About the Guest
Jason Greenwood is the founder of Greenwood Consulting, an independent B2B e-commerce consultancy, and host of The Ecommerce Edge podcast — with over 570 episodes and three releases per week. With 25+ years across brands, agencies, and consulting, he works with manufacturers, distributors, platform vendors, and investors on the B2B digital commerce journey.
In this episode, Jason shares his provocative take: headless and composable commerce will have a bigger, more lasting impact in B2B than they ever did in B2C or D2C — and multimodal AI is exactly why. He walks through how AI agents are turning everyday channels like WhatsApp, email, and Teams into fully functional headless commerce channels, and why B2B suppliers who don’t modernize will soon have a digital gun to their head from agentic buyers on the other side of the transaction.
The conversation covers the data foundation B2B companies actually need before AI can deliver on its promise, why product data normalization as we know it may soon be obsolete, and how the crawl-walk-run approach still applies even in a world moving at 10x the speed of the mobile revolution. Jason also shares how he’s personally living this shift — building AI-powered micro-apps and a RAG-based “Jason AI” consulting tool for a fraction of what it would have cost even two years ago.
Show Highlights
- Why headless and composable commerce will matter more in B2B than they ever did in B2C — and how multimodal AI is the reason why.
- How email, WhatsApp, Slack, and Teams are becoming headless commerce channels through AI middleware that can parse a handwritten order photo or a PDF purchase order.
- The coming shift to agent-to-agent commerce: agentic buyers already exist; B2B sellers need an agentic seller to meet them.
- Why B2B companies can’t skip to AI without first building the data foundation — and what that discovery process actually looks like.
- How LLMs are set to eliminate the need for manual product data normalization within one to three years.
- AI governance as shadow IT: why blanket corporate LLM bans are already failing in the trenches.
- The leapfrog opportunity: why B2Bs implementing commerce technology today can skip straight to AI-first, SaaS-based platforms — and never look back.
- How vibe coding tools like Lovable are letting non-developers build tailor-made micro-apps for a fraction of the traditional cost.
Transcript
00:00.65
Natalija Pavic
Hello, welcome to the show. I’m so excited by you joining us today. don’t you tell me a little bit about who you are and what you do?
00:07.95
Jason Greenwood
Jason Greenwood, been working in e-commerce for 25 plus years. Absolutely love it. Can’t imagine waking up every day and doing anything different to what I do now. I’ve run and built brands. I’ve run and built agencies. And for the last five plus years, I’ve run my own independent B2B focused e-commerce consultancy.
00:25.51
Jason Greenwood
I work with manufacturers, distributors, platform vendors, SIs, and investors, all focused on the B2B e-commerce space. And I also am very fortunate in that I get to do lots of public speaking.
00:37.81
Jason Greenwood
I get to emcee events and and conferences and webinars and all sorts of stuff in our space. And I’m a big, big content producer. have my own podcast with over 570 episodes released called The Ecommerce Edge.
00:48.04
Natalija Pavic
Thank you.
00:50.76
Jason Greenwood
And we release three episodes a week. So I get to do what you do, which is to talk to amazing guests and learn amazing things along the way. So, yeah, I’m just deeply immersed in every aspect of our industry.
01:00.97
Natalija Pavic
When you told me you had 560 episodes on the e-commerce edge, I was like, wow, that’s amazing. What a power, like so much effort and energy. You know, you’re obviously and like, like and you know, anybody listening to this, you’re a a celebrity in the industry. So I’m just like yeah a, a goss, you know, a gosh.
01:20.09
Natalija Pavic
I’m just excited that you’re joining us here today. So, um, but I have going to start, going to let you off easy. So what is an unpopular opinion that you have?
01:30.12
Jason Greenwood
um My unpopular opinion that I’m going to share today is that I think that headless commerce or composable commerce, both of those are actually going to end up being more impactful, more ah more highly deployed, more consequential in the B2B e-commerce space then they and than it ever has been and probably ever will be in the B2C and D2C world.
01:54.14
Jason Greenwood
And we’ll talk about that a little bit more in terms of
01:54.48
Natalija Pavic
Wow.
01:58.43
Jason Greenwood
AI and why i think that AI in particular and its impact on the B2B world is creating this unique opportunity.
02:02.02
Natalija Pavic
Yeah.
02:06.74
Jason Greenwood
And I don’t mean headless storefronts, I mean other headless channels.
02:07.18
Natalija Pavic
Yeah.
02:10.53
Jason Greenwood
And we’ll get into that, obviously, as we go along.
02:12.92
Natalija Pavic
Okay, well, you know, that is crazy, right? That’s crazy because headless and composable are such techie terms. And the question is how could an industry that doesn’t even know what they are all of a sudden realize is the most important thing to them and why?
02:27.07
Natalija Pavic
so those are some questions that I have. um On the other hand, I actually fully agree with you. And I have my reasons, but let’s dive into that right now ah because you’ve opened a can of worms and we didn’t prepare this, which is great.
02:38.93
Jason Greenwood
Yep.
02:39.00
Natalija Pavic
So like, why is Headless and Composable going to be important? And the other question I have is how are we going to get there?
02:45.54
Jason Greenwood
Yeah, I totally feel that question. ah Look, it’s been an interesting journey and an interesting evolution for me, even over the last 12 to months with the impacts of AI and seeing the emerging impacts of of AI on the B2B e-commerce world.
03:03.10
Jason Greenwood
And I think I started to get religion around the headless opportunity for B2B when I started to recognize the power of multimodal AI.
03:14.28
Jason Greenwood
And I’m going to give you an example. So in the beginning, when when I first started using ChatGPT and some of these answer engines, it was all text.
03:18.70
Natalija Pavic
Yeah.
03:20.70
Jason Greenwood
Like if you wanted to translate a bit of text, you’d copy and paste a bit of text in there and you’d say, translate this to Spanish or English or whatever it is. um now i can I can take a photo of a recipe or I can take a photo of a can of something or I can take a photo when I’m out because I live in Mexico now. I can take a photo of something, a sign, ah you know something like that.
03:40.56
Jason Greenwood
I can upload the image to ChatGPT and I can say, translate this to English. Or I’ve used this example on my own podcast before. When I was i buying a dirt bike here in Mexico and I was looking on Facebook Marketplace and I was looking at dirt bikes and oftentimes they would have the the model and sometimes the make, but oftentimes not the year of the motorcycle.
03:58.51
Natalija Pavic
Yeah. Yeah.
03:59.28
Jason Greenwood
So download the images from from Meta ah Facebook, then I’d upload those chat to you and say, tell me what this motorcycle is.
04:00.86
Natalija Pavic
yeah
04:07.25
Jason Greenwood
And it would say, oh, it’s it you know based on these 53 things that I can see in the image. It’s it’s this make, it’s this model, it’s the year this year. And oh and by the way, it’s got an aftermarket seat. It’s got to have to market this and it’s got to have to market that.
