Pulling the Curtain Back on WordPress, WooCommerce and AI

Dave: Hello, and welcome to Do the Woo shiny, exciting tech podcast. I’m Dave Lockey, one of the co-hosts, and today I have only one other person on the podcast with me, because the Americans are Americaning, as they generally do on July the 4th. But we’ve got a… How do you describe yourself? You are an Australian but in America, so we get to access you, Dan. I’ve got Dan Walmsley from Automattic.

Dan: That’s right. I’m being actively sacrilegious by working on the 4th of July in America.

Dave: Yeah, that is subversive.

Dan: Oh, absolutely, yes. I like to stick it to the man by working harder for the man.

Dave: Joyful subversion. I’m not sure that’s quite what they intended by that, friends. Thanks for joining me today, Dan. We talk quite a lot behind the velvet rope of Automattic, but I thought it’d be really fun to have a conversation out in the open for the broader Woo ecosystem, and to pick up on some stuff that we’ve talked about before, but also to have a broader chat, hopefully not mentioning any company secrets. So tough.

Dan: It’s going to be difficult for us because we’re totally used to being like, “Yeah. When we build those rocket ships to Mars, everyone’s really going to be like, ‘What happened to Automattic?’”

Dave: Yeah. Poor brains having to handle different context. So I think first things first, let’s do some hygiene. Would you please introduce yourself?

Dan: Absolutely. My name is Dan Walmsley, although my girlfriend calls me Daniel, which is adorable. Only she and my mother get to use my full name. I am the interim head of AI at Automattic. I work on the applied AI team, mostly as an engineer, architect kind of person, but also doing a little wrangling right now while Andrew Spittle is on sabbatical. He is the true and canonical head of AI.

Dave: How did you end up at Automattic? Give us a little brief history then here, please.

Dan: Yeah, so I had bounced around various startups for a few years, mostly in a CTO capacity and mostly around activism and sort of democratizing certain technologies. Specifically, so when Obama won in 2008, there was a sort of big surge of interest in digital organizing and the impact it could have both for nonprofits and for brands, and so I worked at Purpose as a CTO trying to sort of platformize digital organizing there and then more directly at NationBuilder where we got venture funding and were able to turn that into a big product that served lots of people. So there’s always been sort of strains in my history of sort of activism and technology and sort of entertainment. I was an improviser for many years and play music and things.

When I had a baby, I decided the best place for me to work was Automattic because I really wanted to work still on democratizing technology, and so there was a strong alignment of values there. Also I wanted to work remotely so we could live near family in rural California, which we do, and it’s a magnificent life hack, I have to say. So I’ve been here for about, gosh, coming in on nine years perhaps. Yeah, and it’s the longest I’ve worked anywhere by a factor of over two. I don’t think I’ve ever worked anywhere longer than four years, so I think that says a lot for the company. I have a short attention span, but they’ve managed to hold me.

Dave: Yeah, that’s impressive, I think, for somebody that’s used to flitting around and chasing squirrels to have your attention met. I imagine you’ve done a few different roles over your time. Perhaps we’ll speak to some of those. But I think probably what folks will want to get out of this episode, if nothing else, is what’s Automattic doing with AI? What are we going to see changing with WooCommerce or WordPress.com or Tumblr? What are we hatching that you can talk about?

Dan: Yeah, so I mean the first thing to say is something that I think everybody is aware of, which is that the AI industry is moving at lightning speed and even the most fundamental aspects of AI see new announcements every week, whether that’s in the usefulness of models or their size or the ability to run things that used to take huge compute clusters on a single laptop using things like stable diffusion. So just the first thing I would say is everything has a giant asterisk next to it, which is it could all be upended next week by some new technology. But a lot of what our team exists to do is to take the technologies that we can take to the bank in the present moment, so say like GPT-4 with functions or something like that, this APIs and say, “Well, what can we do with this today that wasn’t possible six months ago?”

