Podcast: Play in new window | Download
Subscribe: Apple Podcasts | Spotify | Android | TuneIn | RSS
Jason BarnardLike many digital practices, search engine optimization is becoming more conversational.
Not long ago, SEOs had to make their best educated guesses about what was working to get their websites to rank better. Now, by focusing on both feeding information to and gleaning feedback from Google’s knowledge graph, Jason Barnard helps companies craft content strategies and messaging architectures that keep their brand prominent in Google’s search results.
We talked about:
- his diverse background as an economist, musician, cartoon dog, and brand-SERPs expert
- how he got interested in Google’s knowledge graph
- how Google can identify the author of an article, even without a byline
- how your content helps Google understand what your business is about
- how he uses Google’s understanding of a business to plan content that can clarify that understanding
- his observation that “SEO is just packaging content you should be creating anyway, packaging it for Google”
- the importance of well-structured, consistent content
- the challenges of aligning human communication quirks with Google’s machine-precise evaluation of web content
- the reliance of Google’s new Search Generative Experience on their knowledge graph
Jason’s bio
Jason Barnard is the CEO of Kalicube. Jason is also an entrepreneur, author and digital marketer who specialises in Brand SERP optimisation and Knowledge Panel management. Jason uses the pseudonym “The Brand SERP Guy” for his professional work.
Connect with Jason online
Video
Here’s the video version of our conversation:
Podcast intro transcript
This is the Content Strategy Insights podcast, episode number 172. Practitioners of search engine optimization are famous for their fealty to Google. Nowadays, though, what used to be a guessing game with SEO’s trying to divine what Google wants to know about a website can now be more of a conversation between a brand’s messaging and content teams and Google’s knowledge graph and search engine. Jason Barnard knows more than anyone about the content strategies and messaging architectures behind these advanced search-marketing practices.
Interview transcript
Larry:
Hi everyone. Welcome to episode number 172 of the Content Strategy Insights podcast. I am really happy today to welcome to the show Jason Barnard. Jason is the founder and CEO at Kalicube. But Jason, Google knows you by some other names as well. Can you tell us a little… Well, first of all, welcome and tell the folks what Google thinks you are.
Jason:
Yeah, Google’s had a lot of different opinions about me and thank you for, A, inviting me and, B, asking me that incredibly delightful question to start. My career in the past was a musician. I was a professional musician for years. Before that, I had an economics degree with statistical analysis and I was going to be an economist, that didn’t happen. And I joined a rock band playing double bass, punk folk music. And then I became a cartoon blue dog called Bua with a hugely successful website and a TV series that was aired around the world. And then I tried to become a digital marketer and I pitched to clients and clients would seem incredibly interested and then they wouldn’t sign, and I couldn’t figure out why they weren’t signing.
Jason:
And then one day, one of my clients who actually became a client said, “Well, we searched your name after you left the office and it said at the top Jason Barnard is a cartoon blue dog. And we think that’s funny, but most people probably wouldn’t want to give their digital marketing strategy to a cartoon blue dog.” At which point I thought Google is a child. It hasn’t understood what I’m trying to project to my audience, I need to educate that child so it understands the cartoon blue dog is now in the past, and that today I’m a digital marketer and I want to be represented primarily as a digital marketer. So there was a whole kind of awakening moment, and that was in 2013. And then two years later, I created Kalicube, built the database, a billion data points of branded searches, which we use to drive our client’s strategy and our own strategy indeed. And one of the really nice things… Oh, sorry, I was going to answer the question finally.
Larry:
Yeah, right.
Jason:
I was building up to the question because one of the things is that knowledge graph, Google’s understanding of the world, this child understands the world based on Wikipedia, Wikidata, IMDB, MusicBrainz and other often entertainment-based databases that are half human curated. So if you put Wikipedia to one side and you take IMDB, MusicBrainz, Rotten Tomatoes, all of these have been used by Google to understand the world, fill up its knowledge graph alongside Wikipedia, Wikidata and Freebase.
Jason:
Now the problem with that is, of course, if you’re a musician, a lot of your information is in those databases so Google thinks, “He’s a musician,” because it’s been told to rely on that information more than it relies on the open web. So making it start thinking, “Well, actually not a musician, he’s something else,” is very difficult. Then you add on top of that, “He’s a musical blue dog with a TV series,” you’re really struggling now to re-educate this child and say, “Well, that might be the information you have. It might be the information you’ve been told to trust. It might be dominant in your trusted source set, but the rest of the web is full of information that describes me as a digital marketer.”