04:17.66
Jason Greenwood
Would you like me to you know tell you tell you more kind of thing? And so nowadays i can i can upload a 200 page PDF to ChatGPT and I can ask it to summarize it for me.
04:28.59
Jason Greenwood
I can ask ah for a piece of very specific specific information from the PDF. It’ll tell me what page it’s on. It will summarize it down into two paragraphs for me. it will It will do all that for me it will do it in a matter of seconds. So this the beautiful thing about multimodal AI, and and you can upload any format of of data, any format of document.
04:45.91
Jason Greenwood
It can interpret it. It can parse it. And then it it can tell you information about that that data. Now, what’s super interesting about the mo multimodal aspects of this is that in the B2B world, we have a scenario where e-commerce is still not that common.
04:55.61
Natalija Pavic
Yeah.
05:02.48
Jason Greenwood
ah And where e-commerce is common, oftentimes manufacturers and distributors will consider EDI and punch out their version of e-commerce, not a self-service e-commerce portal. And so what happens in the B2B world?
05:16.53
Jason Greenwood
ah Account management, ah SDRs, BDRs, BDMs, ah customer service, they still play a very significant role in client communication and services.
05:26.61
Natalija Pavic
Yeah.
05:27.08
Jason Greenwood
And what that means is, is that the buyer is oftentimes communicating with an account manager via email, Slack, Teams, WhatsApp, ah text message.
05:37.51
Jason Greenwood
They are communicating with an account manager and they’re oftentimes generating, say, for example, a PDF of a PO on their side, the buyer’s side. They’re emailing it to their account manager or they’re emailing it to customer service. They’re saying, can you place this order for us? And then they manually key it into an yeah ERP system.
05:52.54
Jason Greenwood
And the same happens over WhatsApp and Slack and Teams. They’re asking for quotes. They’re asking for inventory availability. They’re asking for pricing. They’re asking for, um you know, what’s my MOQ? What’s my buying increments? They’re asking all the same questions that they would normally get answered by logging into a self-service e-commerce portal where they would see their own pricing.
06:11.29
Jason Greenwood
They would see their own catalog. They would see, you know, which inventory locations they’re able to to buy or do will call from. And now because of this multimodal AI capability,
06:23.13
Jason Greenwood
We now can put a man in the middle, but it’s not a man. It’s a robot, effectively, a multimodal AI agent smack in the middle of these traditional communications channels, and they now become headless commerce channels.
06:35.09
Jason Greenwood
Because if we have a commerce solution in place, we now have API endpoints that we can farm out these messages to.
06:38.88
Natalija Pavic
Yeah.
06:43.00
Jason Greenwood
We can get meaningful answers from it. We can present them directly to the customer in those channels. So I’ve seen, for example, There are platforms out there. One of them is called Motivate just as an example. There’s many examples out there, but Motivate’s one where, you know, someone can send in an email with a POPDF attached.
07:01.14
Jason Greenwood
This system that sits in the middle, which is effectively a middleware tool, will strip out the the PDF. It will parse it using multimodal AI. It will map it directly into the commerce system as if the customer…
07:12.56
Jason Greenwood
logged in to place that order in the portal. So what’s great about this, and it can also receive images, it can receive anything. So if I’m a, let’s say i’m a builder and I’m on a job site and I realized, geez, okay, I’m short of, ah you know, 20 sheets of drywall, or I need, you know, some two by four, or I need whatever for this building project.
07:28.89
Jason Greenwood
you know I’m not going to probably log onto my phone and log into a portal and and search for something and place an order. I might write it. I might write my order, literally handwrite my order on a piece of drywall or on a building tile, and then I might snap a photo of that. And where historically I would have messaged that to my account manager via WhatsApp and said, hey, hey, mate, you know hey, Bob, hey, Jane, can you can you send me you know these these three things I need for the building site? How fast can you get it to me?
07:52.83
Jason Greenwood
And what’s my price? And then they’d message back, right? Well, now we’ve got these multimodal AI systems that effectively have turned these channels into headless commerce channels. And it is really only the multimodal AI capability that is facilitating this to happen.
08:08.56
Jason Greenwood
Whereas historically, if if suppliers wanted to do anything with data exchange, it was either via EDI or it was via OCR, were they and and it was just dumb OCR. it was not smart enabled OCR.
08:21.83
Jason Greenwood
And oftentimes that data that got extracted out of documents was was inaccurate and needed to be reviewed by a human ah anyway. And so this is a very, very exciting time for B2Bs because when they make an investment into a commerce system,
08:36.46
Jason Greenwood
The less that they need to require their customers to change their behavior, to take advantage of that commerce system, the better, because it automatically means high adoption rates, right? Sure, you can go and sit by the customer and help them log in and help them figure out how to use this new portal and all that sort of stuff, but there’s a percentage of those customers that no matter how good of a job you do trying to onboard them and train them, they will never use a portal.
08:58.93
Jason Greenwood
um But they still want to do business with you at scale, and they and they want to just keep doing business with you. ah They love Bob, they love Jane, their account managers, and they want to continue to have that relationship with their account manager.
09:09.53
Jason Greenwood
But we need to unburden the account managers from all this manual effort of keying in orders, sending across product information to the customer in a zip file, searching a network drive, all that sort of stuff.
09:21.29
Jason Greenwood
And multimodal AI is facilitating every single piece of functionality that I’ve just described.
09:28.70
Natalija Pavic
Okay. Well, I mean, I have a lot of follow up questions. ah i don’t know where to start actually. So what you’re saying is that we should forget the website and leapfrog into AI, primarily AI interactions. Am I hearing that correct?
09:40.72
Jason Greenwood
I think for for some, for some customers, there are some customers that still will prefer to log in.
09:43.35
Natalija Pavic
Yeah.
09:45.36
Jason Greenwood
If they’re sitting at their, they’re sitting at their desk, they’ll prefer to log into a e-commerce website, portalized experience where they can download the previous invoices. They can check the status of their credit account.
09:56.30
Jason Greenwood
They can log a quote request. They can, um you know, they can place an order. They can check their MOQs, their pricing, their catalog availability, all that sort of stuff. But there is a certain percentage of most customer cohorts in the B2B world that will not transact with you in that way.
10:11.48
Jason Greenwood
But we still want to serve them digitally and we still want to automate as many of those interactions as we can to make us more efficient as a supplier and to give better service to our customers, meaning that we we we don’t we’re not forcing them to change their behavior unless they want to.
10:26.59
Jason Greenwood
ah Now, there’s a certain cohort of those customers that will be telling you already I want you to have a portal. I want you to give me an e-commerce website that I can log into and do all these things.
10:36.45
Natalija Pavic
Yeah.
10:36.58
Jason Greenwood
so So I think meeting the customer where they are, not where we expect them to be, is the most important thing we can do.