If we build a lot of different prototypes and approaches, can we find one that creates an inflection point for the company, whether that’s in the cost of support or in helping users to shortcut actions by composing tool-based workflows from a single statement, or whether that’s providing customized support docs that are tailored for the problem a user is having and the features they have enabled? Obviously editors say… Then there’s a lot we’ve already announced on that front. There’s a lot we’re already doing. People might have seen that we’re generating answers in the support widget. They probably have seen that the Jetpack AI plugin, which generates layouts and images and things like that.

What we’re working on internally is basically subsequent generations of all of those things. Like a lot of other companies, we’re trying to do it with an eye towards business impact. The first pass might have been more about feasibility or learning, but the real meat of the work is in how does this integrate with the business, how does it support the business, how does it affect the fundamental metrics that we care about? Because these technologies are not cheap. GPT-4 is an incredibly expensive and terribly bad database, if you use it for that, and so you have to make sure you’re using the things it’s good at.

Dave: Right, okay. “The things that we’re doing are the things that you would expect us to be doing and we are going to keep doing them.” Is that the bottom line?

Dan: Yes. Well, I would say on the more interesting front, where I see opportunity and momentum for us is in building classifiers out of the data that we sort of uniquely have, and some of the data that we uniquely have is that if you are using our hosting platform or our tools to build a website, we have a real-time stream of the data that is necessary to provide that functionality, whether it’s indexing or helpful suggestions or other things. That data provides an incredibly powerful foundation for AI assistance because we know what you’re trying to do while you’re doing it, you’re using our tools. That puts us in a position to help in a way that other tool vendors might not be able to. We’ve invested the last 10 years in Jetpack Sync, which is an underlying data synchronization architecture that means that all of these cool indexing things and stuff are just fire and forget features that work on any site.

When I originally architected Jetpack Sync, I said it needed to be able to run on a Raspberry Pi strapped to the roof of a four-wheel drive going through a tunnel connected via 3G, and so if there’s ever a glitch, it needs to be able to be bulletproof to that. Now that that foundation is bulletproof, and it really, really is night and day compared to how it used to be, there’s so many AI features you can build on top, many of which are pretty obvious and don’t need to be spelled out, but just to help you get where you’re clearly trying to go at any given time.

Dave: Yeah, that’s amazing, isn’t it? I think there’s something about having made that investment in digital infrastructure, and I guess the business change always comes with integrating new technologies at that time 10 years ago and continuing to invest in that. That then allows for this next wave of evolution as we layer in these new emergent technologies. I sort of feel that despite the fast-changing nature of this current generation, this week’s generation of a AI tech, and the fact that to some extent we have to ship about and find out as the context and the technology’s changed beneath our feet, investing in the learning and the tools and the nuance that comes with actually being hands-on with product will perform that same function. It’s good hygiene, it’s change management within the business, and it’s scaffolding for whatever comes next, whether that be in the next month or the next week or the next year, and probably, at this rate, tomorrow.

Dan: Yeah, generally speaking, when it comes to contemplating how to shape a feature, for me there’s at least three major considerations. One is over the long term, is there something foundational about this? Is there something fundamental about this? For example, we worked on this chat feature for Seth Godin’s blog and there’s a few other blogs it’s on too, and there there’s a lot of improvements of that coming down the pike. It was very much a first pass, but we believe that chatting with a blog, making it possible for someone to take any knowledge base, whether that’s something they’re curating or it’s a long tale of their blog history and making those recipes, those diary entries, those words of advice or whatever else that you filled your blog with accessible through a chat interface, we believe is, well, I believe certainly is a fundamental new both user interface paradigm and a much more powerful way to interact with content than we’ve had in the past.