Larry:
That’s so fascinating, that evolution… Not evolution, but the different views of your identity that feeds in there. Hey, but I want to set a couple of contextual things here because some of our guests may not know about Google’s knowledge graph and the importance of that in getting found online and how it helps Google contextualize you. But I also want to make sure that we keep content, because this is a content strategy podcast, but I want to make sure, because I think any content person of any kind, you want your content to be found so I think it’s important to understand this stuff. First, let’s talk a little bit about Google’s knowledge graph, because you said 2013, that’s only a year or two after the knowledge graph was introduced. Tell me a little bit about your discovery of the knowledge graph and how you’ve come to understand it so well.
Jason:
Yeah, well, I mean it’s lovely that you mentioned that it was a lucky break that at exactly the time I had this problem, Google was creating a knowledge graph, so I immediately had to look into that. So I started SEO in 1998, so I’ve grown up with Google. Then it started its knowledge graph at exactly the time I was trying to reeducate it about me and present me in the way I wanted so I’ve grown up with the knowledge graph and seen it evolve. So obviously Google and the knowledge graph don’t know that I’ve been evolving and growing up with them, but I’ve been growing up alongside them for the entire history. And that’s given me a really nice insight because I’ve seen all of the different stages, I’ve seen the different data sets being introduced, and I’ve seen now the engineers at Google letting this knowledge graph child that’s trying to learn the world, they’ve let it loose and it’s now looking at the World Wide Web, the unstructured data and trying to make sense of it.
Jason:
And one really interesting thing, and we’ll come to writers in a moment, but recently it’s been trying to guess at who has written an article, and that is huge. It had a two-day window, and it’s in the KaiCube Pro database where we had a two-day window where it was guessing at the articles I’d written and it got some of them wrong. It attributed some of Barry Schwartz’s articles to me, which is delightful, thank you, Barry, now I look like I know all about the history of Google and search in general. But it was looking at the page, because Barry didn’t put his signature at the bottom, his tagline, my article was the first one underneath the article, and Google assumed that the first author it saw and recognized on the page must be the author of that article, and after two days it disappeared. So Google are doing it, but it’s testing at the moment and it isn’t yet introduced fully, but guessing at who the author is huge in terms of-
Larry:
That’s fascinating. One thing I have to point out here is that’s conceptually the opposite of a large concern of a lot of content strategists, which is this notion of voice and tone and style and governing that and making sure that it’s consistent across all of your writing. And if you’re really good at that, you don’t even have to put a byline on your story, Google can figure out who wrote it.
Jason:
Sure. And Bill Slawski was talking about, is it Author Vectors I think they’re called? And Google can recognize who’s written an article by the style, and it’s obvious with AI, it’s incredibly simple to figure out what the style is. And if an AI can write in the style of Shakespeare, it can figure out whether something’s written by Shakespeare or by Tennyson.
Jason:
So in the same way we all have a footprint or a fingerprint in the way we write, but imagine the amount of resources Google would need to attribute to that to do it at scale. And also imagine that it wants to be confident, and that’s something everybody forgets about Google, is understanding is one thing, confidence is the most important thing. Getting it to understand something as simple as pie, getting it to be confident in that understanding is the real trick. So what it’s done is it’s looked at the article, it hasn’t checked my style against Barry’s style, it’s just looked at the name at the bottom and said, “I recognize Jason Barnard, let’s go with that.” And so what we need is both, we need both. Sorry, I’m getting over excited.
Larry:
No, no, no. But I want to back up just a little bit because you mentioned Bill Slawski, and rest in peace, Bill, he was this awesome guy who read every Google patent the day it was published and would share the discoveries in there. I’m not familiar with that, was there a patent related specifically to that author identity? You said, what did you call it? Author…?
Jason:
Author vectors, I think it’s called.
Larry:
Author vectors? Okay. Yeah, fascinating.
Jason:
That was ages ago. I think that was 10 years ago. And that’s the thing with all these patents, they patent stuff because they’ve invented it, but they’re not using it. And then at some point it gets integrated. We don’t know which ones have been integrated, which ones haven’t, when they get integrated.
Jason:
But what you can do or what we’ve been able to a lot of the time with Kalicube is by tracking over time, we’ve got eight years of data, and by keeping an eye on the brand SERPs in particular, which are the search results for a brand name, which are incredibly insightful into how does Google understand a brand, an entity, a person, a company, a record, a cartoon character, a music group, a music album, a music song, all of these are understood by Google. And the result for the name of each of these entities, these things is hugely insightful about not only what it understands, how it perceives it, how it perceives the world’s opinion of that thing, and also how it’s learning as time goes by. And this is why I get so enthusiastic about it, because it goes so deep so quickly.