10:44.66
Natalija Pavic
Right. And to your point, you know, what’s funny is that B2B as an industry hasn’t even jumped onto fully commerce bandwagon. So there’s still a lot of headway and yet the market’s moving on to AI.
10:55.62
Natalija Pavic
And to your point, buyers are going to, they’re, contacting their sellers all the time. They want to talk to somebody. They’re not in the like, I want to go self-service my own return and the in the account management folder, right? so um But isn’t that putting the cart before the horse? Like how how are the B2B organizations going to get there if they don’t have that commerce base? How are they going to take advantage of all the AI that’s out there?
11:19.03
Jason Greenwood
Yeah, and that’s why with my clients, I really encourage them to take a a crawl, walk, run approach. Because oftentimes a ah client will come to me and in the first instance, kind of the first words out of their mouth or the first question out of their mouth is, Jason, Jason, how can we how can we use a on a on our AI in our business?
11:33.34
Jason Greenwood
And it’s like, whoa, whoa, whoa, slow down there, chief.
11:34.00
Natalija Pavic
Yeah.
11:36.14
Jason Greenwood
um you know you don’t you You don’t have a PIM system. You don’t do e-commerce today. You do everything in a green screen ERP. You, you know… You know, you have you’ve undergone no digital transformation in your business for the last 15 years, 20 years.
11:49.98
Jason Greenwood
ah You’ve got no digital enablement. You’ve got no digital capability in terms of people inside your business that even understand digital.
11:55.09
Natalija Pavic
Yeah. yeah
11:57.20
Jason Greenwood
ah You know, you’ve got an aging sales force that, you know, are all 40 plus. Oftentimes they know their product inside and out. They know their industry inside and out, but they don’t know digital inside and out.
12:08.21
Jason Greenwood
And so we have to try to sort of I try not to be a wet blanket in the sense that it excites me that they’re excited about AI. And I’m not trying to quench that thirst, but I’m also trying to say, look, we have to start back at the beginning. We have to build on a solid foundation and all this amazing AI tooling available through all the vendors that you could choose to work with out there in the market.
12:30.51
Jason Greenwood
They all rely certain data building blocks being available in your business before they can do their magic.
12:34.20
Natalija Pavic
Yeah.
12:37.38
Natalija Pavic
Yeah.
12:37.65
Jason Greenwood
There’s certain foundational components that actually already have to be there, and you have none of those in place. So how about we start back at the beginning? We look at your you know your product data. We look at how it’s organized.
12:46.40
Natalija Pavic
Yeah.
12:47.10
Jason Greenwood
We look at how complete it is. We look at all where all the data is housed and ah what format it’s in and is there consistency ah around it.
12:52.56
Natalija Pavic
yeah
12:54.73
Jason Greenwood
ah We look at your customer data. We look how that’s organized. We look at you know the parent account versus the buyers within the account, buyer permissions, credit limits, you know what kind of shipping you provide for them and whether they can you know do will call and come and pick up from your warehouse and You know, my discovery is like you know several hundred questions long when I go in and I’m starting to work with these manufacturers and distributors.
13:13.36
Natalija Pavic
Yeah. Yeah.
13:15.88
Natalija Pavic
yeah
13:16.03
Jason Greenwood
But the the whole point I’m trying to make here is is that ah they’re oftentimes B2Bs are trying to leapfrog all the hard work and jump straight to the shiny new thing, which is ai And they they just it’s really hard for them to grasp why they need to put in the foundational effort first.
13:34.92
Natalija Pavic
Well, yeah, it’s and it’s interesting because there’s so much work required to get to that scalable you know AI service, get the LLMs actually working for you. um But there’s also an opportunity here, right? Because instead of trying to redo what the journey that B2C has been on, B2B has the opportunity to leapfrog B2C and move straight to that AI first kind of world.
13:55.18
Natalija Pavic
They still have to do all the same legwork, it sounds like. And we’ll get into that. um But you know I wanted to ask you, One of the things that’s clear here, to to me, this is obvious, but maybe not to everybody, and and maybe you can tell me more, is headless incomposable and how important that is for an agentic AI-based future. Maybe you could talk a little bit about that.
14:15.35
Jason Greenwood
Yeah, so let’s let’s go back to why this is going to become, and it’s starting to become already a very important capability that the suppliers have.
14:23.10
Natalija Pavic
Yeah.
14:25.93
Natalija Pavic
Yeah.
14:26.30
Jason Greenwood
And that’s because on the buy side, the the buyer is maturing and their technology is maturing.
14:32.40
Natalija Pavic
Yeah.
14:32.67
Jason Greenwood
There’s a lot of buyers out there that use procurement platforms like the Jaggers, the Coopas, the Aribas of the world, that they do all their procurement either in the standalone dedicated procurement platform or they have a procurement module inside of their yeah ERP.
14:48.92
Jason Greenwood
where they’re creating all their POs, they’re creating their buying lists, they store their price lists from their suppliers in there. they They do all this this whole entire procurement function and usually led by a procurement manager on the buy side.
15:02.64
Jason Greenwood
And they want to automate that process as much as humanly possible. Now, the reason why they want to automate this is because if they can claw back 100, 200, 300 basis points of margin, That is substantial for their business because in the B2B world is extremely common to run anywhere from two to 4% net margins.
15:20.44
Jason Greenwood
Like that’s very common. And as a result of that, anything that they can save from a procurement cost and complexity perspective, making them more efficient, that’s that’s direct margin they can put into their back pocket.
15:32.46
Jason Greenwood
Okay? So what what’s going to eventually come, and these these platforms are already talking about agentic procurement, but what that means is that on the buy side, there’s not going to be a human being anymore.
15:45.14
Natalija Pavic
Yeah.
15:46.28
Jason Greenwood
There’s going to be ah a digital agent doing the buying.
15:48.37
Natalija Pavic
Mm-hmm.
15:49.02
Jason Greenwood
which means that they have to have a digital sales agent on the selling side to be able to communicate with.
15:54.07
Natalija Pavic
Yeah.
15:54.73
Jason Greenwood
otherwise
15:54.88
Natalija Pavic
Right.
15:55.53
Jason Greenwood
ah Otherwise, we can’t do this. We can’t create this agentic procurement pipeline if there’s if there’s not the supplier and the the buyer working together to to build that pipeline to communicate with each other electronically.
16:09.32
Jason Greenwood
So I look at agentic buying or agentic procurement or agentic commerce.
16:12.45
Natalija Pavic
Yeah. Yeah. yeah
16:13.92
Jason Greenwood
I look at that as kind of the the Novo version of EDI, right? It’s the it’s the super modern version of EDI. And just like in an EDI mediated world, you’ve got to have you’ve got to speak the language as the seller. You have to speak the language of EDI and you have the technology to receive the EDI buying messages, electronic data interchange message from the buyer.