We believe that this is something that every site owner should be able to tune for their own use case. Much like with ChatGPT, how you can plug in tools, what does it look like to plug AI tools into a chat experience powered by a WordPress website? I think a lot of people are thinking about that and we’re thinking about that too. We just think there’s a huge amount of value. Is there something fundamental, is one question, and then are there many possible different use cases, all the way from a free site up to an enterprise site? Can we imagine that same feature, say chatting with a blog, being useful? Well, yes, if you provide customer support, if you provide advice, if you provide recipes, if you provide stories for kids, those are all different personalities and knowledge bases that can be delivered through exactly the same user interface paradigm and set of tooling. The adaptability of the GPT series of models means that we can use the same stuff to serve all those things, so that’s great.

Then in the short term, the question is, and this is the third piece, does it solve a pain point people have today? Is there something missing in their lives? That part is a little harder. I feel like we are in a moment where people aren’t quite sure what this AI thing is for yet or how much to invest in it, but I do think that for many products, people are intrigued by AI and willing to try it as an add-on to something they already know. Oh, I was going to buy the pro plan. Oh, it comes with an AI thing. Well, I’m definitely going to get it now and try it out for a week and whatever and see what that’s about because I want to know what AI can do for me.” So we’re in a moment like that, that that’s kind of a bubble where I think that’s going to tip a lot of AI companies’ conversions over the edge in the time being.

But the long-term test, the long-term tell is are you actually providing value? Do you actually have a moat? Can you be eaten alive by someone else with better data? I think at Automattic we have many, many, many possible things we can try that all of which look really, really promising and are based on, again, the reason why we invested in Sync is we asked the question, “Is it fundamental? Does it enable multiple potential use cases and does it scale from simple sites all the way to the enterprise?” And the answer for all three was yes.

Dave: Yeah, does it take away pain?

Dan: Yeah, since that time we have rolled out at least four or five different technologies on top of that same backbone, including security scanning, indexing with Elasticsearch and a whole bunch of other things. I feel like some of these chat interfaces and other things will be the same way, the universal platform on which you can build quite distinct experiences for different parts of the market.

Dave: It’s kind of difficult to imagine generative AI now being anything other than table stakes for any digital experience. I don’t think that’s true yet because the adoption curve for this stuff, whilst it’s impressive, there’s still a lot of people that aren’t using it or at least don’t know they’re using it. It’s just like, “Oh, wow, this got better and easier.” Do you agree? Do you think this is sort of a differentiator or that it’s just hygiene, or a bit of both?

Dan: I think we’re finding out what’s possible. Even at Automattic, where we have more resources and more, I think, opportunities perhaps than other people just because of the sheer variety of different products, we’re still finding out where it can be a real differentiator. Part of that is the big asterisk at the beginning. What you thought was a moat yesterday is gone today and what you thought was impossible yesterday is possible today. All you can do I think is just take your best shot every day at what’s the most important thing to work on, like, “Where can we move the needle in the short term while not forgetting the long term, while considering what’s fundamental, while considering what’s universal, but what can we ship today?” We ship a lot of stuff internally for Automatticians that never sees the light of day, but are attempts at different ways of solving a problem. Our team is no different. We’re shipping all kinds of craziness at short notice with no tests or whatever for Automatticians and just seeing what happens and gathering feedback and going for it, which it is the Automattic way.

Sometimes there’s different lengths of time between when something turns out to be a good idea and when consumers get it, depending on how it scales and all that kind of stuff. But it’s one of the funnest parts of my job is just like, “Hey, what do you think of this?” Take some feature and deploy it into Slack so people can quickly interact with it before we throw it on a website and that kind of thing. I remember a story from many years ago about how they designed the Palm Pilot, and apparently the designer of the Palm Pilot carried different sized blocks of wood in his pocket and put them in different pockets and pulled them out at different times and asked himself, “What would be the most useful thing this block of wood could do right now?”

Because of the way that humans tend to anthropomorphize AI, you can very quickly trick a human into asking an AI to do something that is currently impossible, but which ought to be possible. You can say, “I’m an AI and I’m capable of this or that, what would you like to do?” Of course, the human will always, based on what the AI says, will ask something that’s either possible or impossible, and then you just have to classify that and see whether those impossible tasks ladder up into something that you have to bring into the realm of the possible.