Larry:
It does. And we were talking before we went on the air, and I want to explore this right now, you were talking about how a lot of the publishing and content activities that we do are educating Google about us, but they’re also educating us. There’s, sounds like a feedback loop or something, tell me more about that.
Jason:
Yeah, a feedback loop in music is horrible because it’s the music going round and then it goes really, really loud in your ear and it bursts your eardrum. So from a musical perspective, feedback is bad unless you’re a really good hard rock guitarist but that’s a different topic. But the feedback loop is actually really interesting. It’s how we’ve built the Kalicube process, which is how we design and implement digital marketing strategies for companies.
Jason:
And it sounds a bit strange, but we figure out what the perfect digital marketing strategy is, and we implement it with a company by looking at Google. So we will look at the results from Google for that company’s name. We look at the first 10 pages, we analyze it and we figure out what has Google understood? What has it misunderstood? What is it seeing? What isn’t it seeing? What is the company doing right? What’s it doing wrong? What’s missing? Which are the missing pieces? How is it unclear to Google what it is the brand is trying to achieve? And then we can build from that what they’re doing right, what they’re doing wrong, and create a digital marketing strategy for them.
Jason:
At which point we are educating Google because what we do is place the company in the right places at the right time, so they’re standing where the audience is looking with the right piece of content to explain to that audience that they have the solution to the problem and invite them down the funnel. Google then sees this and understands what the company is offering to whom and that they have the right solution, at which point Google then clarifies in its mind, let’s call it a mind, all of that information. And that’s then reflected in the search result for the company when you search for the company name because it improves and it reflects more accurately the brand narrative and the world’s opinion of that brand.
Jason:
And then you start again and you look at what else is wrong, and then you keep going and keep going. And as you said, it’s a feedback loop that’s never ending and it’s brilliant. And it’s not just about brand, it’s about brand, positioning, marketing, standing where your audience is looking, showing them the right solution with the content, making sure the content resonates and works for bringing people down the funnel because ultimately that’s what Google’s looking for is that you provide the solution to the user.
Larry:
Yeah, I remember when Google first started talking about that about 15 years ago, I took it to heart and I became a UX practitioner and kind of abandoned SEO. But the way you’re describing that, and Google has always professed to if you’re doing right by the user, you’re going to show up better here. And what you’re saying is that by properly educating Google about it… But tell me more about the other side of that? Well, you did tell me about how you can infer from those first 10 pages of search results, like, “Oh, this is what they understand.” So how long are your engagements with clients? How long does it take to say, “Okay, I see what’s going on here, let’s say this stuff for a while and see how it changes,” what does that process look like?
Jason:
I’m going to go back a step now because I’d like to make a point because you’ve just made me realize something. Google have said for years, focus on the user, create for the user, and we’ll figure it out. And it’s this really vacuous comment that doesn’t mean anything. And it’s totally useless advice, not because it’s bad advice, but because it doesn’t have anything pragmatic in it that you can actually implement or do.
Jason:
And what we’re doing at Kalicube is looking at the brand SERP because Google’s brand SERP for your company tells us exactly what your user is looking for. We also identify within your industry. So we’ll take 70 competitors or 50 competitors or 40 or however many competitors we can find, and we analyze them and we can template the entire industry, at which point we know what Google thinks is useful and helpful to the user so we can create digital marketing strategy that corresponds to what Google is saying that’s helpful to the user. And with Kalicube, we’re making it pragmatic. It’s not that vacuous phrase because we can say, “Well, here’s what’s useful, here’s what’s useful, here’s what’s helpful, here’s what’s valuable, let’s do it.”
Jason:
And to the question, 24 months. So it takes about two years for us to implement the whole lot, but to actually identify the perfect digital marketing strategy takes less than a month. So in the first day, I could probably tell you most of what you need to know, but then we need to talk to the client to understand what are your business goals? What are you currently doing? What are your resources? Because if you don’t have anybody who could podcast, there’s no point in us telling you to do a podcast. So if you don’t have that resource, we’re just not going to talk about it, even if it’s something helpful for you or valuable for your company.