16:35.13
Jason Greenwood
you’re gonna have to do the exact same thing in an agentic commerce world. Now, the beauty of agentic commerce is that in the same way that I can upload a PDF today to ChatGPT and it can give me meaningful information, those agentic buying technologies, we’re gonna get to a place in the in the very near future
16:51.45
Natalija Pavic
Yeah. Mm-hmm.
16:51.64
Jason Greenwood
Where again, if I’m a, if let’s say I’m a builder and I need to buy some building supplies, I’ll be able to upload my plans, literally the plans of what I’m building to my agentic buying platform.
17:03.37
Jason Greenwood
I’ll be able to parse that. I’ll be able to interpret that. And I’ll be able to send that down the pipeline to my agentic seller. They’ll be able to interpret that and parse that and say, okay, cool. ah You know, the agentic buyer will say, based on these plans, what can you supply?
17:17.54
Jason Greenwood
What is my price and when can you get it to me? Okay. The agentic seller will go, okay, cool. Based on the plans and our review, digital review of the plans, here’s the, the, the, basically the buying list, the procurement list that we need to fulfill.
17:30.41
Jason Greenwood
We can fulfill 80% of the components on this list based on the plans. Here’s your price. Here’s our warehouse availability. And here’s how fast we can get it to you. Or you tell us the date you want us to ship and for it to arrive on your building site. And we can make that happen too.
17:42.96
Jason Greenwood
So this, is this is all being enabled. buy AI. But again, the procurement side of the industry is currently further ahead on this sort of digitalization journey than the sell side.
17:57.75
Natalija Pavic
Yeah.
17:58.94
Jason Greenwood
And so the sell side is is still and has been playing catch up for the last decade. And so this is where there’s this friction where the The suppliers and sellers are going to be forced to modernize very soon. If if if they’re not if they don’t don’t already have a digital gun to their head, so to speak, from the buyers that say, hey, why haven’t you given us a portal yet that we can buy through, a self-service portal?
18:24.60
Jason Greenwood
They are really going to get the heat put on them once agentic buying becomes the norm.
18:29.68
Natalija Pavic
I mean, you’re talking about agent to agent commerce, right?
18:32.46
Jason Greenwood
Yes, 100%.
18:32.95
Natalija Pavic
So you’re talking about buyer as the agent, seller as the agent.
18:33.28
Jason Greenwood
hundred percent
18:36.49
Natalija Pavic
We both know that in order for this to work, APIs have to be really robust from an architecture perspective because they can not be able to pull up all the product information or…
18:47.78
Natalija Pavic
not be able to pull up the entire catalog, have to go category by category. There’s so many limitations that some platforms have. um So that’s going to be key. I mean, yeah out of curiosity, do you think that these agents are going speak English with each other?
19:00.92
Natalija Pavic
They might have their own agentic language, right?
19:01.16
Jason Greenwood
well
19:04.16
Natalija Pavic
Yeah.
19:04.67
Jason Greenwood
Well, the thing is, is that already AI and these large language models, you can, you know, if I’m if i’m a native Spanish speaker, I can speak to ChatGPT in Spanish.
19:11.55
Natalija Pavic
Yeah.
19:12.91
Jason Greenwood
if i’m
19:13.63
Natalija Pavic
Yeah.
19:13.79
Jason Greenwood
If I’m a native French speaker, I can speak to it in French. I can speak to in English. doesn’t really matter.
19:17.40
Natalija Pavic
Yeah.
19:18.25
Jason Greenwood
But it will be it will be primarily human language-based because that’s the beauty of LLMs is they are –
19:23.55
Natalija Pavic
Yeah.
19:23.53
Jason Greenwood
used to human language and and not having to speak some esoteric language like EDI with this custom syntax and these custom rules. um That’s the beauty is that we can use plain language.
19:36.15
Natalija Pavic
Yeah.
19:37.24
Jason Greenwood
But the the other beautiful thing is, is that all of this information gets turned into vector data so that so that
19:37.45
Natalija Pavic
Yes.
19:43.95
Natalija Pavic
yes
19:46.16
Jason Greenwood
Like if I upload some plans, that gets vectorized. So that way, when I’m comparing my vectorized product data using cosine ah linkages, which is how LLMs work, they base it on proximity and prediction of what should come next.
19:49.64
Natalija Pavic
Right.
20:02.55
Jason Greenwood
It’s like, think of it as an ultra smart autocomplete mechanism. that that’s That’s how the search and find from the query to the result get matched together for relevance.
20:13.31
Jason Greenwood
and so being able to turn
20:13.40
Natalija Pavic
Right.
20:15.43
Jason Greenwood
a plan or a PDF or a CSV file or whatever, being able to turn that into a vectorized set of data that then um ah I then can search my product data set, my pricing data set, all this information that I’ve got in my catalog, my digital catalog.
20:19.35
Natalija Pavic
Yeah. No.
20:32.79
Jason Greenwood
I then now am finding vectors that would auto complete what I would typically need to fulfill that order. And so in a nutshell, that sort of leads very nicely onto what I think is going to be a massive change in the way that PEM systems are designed, built and maintained because we’re now in a world where instead of having to feed these LLMs highly structured product data,
20:48.36
Natalija Pavic
no
20:58.11
Jason Greenwood
We now can just dump it any format of data into a just a gigantic data store that can be indexed with vendor ah vector indexing, right?
21:00.45
Natalija Pavic
Right.
21:06.73
Jason Greenwood
we don’t We don’t have to structure it in any human meaningful way because the LLMs can query these massive data stores with tens of thousands of documents and PDFs and CSVs and images and zip files and and everything else and pull meaningful responses from unstructured data.
21:24.65
Jason Greenwood
and and certainly not consistently structured data. And so i think I think all of this capability around ai is going to start affecting even the way in which we structure our data to begin with ah foundationally.
21:39.69
Jason Greenwood
I think that’s going to change significantly over the next one to three years.
21:44.33
Natalija Pavic
Real quick, define a vector for me. Yeah.
21:47.33
Jason Greenwood
Yeah, well, it’s it’s it’s a i mean ah it’s it’s basically a numerical representation of a piece of of data. So we might recognize it as an image, but effectively we’re breaking the image up or the LLMs when they ingest it.
22:03.74
Natalija Pavic
yeah
22:03.81
Jason Greenwood
They are breaking it up into effectively ones and zeros or coordinates. They’re breaking it up into coordinates of information so that – So that instead of just being able to autocomplete words, we now can autocomplete queries based on ah digitalized ah version of that piece of data.
22:15.19
Natalija Pavic
Yeah.
22:23.63
Jason Greenwood
Whatever type of data that is, it could be a JPEG, it could be a CSV, it could be a PDF. So that’s in a nutshell, we’re turning into coordinates coordinates that that almost like a map ah that that allows the the response to be a vectorized response from an LLM so that that matches what we’re looking to get out of the vectorized originating data, the raw data.