Dave: Yes. We are good at asking for the impossible as people.

Dan: Yes.

Don’t miss Avalara’s Next conference for developers

Dave: Let me change gears a little bit. One of the things that I think we’re still passing out as people is to what extent does ChatGPT become the interface through which you then route to many or all other applications in your life? When do you go to a specific application with embedded ChatGPT? And I use ChatGPT here as just a generic sort of, like, Hoover being a vacuum cleaner.

Dan: Right, so AI-based chat, the paradigm, versus ChatGPT, the implementation, which we probably should be clear about that, because I think it’s clear to me that chat is really useful, has been around for a long time, and is going to become even more prevalent now that you don’t have to have a human on the other end of it. It’s just closer to the way humans natively communicate with each other, and that reduces the barrier to success. Rather than Google where you have a single prompt and you refine it over time, just being able to ask questions of whatever data sets sit behind the model is really powerful.

But ChatGPT itself, the company and whatever, it’s possible that, say, Apple comes along and says, “We’ve shoved a big enough GPU in every one of our devices, that you can run an instruct level language model, say a 30 billion peran language model or something, maybe not in your phone, but maybe in a brick in your house. You could imagine the Mac Mini getting stuffed to the freaking gills with M2s, with neural processing units on them or something, and sitting there running a language model that is totally private to you and ensuring that no personally identifying information ever leaves your network.

If you’re going to take full advantage of the power of an LLM, you might feel uneasy being entirely dependent on one company like OpenAI. You might feel like, “I want own the hardware this runs on.” That’s a possibility. I mentioned that only to say, if that becomes the paradigm, then people who make hardware win, and that’s not OpenAI, and that’s not necessarily Microsoft. If Apple is able to say all the important stuff actually happens at the edge, then OpenAI has no nothing at the edge. Whereas if they’re able to say, “Well, all the important stuff happens in the cloud,” OpenAI wins because Apple is at a disadvantage when it comes to those hosted models right now.

Dave: Yeah, and I guess I tend not think of this so much as winners or losers, but just the different roles that these companies are going to play in the ecosystem. There’s almost certainly going to be space for both of the above. There’s going to be room for cloud-based OpenAI, and there’s also going to be, I think where Apple’s proven the value of its investment in hardware is when it wants to do stuff like Vision, that just enable experiences that nobody else could build and potentially open up a whole new… I don’t even know how you describe it, sectors, categories, paradigms of experience like genuine innovation, that I think then a lot of value accrues to as they become the platform for that.

Dan: Yeah. I will say, though, it does seem like this plugin idea where the large language model kind of exists as a orchestra conductor or a railroad switch operator between different AI tools that may be published by different vendors. One of the most interesting developments at Microsoft Build was that they adopted open AI’s specification for AI plugins, and it’s very webby. It’s like a file hosted at .well known/aiplugin.json. It describes authentication endpoints like OAuth that maybe obviously use HTTP and JSON. It points to a Swagger like YAML or JSON file describing the endpoints and their metadata, their inputs and outputs in a way that can be paused and made useful by an LLM. With context, Windows are getting to the size where you can shove a whole lot of these tools in there at once.

The only role in those situations that the LLM plays is marshaling data in and out of these… deciding on the best sequencing of actions to solve the problem using those tools and then marshaling data in and out of them based on the description. So, “Oh, this date’s in UTC, this date’s in GMT, maybe it has to do some conversion and stuff.” In that situation, the orchestration AI can be pretty dumb, you know what I mean, in conceptual terms. You’re not using it as a database, you’re not using it as a therapist. You’re just using it as a thing that says, “Given this set of tools, what’s the right set of steps to solve this problem, the problem described by the user?” That’s a really cool thing because it actually is much more decentralized in a weird kind of way than the current web where so much is dependent on a handful of social networks and Google as the kind of attention aggregators.

