside of the day-to-day. And, and also maybe if, if there was a good financial aspect to it, then it just means us being able to pay the staff even more, which is, is something that I have a lot in my mind, you know?
[00:44:13] Stewart: Awesome. That’s really great. I look forward to, to trying it. I, I can think of somewhere. We could definitely, definitely use this one. awesome. Thank you so much for your time, Jason. if people wanna connect, find out more about you, big bite, content query, or if they’re interested in becoming your engineering mentor, engineering teacher.
[00:44:35] Cause it doesn’t, sounds like you’re being mentored. I think we need another way to address that. But, where can they find you? Where do you want them to go if you want them to at all?
[00:44:44] Jason: well, if you’re looking at the job, I think it’s probably best going through big bite net and going to the career section just because,I’m just a bit messy, so sometimes I might forget the responder stuff. so if you, if you do wanna go through the form process, you’re probably gonna get better taken care of by the team and, and they’ll make sure everything’s fair and, and done the right way.
[00:45:00] Whereas you contact me directly, it’s, it’s, it’s probably not the, the best route for everyone cuz some people might not contact me directly, wanna keep this as fair as possible. but if you, you know, like I’m on Twitter a little bit these days, so, it’s a stupid user name cuz it’s just Agnew with two underscores.
[00:45:15] But, I’m sure people can find me. And, if someone wants my email, it’s firstname.lastname@example.org. So if, you know, if, if you, if you really do wanna reach out, you’re more than welcome to. Can’t promise a response because, I’m just a bit of chaos sometimes with this stuff, but I will do my best.
[00:45:30] Stewart: Of course. thanks again.
[00:45:32] well thank you all for, for listening today. If you’ve enjoyed the episode, reach out on Twitter. You can find me, at Chair Richie. there is also a ma on handle that I can’t pronounce within my Twitter bio, so feel free to follow me there. we’d love to know what you think about the episode, and it really helps a lot.
[00:45:50] If you could leave a review, and rating on iTunes or whatever it is you listen to your podcasts. and we will see you again in about two weeks with our next episode. Thanks for listen.
A modern media podcast
hosted by Stewart Ritchie
In this week’s episode, I had the pleasure of speaking with Jason Agnew, the founder of Big Bite, about the impact of technology on media. Jason shared his experiences working with major media brands and the challenges of larger and more complex projects. We discussed how having a larger team can make difficult tasks more manageable and delved into the differences between smaller and larger sites.