22:49.77
Natalija Pavic
Yeah, so you’re you’re getting deep into like how LLMs work and I actually didn’t know that, but that’s fascinating. um i don’t know a lot of things, so it’s not surprising.
22:54.60
Jason Greenwood
Yep.
22:57.11
Natalija Pavic
But um when you think about a vector, like you said, it’s almost like translating data into a unified framework. It’s saying like, well, everything, an image, an audio file, a transcript can all become a vector.
23:06.04
Jason Greenwood
Yes.
23:06.21
Natalija Pavic
right And that vector has multi dimensions, it has values on those dimensions, it has a direction. So mathematically speaking, now you’re playing with something where you take that unstructured data, now everything’s a vector, now you can start to to you know move it around and and put it all together.
23:21.09
Jason Greenwood
correct
23:21.33
Natalija Pavic
What’s interesting to me, i think this gets lost in the shuffle a lot, is you have these LLMs that are able to understand contextually what customers are saying. And, you know, obviously you see the rise of Agentec AI, Kibo. At Kibo, we have our own Agentec Commerce offering.
23:34.62
Natalija Pavic
We have shopper agents, merchandising agents, et cetera. um What’s interesting to me is a lot of people miss the fact that your search engine on the site also has to be pretty good for it to actually work.
23:45.75
Natalija Pavic
Because even though the LLM can understand what you’re saying, it doesn’t mean they can find it.
23:49.86
Jason Greenwood
yes
23:50.72
Natalija Pavic
And I think that gets missed. So it’s funny in the sense that it needs to break it down by vectors. But then if you’re using synonym or keyword search, I like to say, which is the most common search out there, by the way, which is where you put in something, it picks up a few keywords and looks for those keywords in the description.
24:04.91
Natalija Pavic
um It’s like putting lipstick on a pig, quite frankly. um And so I think what is your search engine gets misunderstood. It’s not a question that people are asking when they’re buying a gent of commerce, not understanding they have to bundle those two together.
24:17.70
Natalija Pavic
ah was like a comment. um I don’t have a question. but if you have If you want to respond, that’s fine.
24:23.03
Jason Greenwood
Yeah, well, i think I think what’s super interesting is that instead of having, um you know this this goes back to the fundamental way in which we architect, the fundamental way in which we architect commerce stacks today is, okay, we have a PIM system, for example, product information management system.
24:31.51
Natalija Pavic
Yeah.
24:40.34
Jason Greenwood
We take all of this weird and wonderful data, we normalize it, we standardize it, we structure it. We have ah buckets of both structured and unstructured data. So we have arrays and attributes that could be reused across more than one product, that’s structured data.
24:53.75
Jason Greenwood
And then we have certain yeah product-specific data that’s like ah a long description, which is like might be a paragraph. And that only applies to one product because that’s describing that product in detail. in a narrative detail that is unstructured data because by definition can only be applied to that one product.
25:10.23
Natalija Pavic
Yeah.
25:10.22
Jason Greenwood
Right.
25:10.95
Natalija Pavic
Mm-hmm.
25:10.99
Jason Greenwood
And so ah there’s a whole ton of work, especially in the distribution world where 200,000 items to 2 million items is pretty common in the distribution world.
25:11.11
Natalija Pavic
Mm-hmm.
25:21.28
Jason Greenwood
And to, to, Do that effort to standardize, normalize, map, create structured and unstructured data, ah put it into your PIM system in a meaningful way, create all of your taxonomies in terms of categorization, subcategorization, attributes, filters, facets, ah linkages between cross-sells, upsells, related products, um all that sort of stuff.
25:43.94
Jason Greenwood
That is a massive lift today. And yes, there are some AI tools that can help ingest that data, then normalize it on the way out and map it into something like a PIM. So there are AI tools that assist in this process, but there is usually still a fairly significant amount of human effort required to answer questions like, for example,
26:05.72
Jason Greenwood
Okay, I’ve got for for the for the reference of an inch ah attribute of an item, I might have one product that has I capital I lowercase n dot.
26:17.15
Jason Greenwood
I might have another item that comes from another supplier that is I lowercase lowercase n dot. I might have uppercase I lowercase n no dot. I might have lowercase I lowercase n no dot.
26:28.57
Natalija Pavic
um no
26:30.07
Jason Greenwood
Then I might have for some inch, literally the word inch. Now, when i Now, they all refer to the same attribute, but I don’t want to have that attribute duplicated 10 different times in my PIM system.
26:37.90
Natalija Pavic
Yes.
26:43.20
Natalija Pavic
Yeah.
26:43.41
Jason Greenwood
I want to normalize, standardize, and map that so that every product that has an inch reference uses the same ah attribute. right Attribute value, actually, is what it is.
26:53.17
Natalija Pavic
Right, right.
26:53.69
Jason Greenwood
So it might be ah the attribute might be length, and then underneath that, you might have values like inch, millimeter, meter, yard, you know, all the different different potential length attributes you have of a product.
27:07.60
Natalija Pavic
Yeah. Yeah.
27:07.87
Jason Greenwood
You might have all those, but we need to standardize those where they’re where they’re referring to the same attribute value. And so that normalization, there are AI tools that can help with that, but there usually needs to be a human in the middle because
27:17.80
Natalija Pavic
yeah
27:22.21
Jason Greenwood
At some point, we have to decide which version of that attribute value we’re going to use and standardize on. And so this is where LLMs in the future, i think we’re not going to need to normalize and standardize that because what’s going to happen is an LLM knows that IN.
27:40.25
Jason Greenwood
capital IN dot inch, ah you know, IN with no dot, they know that that all refers to an inch. So we won’t need to standardize that because it knows that if I’m referring to M or meter, it’s referring to the same thing, right?
27:55.44
Natalija Pavic
Right, right.
27:55.82
Jason Greenwood
In relation to length. So I think that’s the really exciting thing for me is that we still have this intermediate step today. We have to standardize and normalize that data before we can inject it into something like a commerce platform.
28:09.48
Natalija Pavic
Yeah.
28:09.61
Jason Greenwood
So the people, when they’re doing a search for a, you know, when they’re doing ah ah a search for a valve, for example, on ah on an industrial supplier’s website, ah you know, we have to know the the inside diameter, the outside diameter, the the flow rate ah capability, all these different things.
28:12.02
Natalija Pavic
Yeah.
28:18.41
Natalija Pavic
Yeah.
28:25.17
Jason Greenwood
We have to know all these different attributes around that product because we’re speccing it into ah specific job and it has to be able to to to be able to do that job. but um So today we still have to standardize this data before we even get it into our commerce systems in the future.
28:39.39
Jason Greenwood
i’m just I’m pretty confident we won’t have to do that anymore.