I think it’s possible that if this AI plugin standard takes off, then you could have multiple engines consuming the same standard. You could today, and in fact LangChain already has a way of just consuming AI plug-in spec and integrating it into any workflow you like, and you can run that on models hosted entirely on Colab notebooks or on your own laptop or anything like that. It’s a completely open spec, and so every single AI plugin that’s published is accessible not just to ChatGPT, and not just to Microsoft, but to a web GPU-based LLM hosted on your own WordPress website, if you want, that you could host the actual model, fine-tuned for your own use cases, on your own website and plug in whatever plugins you want, so just wanted to mention that.

Dave: It’s crazy, isn’t it? I mean, I’m just trying to think of all of the things in my life that are on my to-do list, they could just get done without me needing to do it, or at least 90% done.

Dan: Yeah, exactly. It’s ultimately what it breaks down to is what are the things that are getting done that weren’t getting done before? If we want to talk about economic bread and butter sort of issues, it’s like in my own life, if I had the perfect assistant, it would be able to listen to my inbox, and I would never miss that important message from school that was buried at the bottom of the newsletter, like, “Bring your bathers tomorrow because it’s a pool day.” Because I don’t know about you… you might have kids in… but I drown in incoming information from the kids’ school about volunteering opportunities and days off and half days and this and that.

Dave: School chat and school tech, it’s overwhelming for sure.

Dan: Yeah. I just want my assistant to let me know, to be smart enough to let me know on the morning that they have to wear some different clothes to school, that they have to wear some different clothes to school, just the salient facts delivered at the right time. To me, that’s the promise of a bot based world, an assistant based world, is if an assistant makes that promise and keeps it, then that’s the assistant for me and I will shovel $50, $100 a month at them.

Dave: Well, I think on an individual basis that is, yeah, I mean, I would pay that money too, for sure, even just not to have to read school news out is I would probably pay that. But I think where this potentially becomes more… not more, but another way of thinking about this is what does that mean if you are running a store on WooCommerce and you are getting a bot that is acting on your behalf. It’s like your e-commerce expert and it’s saying, “Given last month’s open rates and the new products you’ve uploaded, and I’d recommend this discount code and this language, this creative. Are you happy with this month’s marketing plan?” Click, go.

How much more value does that generate for use of WooCommerce? Then if you believe, as I do, that people that doing trading, selling to each other binds us together as people and strengthens our civilization even during these sort of fractious times of change, is a net positive for the world. How much good can we do by helping merchants to sell the right stuff to the right people at the right time in the right way, even if you just move that needle 1%?

Dan: I think for me what’s really exciting, and it always starts with the alignment of values and incentives, is that Woo is all about supporting small business growing into big business. It’s all about people at any size of business being able to just start and not have the lack of a website get in their way or not have Amazon’s terms and conditions get in their way or whatever. Nobody can stand in your way, but there is complexity involved in every aspect of that from, “Okay, what if I need to customize my website in a way that’s not yet supported by the plugin or my theme, or what are all the different things I need to learn just so I can do the two or three things that really matter to me? Oh, I’ve got to learn about the theming system and this and that.” There’s all sorts of places where AI can help, all the way from helping the coders who make these plugins in the first place code more quickly because they’ve got AI assisted coding tools, to AI assisted obviously generative content.

The Jetpack AI plugin is a great example of this. You can generate content and then you can reformat it to be longer and shorter and more formal and more whimsical and more whatever. I think it’s going to start feeling we just have more ways of manipulating the content we’ve already created. Maybe you’ve already created a bunch of products and you want to go in there and just make the descriptions punchier. Well, Jetpack AI can already help you with that. You can go in and just punch something up and there’s going to be more and more of that sort of thing all the way to… We have a really, really on WordPress.com powerful advertising system called WordPress.com Blaze, and it’s actually the twin of Tumblr Blaze, and so it’s a place where you can go and create ads based on your own content to promote your own products and services.