Jason:
So from that perspective, we can give you the strategy 24-month Gantt chart strategy within the first month. And then either you stick with us and we implement it over two years, we train you, we give the SOPs for each and every mini strategy, LinkedIn, podcasting, videos, PR, FAQs, schema markup, whatever it is that you need to do for your digital marketing strategy. And we’ll give you the SOPs that we’ve already used ourselves on Kalicube over 24 months. After 24 months, we’ve taught you to fish.
Larry:
And then you do your own fishing after that?
Jason:
Yeah. And the brand SERP, once it’s stable doesn’t change very much. So after 24 months, you’re going to see an incredibly stable brand SERP that represents your brand narrative, you know that you’re on the right track with the digital marketing, your team has understood how to implement all of these mini strategies within the strategy. And they all focus on branding, marketing with SEO baked in. So branding first. When you’ve got a brand, you’ve got something to market. Create your marketing materials for your audience where they’re standing, where they’re looking and package it for SEO. SEO is just packaging content you should be creating anyway, packaging it for Google.
Larry:
Yeah, I’ve always thought about, it’s the meta information about, “Hey, here’s what we got,” and to help people find it. Yeah, you mentioned just a minute ago you were talking about the implementation of that strategy a lot of communication channels, and I’m wondering is that kind of up to your clients? I do a lot of omnichannel content stuff and we’re always concerned with all that content strategy stuff that goes into having nicely reusable structured content that you can do a lot of different things with.
Larry:
Well, I guess let me ask you about that. How important is that? We in the structured content, content modeling world, think it’s super important, and then one of the things we’re structuring it for is so that we can create that structured data that goes out along with web content. What does the content need to look like to fulfill this brand SERPs promise you’re doing?
Jason:
Yeah, I like that question because there are multiple things in there. Number one is it needs to be structured because structure will help build confidence. Google can understand if it’s not structured, but it won’t be confident. Confidence is key.
Jason:
The next thing along is your brand narrative, your brand voice. You’re creating masses of content, maintaining that brand voice is hugely difficult. So the advantage of what we do at Kalicube is we’re constantly tracking what Google’s saying about it, and we track all of those results. So any anomalies immediately leap out to us. For example, you create a piece of content you think that’s going to perform incredibly well and we can’t see it, it means that there’s something wrong with it, it doesn’t fit with the package that you’ve been presenting to Google up until now. So we can identify those mistakes and those holes.
Jason:
And we are having a lot of fun as well with we are building KaliBot based on GPT4 Turbo, and Damry, who’s the guy who’s doing the machine learning at Kalicube, has created a scoring system whereby every piece of content we put into the bot, it gets scored on a score from one to five of whether or not it respects the Kalicube voice. And I tried it out with a piece of content I wrote 15 years ago for Bua Koala for the children, and it got a score of one out five. And then I just wrote another piece of content that I would naturally write today, it got five out of five. It’s really cool.
Larry:
Interesting. And that’s a model that you created yourself there at Kalicube?
Jason:
Yeah, I mean, I didn’t personally create it, but we’ve got a guy called Leo who worked on Siri, who’s coaching Damry, and Damry is absolutely going to town and having a delightful time with machine learning and ChatGPT, and it’s lovely to watch and we’re building some really fun stuff.
Larry:
Yeah, that’s why my ears perked up around that because I actually have a second podcast now about content and AI, and I think that’s going to be of interest to a lot of people, just that whole ecosystem of like, oh, “Hey, we want to understand, we want to do this, and an LLM can help us with that.” And then boom, you’re off to just go to a Hugging Face work some LLM and do your thing and there you go. Yeah, very cool.
Larry:
Hey, I wanted to go back a little bit again. So much of this, it’s occurring to me as we talk this so much of the results of brand marketing efforts or marketing or SEO is you’re kind of inferring that the stuff that the content strategists are tasked with is working in the sense of if you achieve that consistency that gives Google trust and faith in what you’re doing and then you show up in the brand SERPs because of that, how far upstream do you go with your clients on working around that governance of voice and tone and style and just making sure, again, another feedback loop, I’m going to guess that like, “Oh wait, you’re falling short here. You need to educate these folks to do…” Is that what’s going on?
Jason:
Now here what you’ve done is described what we could do if our clients could implement it. So fact is with all of the data, we’ve got a billion data points, with a simple query, we can pull out all of this information and say, “That specific piece of content needs something doing to it.” But right now, the clients are just struggling to implement a LinkedIn strategy that makes sense and is consistent with the Facebook strategy, which is consistent with what they’re writing on their website. And that’s stunning is what’s being put on the website is not consistent with all the stuff, especially historical stuff. So the first thing is clean up your digital ecosystem.