28:42.68
Natalija Pavic
I mean, you’re kind of blowing my mind um because, you know, you’re really bringing into ah view sort of like this huge data issue that we have. and And to your point, it’s it’s been going on forever. Like, hey, you know, we’ve got this product. Now we have to list it on several different marketplaces.
28:58.17
Natalija Pavic
um And if we’re going to move into this world where dropshipping and marketplaces and third parties, that becomes more and more important than the problem becomes incredible, incredible in complexity.
29:08.35
Natalija Pavic
Right. um I also like how you’re geeking out and like the definition of an inch. You get you very excited about this.
29:12.79
Jason Greenwood
Yeah. Yeah.
29:15.13
Natalija Pavic
But I love that because, um that you know, for a long time, we thought the solution was going to be standardizing on on terms and attributes across industries. And now this has flipped on its head. We don’t have to do that anymore.
29:28.75
Natalija Pavic
AI is here to structure the data for us. Nobody likes to clean up data, right?
29:32.23
Jason Greenwood
Yeah.
29:32.33
Natalija Pavic
So being able to have AI tools to help us do that is tremendous. um So, and and where do you think we are right now? Like, I feel like we’re in this like special moment, like a tipping point almost.
29:43.89
Natalija Pavic
Like have people started to realize or had the insight that ah you’re having right now?
29:50.33
Jason Greenwood
Look, it is definitely getting spoken about, but there is still massive challenges of governance, adoption, approval, investment, roadmapping, enterprise architecture.
29:57.36
Natalija Pavic
Yeah. Yes. yes
30:04.97
Jason Greenwood
There’s still ah this laundry list. I was just just two weeks ago, I was emceeing the B2B Commerce World APAC event down in Melbourne. And there was a number of AI ah keynotes been given from the stage. And there was a couple of firesides that were given from the stage where some businesses were talking about their AI journeys.
30:24.50
Jason Greenwood
And certainly there’s still this massive disconnect between what AI can do and what it can do for us. And this is this is the but this is the big disconnect. And it’s a real challenge because one of the one of the ah roundtables, they were talking about how they even had a really hard time deciding which AI agent they were going to use internally to to make their teams more efficient. So, okay, are we going to use standardized on ChatGPT? Are we going to standardize on CoPilot? Are we going to standardize on Grok? Are we going to standardize on Gemini? What are we going to standardize on? And they standardized ultimately on CoPilot because they were a Microsoft shop and they were using Microsoft technology across for internal internal use across all their computers.
31:06.19
Jason Greenwood
And so they decided to basically ban any other LLMs in the business other than Copilot. Now, that’s now that’s a major issue. And I’ve experienced it with some of my clients where they go, well, you know, some of our teams, they go, well, Copilot isn’t good for what what we do. Like, for example, in IT, ChatGPT is significantly better for us if we’re trying to troubleshoot something or we’re trying to. ah you know, write some, so let’s say we run NetSuite and we’re trying to write a piece of suite script. Well, ChatGPT is going to give us a much better ah response for that than Copilot doesn’t know. And it’s it’s totally useless ah for that purpose inside our business. And so there is this massive battle whereby
31:45.61
Jason Greenwood
AI is just currently, as of today, as of mid-2025, it is turning out to be just another form of shadow IT, unfortunately. And so there are people inside the business that are bringing their own device, and they’re going, okay, well, you won’t let me use ChatGPT on my work laptop, but I got my personal phone right here, and I got a subscription to ChatGPT, so I’m just going to use my ChatGPT.
32:09.93
Jason Greenwood
because I need this to make me more efficient and to, and so that I keep my job.
32:12.16
Natalija Pavic
yeah.
32:13.55
Jason Greenwood
but and so I’m going to use it ah because I, or I’m going to use, um, you know, Gemini on my phone or whatever it is. and then I’m going to email myself the responses or I’m going I’m going to WhatsApp myself the responses. And so this is, this is, this is,
32:25.92
Jason Greenwood
in the trenches. This is literally happening today because these companies are trying to lock down usage ai of AI under very strict traditional IT governance rules.
32:38.07
Natalija Pavic
Yeah.
32:38.30
Jason Greenwood
And that just doesn’t work in in in a modern, fast moving AI innovation environment.
32:45.90
Natalija Pavic
No, and I’ve you’ve even talked to some other people who talk about the incentives are not correct. The incentive structure is not correct. Like we should be incentivizing experimentation, but we’re incentivizing results.
32:54.55
Jason Greenwood
Yes.
32:54.85
Natalija Pavic
And so the problem is people are afraid to experiment because they’re tied to these results. They’re not incented to be innovative, right?
33:01.39
Jason Greenwood
Yes.
33:01.93
Natalija Pavic
And that’s going to be a challenge. Like the company that figures it out gets ahead because ultimately they save on costs by leveraging AI, right?
33:08.11
Jason Greenwood
Yes.
33:08.53
Natalija Pavic
um what so What do you think about this fear that you know AI is a fad or that he’s going to destroy everything? What’s your take there?
33:18.70
Jason Greenwood
Look, I think i think it’s really interesting because. i guess I’m quite a bit older than you, so I remember. the the I remember what society looked like pre-internet.
33:30.29
Jason Greenwood
I remember what society looked like with the coming of the internet, the mainstreaming of the internet, and then what I like to call the mainlining of the internet. And then I also saw that same pattern play out when mobile became the norm.
33:42.43
Jason Greenwood
The normal, ah you know, it became normalized.
33:42.90
Natalija Pavic
Yeah.
33:44.89
Jason Greenwood
And once we went from you know Once we went from BlackBerry Pearls to the iPhone 3GS when it first came out, I remember, and it was like, oh my God, this is so transformational that, okay, nothing will ever be the same again.
34:03.78
Jason Greenwood
I remember when this… when mobile first started to hit. And all of a sudden we had a meaningful um browser in our hands that that was actually good enough to be used as a real browser.
34:14.90
Jason Greenwood
and And then the app store started to get built and we started to have the apps and then it was all about the apps, stupid. And then it started to meaningfully impact our lives. And I remember how fast that happened versus how fast or how slow it was for people to adopt the original internet, web browsers, search engines, et cetera.
34:32.32
Jason Greenwood
And it was like it was reduced by like a factor of 10 in terms of the adoption rate. like it was The adoption curve went through the roof.
34:37.59
Natalija Pavic
Yeah.
34:39.19
Jason Greenwood
What we’re seeing now is that AI is being adopted at 10x the rate mobile was.
34:43.98
Natalija Pavic
Right.
34:44.01
Jason Greenwood
So this is truly turning out to be the fourth industrial revolution.
34:48.99
Natalija Pavic
Yeah.