There’s obviously already places where we use AI sort of in the ad server for targeting and that sort of thing, and that can always be improved. But also generating ads based on content is obviously a great idea. People do that using ChatGPT today, but there’s just ways that we can make it easier for merchants. All of those things are happening all over the place, not just at Automattic. I think we’re all engaged in this giant depth-first search of what is the right way to expose the power of these tools without leaving users subject to weird hallucinations or having to proofread everything or resulting in content that reads way too boilerplate. Like, “Ugh, this is just another AI generated website. It feels depersonalized.”

It needs to still have the impact that people are looking for, and that I think is where, as we go through various generations of AI products, the companies that measure and improve on the fundamental metrics that define these workflows will be the ones that live to get to the second and third and fourth generation of these tools where the business impact is exponentially higher than what we have right now.

Dave: I know we’ve touched on probably a couple of aspects of this, but something that I always try and come back to and anchor myself with is our mission, Automattic’s, but also WordPress’, to give people the freedom to publish and the freedom to sell online. I’d like to hear your thoughts about how these technologies, AI changes or doesn’t change those freedoms. Does it give freedom a new context, and if so, what is that and what does that mean for the ecosystem and everyone that relies on it for making a living?

Dan: Yeah, actually that’s a good prompt because it brings me back to a point I was trying to make earlier and then completely forgot, which was that I’ve noticed in the age of AI, quite a few people starting up small businesses that wouldn’t have started them up before because AI is able to power some aspect of the business that they found painful, regardless of whether it’s an AI business. But this is somebody who’s like, “Wow, I always wanted to run a yoga retreat, but I was never able to write all the marketing spin and now there’s this AI tool that does it for me and then I can just run my yoga retreat,” and things like that.

I think where there’s values alignment is in this idea that, well, where there’s freedom I should say, which is what you first referred to, it’s in, “I don’t need to ask someone for permission, I don’t need to learn anything new. I don’t need to hire a bunch of people. I have everything I need today. If I take on this one tool, say Woo, then I have everything I need today to get started on my dreams.” To me, that’s what the unlock is.

Dave: Yeah, I think about it in a couple of different ways as well. I think these just feel sort of a bit orthogonal to that point. One is that it’s about the freedom to create, to express yourself. You mentioned not being blocked because you don’t have the skills or the money or the time or the expertise or whatever. That’s about giving people tools that let them connect with their creativity in more profound and meaningful ways. I think that’s probably building really on what you said, but it’s that freedom to just go and take what’s in your head and manifest it in the world. I think that’s just extremely powerful. The other one is then the freedom to make a living from that and to do that at the edges. We’ve kind of talked about this through the episode as well, which is that you can just run the software, do what you want, sell what you want in any way that you want, to anyone at any time, anywhere.

There’s something very powerful about that. This is absolutely not a recommendation to not be compliant with all local laws and regulations, but innovation happens at the edges. I think one continuous advantage that open source will always have over centralized and proprietary is the permissionlessness of it that, sure, you might need to ask for forgiveness, but you never need to ask for permission, because you can take this stuff and you can just start doing what you want to do. Whatever it is that you want to manifest in the world, you can go and do that. Some of those ideas become powerful and they get traction, they kind of get dragged back into the middle and they shift the whole space. I guess we’re talking about the freedom to innovate and create at an individual level, but we’re also talking about the freedom to innovate and create at a more society or civilizational level.

Wrapped up in that for me is the monetization side of things because the freedoms are only important as long as they actually touch creators. I think more and more people are relying on being online to make a living, and if you can’t afford to be creative or to follow your passion because you don’t have the capital or you’re working three jobs or whatever, then the freedom to create and express yourself, all well and good, but irrelevant. So whenever I think about what we’re doing at where I’m trying to think, “Does this help people make a living online.” Not money, but a living, to be feeling like they’ll be able to put food on their table in six months’ time and that’s part of what freedom to me means in this new paradigm.