Jason:
In fact, no, I’ll come back a step. The first step is define what you’re trying to say. The second step is clean it all up so you’re saying what you wanted to say. And then the third step is to look, is Google saying back to you what you intended to say right at the start? As soon as it starts not saying back to you what you intended, you can start digging in and figuring out the details. It’s hugely exciting. We have one client who’s now asking for this so it’s something I can do on a case-by-case basis. We don’t have a machine that can just do it automatically, but as soon as we get two or three clients asking for it, we’ll build the machine. The data’s there.
Larry:
Yeah. And again, we were talking before we went on the air about basically when it comes to content analytics, it’s like basically most of us just have Google Analytics or some other page-level analytics. And this is another way, it’s not as precise as that in terms of what you’re looking at, but it seems like a really important feedback mechanism. And you’ve talked about how to feed that back upstream. That just seems… Yeah, I think a lot of our content folks are-
Jason:
One really, really interesting point now that kind of occurs to me is if you have three or four pieces of content that have a similar purpose, similar, not the same obviously you don’t want to create all five or six different things that do the same thing, they won’t all appear in your brand SERP because the brand SERP isn’t there to show all of your content, it’s there to show a balanced view of your content.
Jason:
So in order to do what you are describing, we would then need to expand out into a tracking of a wider set of keywords that are specifically aimed at, and then compare that to the branded results and bring all that in, which would be hugely interesting. But that’s a step away right now. And I would love to do it. And as I say, looking now at the data, I can pull some of this stuff out, we can do some of it, but I don’t actually now, thinking about it, looking as deeply as you’ve just looked, I don’t think we could promise to do it today, probably not next year either, but certainly it’s on the horizon.
Larry:
Yeah, as you’re talking now, I’m kind of reflecting on your evolution from SEO to branded stuff and now focusing on the highest level of digital marketing strategy. There’s kind of different concerns at each step along the way. And I’m just wondering if there’s a microcosm of the evolution of your career and how this stuff, for example, and the thing that triggered my thinking about this was when you mentioned, “Oh, why would you do the same piece of content?” There’s so many people who in their pursuit of the long tail end up with 10 really similar looking pieces of content and, I mean, apart from the duplicate content penalties and things like that, it can also run the risk of maybe confusing Google, yeah.
Jason:
Well, we’ve had that problem, which Leanne has now sorted out, Leanne and Jean-Marie have sorted out. We had so many articles about knowledge panels, it was different aspects of the knowledge panel that make sense to people because we are presenting different things in different ways to different people. But Google just completely freaked out and it couldn’t figure out which was the primary article so Leanne came in. I don’t actually know in detail what she did, which is the sign that I’m becoming a better boss is I can say she did it and it’s brilliant and it’s working, but I don’t know what she did.
Larry:
Got it. And it’s not hard to infer just from an editorial perspective, the kind of tidying that you could do around something like that. Yeah, and again, thinking about that, especially that transition from the old content-marketing oriented SEO to the more brand oriented, like you just said, it’s more about having content that helps Google really understand you. Well, I guess, and this is one place where I think we can maybe take Google to task a little bit on its integrity, because the way you just described it, it’s like people totally got that there was differences between those articles, but Google didn’t. How close is Google to really evaluating this stuff like a human being, I guess?
Jason:
Well, I think, I don’t know, but I think Google’s algorithms are capable of identifying the very subtle differences between pieces of content. But the problem is people don’t write clearly enough. So for me, it was obvious, but for a machine it couldn’t understand because I’m inferring a lot of things from the visual aspect. I’m inferring a lot of things from the way the phrase is written, and the machine doesn’t necessarily understand. I may be using colloquialisms. So the machine is trying to understand something that is very human, which is perception of an overall piece of content that the machine can’t have. Poetry, humor, colloquialisms, culture, things that the machine doesn’t really get. So I think it’s hugely interesting that Google’s getting closer and closer and closer, but the confidence remains the key because it can potentially see the difference, but it isn’t sufficiently confident in understanding that difference.
Jason:
So you need to reassure it. Schema markup is great for that. How you write your scheme markup, you have to ask Jarno van Driel, the guy who can do it, and I’m sure he could figure out a way to describe these subtle differences in schema markup. And that’s potentially a really fun project to do with Jarno, which I’d love to do. But I was talking about understanding the world, I really want to say this because I don’t want to forget it. We’ve been saying Google understands the world with its knowledge graph. The knowledge graph is a huge encyclopedia, machine-readable, vastly bigger than Wikipedia, hundreds of billions of facts in there. I actually calculated the size a few weeks ago when… Long story, anyway, it doesn’t really matter, but huge, billions and billions of facts.