34:49.00
Jason Greenwood
Now, it’s still very early days. I’m not trying to say that AI is going to replace us all tomorrow by any stretch the imagination. But I tell you, the people that are the most curious… that retain their curiosity and play with these tools now as much as possible, they are going to come out the other side of this so far ahead of the game that you know people that that are are too afraid of it to even play with it. Like I, ah but about nine months ago, I installed a plugin in my Brave browser to where every single Omnibox query ah gets redirected to ChatGPT.
35:20.51
Jason Greenwood
And I realized that, okay, if I try to dabble with these LLMs versus go cold turkey, instead of instead of trying to do like half my searches on Google and then do half my searches via ChatGPT.
35:33.47
Jason Greenwood
No, if I really want to learn how these things work, if I really want to figure out how prompt engineering works, if I really want to figure it out, I need to go cold turkey, no longer do but you know search engine-based searches, and I need to do LLM-based searches.
35:46.88
Jason Greenwood
And so I’m playing i’m playing mostly with ChatGPT, but I also play with Grok, Gemini, and many others ah every day.
35:51.25
Natalija Pavic
Yeah. Yeah.
35:54.57
Jason Greenwood
And I tell you, it’s been so interesting because Even I haven’t plumbed the depths of what they’re capable of yet. And I’m using them literally every single day. And I use many and many additional software tools that have AI at their core.
36:03.47
Natalija Pavic
Yeah.
36:06.96
Natalija Pavic
yeah
36:07.04
Jason Greenwood
so i’m And I’m building apps in Lovable just just to get experience with it and just to understand what vibe coding looks like.
36:11.84
Natalija Pavic
yeah
36:13.55
Jason Greenwood
So I think if if you can stay curious, then you know i I don’t have a crystal ball. I can’t tell you what it’s going to look like exactly in one or two years, but I know that AI is going to be more used more than it is today. And and that that I can be sure of.
36:28.06
Natalija Pavic
All right, it’s time for some predictions. Okay, so I like to do predictions here because I like to look back and see um see whether or not we were right or wrong.
36:30.72
Jason Greenwood
Yep. yep
36:37.56
Natalija Pavic
um If you had a crystal ball, Jason, what do you think is in store for us with AI and B2B and everything we discussed so far?
36:45.80
Jason Greenwood
Look, I think I kind of look at B2B and AI a little bit like I look at India and some other countries that completely skipped the desktop computer revolution.
36:56.86
Jason Greenwood
I have lots of friends and I have lots of clients based in India, um based in Southeast Asia, based all over the world, right?
36:57.37
Natalija Pavic
Yeah.
37:03.53
Jason Greenwood
And in many of these, you know, quote unquote, third world or second world countries or whatever you want to call them that were technically very late ah to the digital party, In many cases, they completely leapfrogged inferior technology, and they went straight to, like for for many Indians, this was the first computer they ever had in their life, and it was instantly wireless. It was connected to the internet 24-7.
37:27.28
Jason Greenwood
They had it in their pocket pocket or within six inches of their body 24-7 from the first computer that they got. And so we have a situation where many Indians, for example – They’re more technically advanced because they didn’t they didn’t deal with all the rubbishy – they never had to deal with dial-up.
37:45.01
Jason Greenwood
They never had to deal with big, clunky desktop computers. in In many cases, they never never even had to deal with laptop computers. They went straight to tablets and mobiles. And so they leapfrogged much of the Western English-speaking world.
37:58.79
Jason Greenwood
Because that’s just the way the technology rolled out in their countries and wireless made all the difference in the world, right? They didn’t have to have terrestrial phone lines, which many, particularly people that live in remote areas, they didn’t have.
38:09.24
Jason Greenwood
They didn’t have terrestrial landlines. And so they didn’t have access to the internet until they got one of these. And so I think that’s exactly what’s happening with AI now is that instead of having to have three e-commerce replatforms before we get to the good stuff.
38:22.15
Jason Greenwood
Now, for the very first time, when B2Bs implement commerce technology, it is going to be AI first. It’s going to be AI ready. It’s going to be AI enabled. And it’s most likely going to be SaaS in in most instances to where they’re going to get the benefit of all these upgrades without having to replatform moving forward.
38:38.73
Jason Greenwood
So they’re going leapfrog a whole bunch of really crappy commerce technology, and they’re going to jump straight to the good stuff. And that’s very exciting for me.
38:48.13
Natalija Pavic
And one thing that we talked about is this possibility that, you know, who knows what our ah economy will look like? You know, obviously we still need consumers and if we don’t have consumers, we don’t have revenue. So there is an element of stasis or stapleness that we will achieve with AI usage.
39:03.73
Natalija Pavic
What I’m excited about is the opportunity for smaller teams, smaller companies, um individuals to do things that were only possible with large organizations before at a fraction of the cost.
39:16.36
Natalija Pavic
Tell me more about that because I know you’re doing some amazing stuff with your usage of AI is insane, obviously.
39:22.38
Jason Greenwood
Yeah, well, look, I wouldn’t be able to put out three podcast episodes a week. I wouldn’t be able to ah you know edit them, post-produce them, distribute them. i wouldn’t be able to do all of that without AI-first technology, right?
39:34.87
Jason Greenwood
Like with my podcast, I use Descript and I use ah other AI-first technologies to help me with that.
39:34.95
Natalija Pavic
Yeah.
39:40.36
Jason Greenwood
ah I’m using ai to build apps that able to that. wouldn’t You know, the beauty of vibe coding using apps like Lovable is that you’re going to see a lot more smaller pieces of software in the world that are tailor made for the business or tailor made for the individual.
39:57.52
Natalija Pavic
Um, yeah.
39:59.13
Jason Greenwood
Right. So I’m not a coder, but I’m a very technical non developer.
40:03.79
Natalija Pavic
Right.
40:03.88
Jason Greenwood
I am now, so so I’ve had lots of software ideas over the years, but I never actioned any of them because to even get to a basic functioning prototype took took yeah it took so long and it took so long and it was so expensive that I just never, i never took action on them.
40:12.20
Natalija Pavic
It’s like 200 K. Yeah.
40:18.08
Natalija Pavic
Yeah.
40:18.24
Jason Greenwood
I am just about at the end of developing an internal use tool that I’m going to
40:18.32
Natalija Pavic
Yeah.
40:23.99
Jason Greenwood
sell. It’s going to be called Jason
40:26.05
Natalija Pavic
yeah
40:26.17
Jason Greenwood
And it is it is trained. It is a chat interface, an AI-based chat interface that’s trained on all of my YouTube content. And ah so basically I’ve created a rag pipeline that’s vectorized all of my content based on the transcripts from YouTube.
40:41.65
Natalija Pavic
Yeah.
40:41.92
Jason Greenwood
I now I’m building a vectorized chat interface over the top of that, that I can query. I can query all my content. yeah know What’s the best B2B e-commerce platform? Then not only does it give me a structured answer based on the content, but it gives me the references of where it’s pulled pulled that information from, which video,
40:58.04
Jason Greenwood
the exact time point in that video where I discuss that topic and it’s I’m making it clickable so that then it opens that video in a new tab at exactly the point where I talk about that subject or whatever wherever me and the guests talk about that subject.