Dan: Yeah, well, I think WordPress has always been about pure freedom of expression, I think, in some fundamental way. Then Woo is about the subset of pure freedom of expression where people are making something you can sell, and that’s for many people, not all people. Some people have a boring job and an exciting time on the weekends and some people want to have an exciting job that they do during the day and also on the weekends and they’re entrepreneurs and that’s totally fine for them to be different subsets of people. Each of them may be WooCommerce customers with different motivations. But I do think some of the joy and freedom aspects, in fact many or most of them from the core WordPress project, are there in Woo, to a fault with all the defaults and options. People are able to opt out of almost every form of externalized tracking or control or auto-updating or whatever. If you want to control the individual bytes on disc, you can, and I think that that’s important.

Yeah, and I also think where there are people who would benefit massively from somebody making that service a bit more managed where you can describe a problem and it can create a solution for you, rather than you having to go look for an expert who can manipulate the bytes on disk, then that should be part of the picture as well. It’s just part of that classic process of how do you scale the benefits of something. I think there was a time with WordPress where the fact that it was open source was the main benefit. It wasn’t more sophisticated, it wasn’t more powerful. It was in some ways user-friendly, but that’s just because it had so few features at the time. But the fact that it was open source and you could download it as a zip file and run it on your own server was like everything.

That’s no longer really the case because now we’re not just trying to reach the early internet adopters who all ran their own servers in the early 2000s or whatever. We’re trying to reach people who don’t care about computers, and that’s just a matter of scale. It’s just a matter of, “Well, if your goal is to democratize this stuff, you’ve just got to scale out in all these ways that may be powered by technology, but are fundamentally more about making technology disappear than putting every bell and whistle in someone’s hands directly, and they can sit at the Boeing 747 of content management systems and control control the lot.” That’s not going to help most people.

Dave: Yeah. I guess sort of stepping outside my role, that’s something that I feel WordPress has sort of been a bit complacent with for a while. Its open source isn’t really… It’s still an answer, it’s just that not many people are asking that question anymore.

Dan: It’s necessary but not sufficient.

Dave: I think that’s true. But there’s a deeper story which is very much tied to open source and does still very much matter. I spent a non-trivial amount of time thinking about what that is, that story, because I think it’s during times of disruption like this where we’ve got new technologies and everyone’s asking questions. I think being able to fall back on principles and missions and values, that’s when those things really matter. When everything’s going… I guess they can seem like difficult distractions to actually get clear, but when everything gets crunchy, being able to fall back on those just sort of key layers, mentally, those key bits of scaffolding, they’ve certainly helped me think about what does this mean, all this AI stuff? What does this crypto stuff mean as well?

Dan: Yeah, it’s absolutely the right layer to go to. It’s like what’s our values? What’s our history? What are the things we have traditionally succeeded at? Because then when you revisit what’s happening in here and now, say technological disruption from AI, it builds your confidence because you’re remembering your strengths. Say, for example, at Automattic we have a number of huge strengths around community building and our involvement in open source and the millions and millions of free blogs that we have and just the sheer infrastructure itself and scale. It’s mind-blowing and growing.

Dave: And the people.

Dan: Right, you can’t turn around and necessarily build something that Apple might build, as much as you want to, because you’ve got the people you’ve got who have a totally different set of tastes and ideas and ones that are actually far better suited for our mission than trying to squeeze someone else’s roadmap into our culture. There’s a version of this that is very Automattic-y and very WordPress-y and very WooCommerce-y, and that’s the exciting stuff to focus on. The nice thing about it is the possibilities are so open that there’s no harm at all in bringing your whole self to the question of “How should I AI?”

There’s nothing to lose, and you may end up creating a moat because other people aren’t able to see the world the way you can see it. Yeah, so I’m really excited about the possibilities and I think open source in general, and WordPress and Automattic and Woo in particular, stand on so much history of open content and community and positive interactions and long tail content that represents every corner of humanity that in the age of AI, where those things are more important than ever, I’m just super excited about what we can do over the next few years.