Jason:
And we talk about understanding, and then some people in the industry will say, “Well, actually, it’s not understanding, it’s vectorial representation in a mathematical multi-factorial,” yeah, yeah, yeah, great. Technically, you’re probably correct, but it isn’t actually helpful to anybody who’s doing marketing or business. It’s only helpful if you’re really geeking out, at which point you can go and sit and geek out with your friend.
Jason:
And I was really surprised. I was talking to somebody who’s doing a PhD at Oxford in robotics, and she said, “We’ve almost understood the world. Our robots have almost understood the world.” And I said, “What do you mean by almost? Do you mean there’s a semi understanding, but it’s not real understanding because it’s acting mathematical representations in multi-dimensional vectorial database?” She said, “No,” she deals with the visual aspect, and she’s saying we can’t get the robots to understand what’s happening in a room over time. And I asked her, “If it’s sitting still and it doesn’t move its eyes, does it understand its environment?” And she said, “To all intents and purposes, yes. And we talk about understanding at that stage. As soon as it gets up and starts walking around, we have a problem, Houston, because of the passage of time, and that’s where we’re struggling but we’re almost there.”
Larry:
Again, we’re coming up close to time, so I have to wrap this up. You just prompted two giant rabbit holes to go down, the notion of time and how difficult it is to represent, especially in a knowledge graph. But anyhow, hey, we are coming up on time, Jason, but before we wrap, is there anything last, anything you want revisit from the conversation or you just want to make sure we talk about before we wrap up?
Jason:
Yeah, I’d love to bring generative AI into the conversation, but generative AI in search, so Search Generative Experience and Bing Chat. And how that’s now developing, and what I’ve been talking about Search Generative Experience and Bing Chat as either you can look at them as dynamic knowledge panels or micro-feature snippets. So what they’re doing is building an assessment of the search results in the little chunks. So it’s a summary of recommendations and facts into a piece of… Sorry, the Search Generative Experience. And what’s interesting for me now is that Google are now citing the knowledge graph as the source of information. So it’s not just the search results being assessed and reproduced in micro-featured snippet or dynamic knowledge panel format, Google are fact checking. And although right now it’s only dates of birth, dates of founding of the company, it’s a start. So Google is building fact checking into Search Generative Experience as of November, 2023. Being in the knowledge graph is huge if you want to have anything to do with Search Generative Experience.
Larry:
Well, you heard it here first folks. No, that’s super important. I get that. It’s also like, I wonder if it’s related, there’s all the notorious tendency of LLMs to hallucinate and the new things like retrieval augmented generation that adds facts from a knowledge graph. And it sounds like Google’s doing that, which is not a surprise given that they invented knowledge graphs.
Jason:
Yeah, well, I mean the hallucination problem is enormous. It’s the idea of large language models just predicting what the next word is and it doesn’t understand what it’s saying, which is a really nice way of saying it, Andrea Volpini from WordLift taught me that one. But then Bing said, “Well, we’re going to take the search results to summarize them, which means we’re going to have a certain basis in fact, assuming our search results are pretty good.” And Google have now taken that another step further and said, “And we’re going to start checking in the knowledge graph.” Huge.
Larry:
Yeah, and it makes perfect sense because I’ve talked to a lot of enterprise people lately about the inadvisability of shipping generative AI live, but if you have the world’s biggest knowledge graph, maybe you can get away with it.
Jason:
Yeah, I mean Google have to, although I mean right now they’ve been forced into this by Bing. And one last point is you get the impression that Google just launched Search Generative Experience completely out of the box in a couple of months. But if you look at knowledge panels over the last two years, there was a primitive version of Search Generative Experience already present in personal knowledge panels in the format of knowledge panel cards. So for the last two years, you’ve already been seeing a first iteration. So Google weren’t ramping up from nothing, they were ramping up from an experiment that was already in place for the last two years.
Larry:
Interesting. Yeah, they’re a good learning organization for sure.
Jason:
To say the least.
Larry:
Exactly. Well, thank you so much, Jason. This was a super fun conversation.
Jason:
Yeah, it was absolutely delightful, Larry. I don’t know where we started, where we ended up, but it was a lovely journey.
Larry:
Well, we’ll have to pick it up again soon.
Jason:
Brilliant. Thank you so much.
Leave a Reply