Dave: Yeah. I guess this is slightly apropos of nothing but I’m going to share it anyway, just as we start closing out today. I guess questing for my personal mission for a long time and something that kind of clicked into place recently, I’m not saying this is the whole thing, but watched a talk at Eat Belgrade, actually one that a friend of mine was giving, and he was talking about win-win games, non zero-sum games. Those are obviously the games that we want to be playing as a society, games where everyone can win rather than some people win at the expense of the others, which seems like a lot of the world at the moment. I mean, haven’t done my research, but his assertion was that there are two types of win-win game. You’ve got innovation, we’ve got cooperation, and those are both things that I’ve always valued deeply.

I work in an innovation role here and I work at a company that has fundamentally deep roots in cooperation, as embodied by the open source ecosystem. I think that’s the story that I’m going to be dwelling on and thinking about when it comes to trying to manifest Automattic’s mission is how do we maintain and strengthen and build upon that cooperation? How do we make that more meaningful whilst also leaning into the innovation? What do these new technologies open up that we can create with on behalf of, in service of our community and our customers so we can leverage those two things together, and let’s just see, let’s do more winning in the world. Dan. Let’s see more winning.

Dan: Definitely. I mean, it’s a complicated question because you have to have winners and losers in capitalism, otherwise you have a bunch of monopolies and a bunch of monopolies get stale and fold the pieces. To a degree, you want ideas to win or lose, but regardless of which idea wins, you want society to be improved. That doesn’t mean that somebody’s startup doesn’t deserve to go underwater if they guessed wrong about the future of user interfaces or something. That person might lose, but then they’ll be living in a society where a better idea than theirs won, and therefore the capabilities that they have as an individual are improved regardless of the fact that their startup went underwater. We have to allow for… But this is where for a company the size of Automattic in an industry where it’s not clear yet where things are going to land, it’s really important to disrupt yourself in various ways.

Also, and this is just advice for anyone, it’s like discard ideas freely, adopt ideas freely and maintain the most heightened level of awareness possible of where something is showing unusual traction. That requires having a pretty careful data and analytics layer that tells you the right things. It’s possible to build a support widget that deflects support 50% better than another support widget simply by hiding buttons deeper, making fonts smaller, having arbitrary [inaudible 00:42:52] or whatever you want to do. At the end of the day, you really have to be careful about measuring the right things or you might trick yourself into thinking you’ve done something positive, when you’ve harmed your customer.

Dave: Nice. All right, well, look, we need to wrap up, but I’ve really enjoyed today’s discussion with you, Dan. I think we’ve largely stayed clear of spilling any secrets or saying anything controversial. I appreciate that we’ve had quite a wide-ranging conversation, but I intended that to be the case because I know that Automattic can be a little bit of a blackout curtain. It’s very difficult for people outside the business to gauge what’s going on inside and obviously it matters to everyone in their ecosystem what is going on inside. I wanted to, I guess, have a slightly more philosophical and values driven and wider-ranging conversation because the mission remains, the values remain. The things to do remain the things to do, and we’re going to need the whole ecosystem working together to manifest those things.

Dan: Absolutely. Absolutely. Focusing on your strengths, staying positive, building new stuff every day. We need everyone to be doing that. It’s a crazy time. We don’t know who in the WordPress ecosystem is going to break out and become a global success on the back of some of these technologies. But what we do share in common is what you just said, is I think a lot of the values, the community and the willingness to help each other. Yeah, hopefully by Automattic sharing what we’ve learned on this journey, then others can learn from that too. Like you say, it’s the win-win. We can help each other be better and we can also help the freedoms that we care about stay relevant in the age of AI, which I think is critically important. I would feel very sad if those disappeared from the internet and from the world.

Dave: Well, let’s make sure they don’t, and in the meantime, let’s enjoy the human connections that our common software gives us, that keeps us all connected. Let’s keep an eye out for more win-win opportunities all around.

Dan: Absolutely. Hear, hear. Thank you for the time.

Dave: Oh, thank you.