41:11.24
Jason Greenwood
Now, that is amazing because I’m coming out with a new um subscription service for my consulting whereby it’ll be a group or cohort based consulting for for for businesses that don’t need that heavy duty one to one based consulting.
41:25.23
Jason Greenwood
But they will get the companion Jason AI app along with that, that they can ask 24 hours a day, seven days a week. They can ask questions of my content and get meaningful responses. So this is the type of thing that like I would never have done that before.
41:38.12
Jason Greenwood
Outside of these vibe coding platforms, I would never have i would never paid the $20,000 or the $50,000 or whatever it would cost to have a ah developer build this, test it, take three months, whatever. I’ve been working on it just for a few weeks, and it’s cost me less than $1,000 so far.
41:50.15
Jason Greenwood
So very exciting.
41:50.94
Natalija Pavic
That’s very exciting. um and And, you know, this speaks to like the possibility that AI will democratize SaaS. Like think about, you’ve got these like huge vendors, these huge monoliths that kind of control, all the have all the power because they can come along and say, there’s no way you, Mr.
41:56.90
Jason Greenwood
Totally. Totally.
42:06.37
Natalija Pavic
Customer, could develop your own e-commerce platform.
42:09.50
Jason Greenwood
Yes.
42:09.69
Natalija Pavic
We’ve got 20 years behind us. You know what i mean?
42:11.39
Jason Greenwood
Yes.
42:11.61
Natalija Pavic
And we’ve got 200 developers. and And the customer says, oh, no, I only have like two developers. And like, I don’t even know what I’m doing in e-commerce, so I’ll give you the money. But now what you’re saying is, especially for B2B, right, where each application is unique, each portal is different, everybody needs something else from their website, and there’s no out-of-the-box, ready-to-go template for B2B, they can and in you know in-house some of that development and be like, actually, I’m not worried anymore because I have AI tools.
42:38.89
Natalija Pavic
And what’s great is that those companies that struggle with governance because of size, because of red tape, because of how they’re structured, will struggle to adopt that technology for productivity reasons. so um So yeah, I think this is kind of optimistic.
42:52.86
Natalija Pavic
Like I think we’re almost going to move away from these um ah behemoth monopolies towards a more fragmented ah vendor, a more competitive vendor structure and where customers are empowered to make their own products, which is interesting because we’ll move back to the maybe but make your own or maybe even like buy two belts kind of paradigm.
43:12.45
Jason Greenwood
Yes. Yes.
43:13.81
Natalija Pavic
So.
43:14.69
Jason Greenwood
Yeah, it’s exciting. maybe If I think back 20 years ago, there was a lot of access-based database apps that people built inside their organizations for very select pieces of work in their business, right?
43:21.92
Natalija Pavic
Yeah.
43:27.20
Jason Greenwood
Specific automations, specific efficiencies. that’s That’s what I think Vibe coding is going to enable. I don’t think the monoliths are going go away anytime soon just because that they they they fulfill so many large-scale, very complex functions in the business.
43:41.48
Jason Greenwood
So don’t think your major ERP players are going to go away. like I wouldn’t. dream of Vibe coding in ERP. I probably wouldn’t even dream of Vibe coding in commerce platform, but it’s the little pieces of functionality that I need that are very unique to my business to do a very specific function that is not covered by the the ERP, the commerce system, and some other ah major apps that I’m already running in my business or the D365s of the world or whatever it is that I’m running in my business.
44:08.12
Jason Greenwood
There are very, very specific functions that historically I would have had to rely on a custom software development house to come in and build for me. Those little pieces of apps or functionality that glue the entire organizations together to make us really efficient.
44:23.20
Jason Greenwood
We can insource a lot of that now. and And I think that’s very exciting. and And I think it keeps a lid on, especially for the micro SaaS apps that are out there that do lots of little things, it’s going to definitely keep the lid on the escalating prices of those.
44:36.73
Natalija Pavic
That’s very good news for customers. So, all right, I have ah one last question for you, which is like, you know, you already talked about Lovable, Vibe, Kodang, you’ve talked about your vectorized, what did you call it, RAG?
44:38.75
Jason Greenwood
Yes.
44:47.19
Jason Greenwood
Vectorize daddy store. yet Yeah.
44:48.50
Natalija Pavic
Okay, something, RAG something.
44:48.47
Jason Greenwood
Yeah.
44:49.82
Natalija Pavic
i don’t know what RAG is exactly, but i’ll we’ll put that in the show notes.
44:50.32
Jason Greenwood
Yes.
44:54.08
Jason Greenwood
Yeah.
44:54.15
Natalija Pavic
ah What’s your favorite or any any other tools or any other way of using AI that you’ve discovered recently?
45:00.74
Jason Greenwood
Uh, Look, i’m I’m using AI literally every single day for everything. i have it review documentation. I have it. It it is it is the most amazing research assistant that I’ve ever had. and And so when I’m out there looking for data, like let’s say I’m doing research on behalf of a customer.
45:21.35
Jason Greenwood
And I need to put together a report. I need to whatever. And I need to get some raw source material, some source data from across the Internet. I can tell you that that chat GPT is far, far better at finding source data than than a standard Google search. Obviously, now Gemini augmented search is a little bit better, but but.
45:41.42
Jason Greenwood
Still, if i its ability to pull down information out of out of images, out of PDFs, out of all these different documents that are scattered around the internet to pull meaningful information and context out of that data so that then I can present it and reference it in my reports, ah it is phenomenal. I’m saving money.
46:01.32
Jason Greenwood
i’m saving I’m saving probably 75% on research time just because of of of answer engines, is what which is what I like. They’re answer engines. They’re not search engines. They’re answer engines. And that’s the difference.
46:15.93
Jason Greenwood
That’s the key difference. So it’s helping me in almost everything. And and even down to, oh, can you know I’ll upload. Okay, here’s my report, my first draft of my report. scan this, give me i give give me any ideas in ways that I can improve this, make it slightly more clear, make it more readable, make it more digestible, you know what you know just to just to get some feedback like a research assistant assistant would do And then ah you know just give me a, okay, we recommend you do this. We recommend you do that. We recommend you do this.
46:44.63
Jason Greenwood
ah you know here’s Here’s a sample so that you can see kind of what we mean. And then boom, I can go and I can make it more readable. I can review it. I can make sure that it is just as perfect as it can be before it gets to the customer. So there there’s a lot of things here.
46:58.04
Jason Greenwood
um I already already use it in content. I will continue to use it in an escalating way. in And in fact, i’m gonna I’ve got like four more ah micro apps already documented that I want to build after I get finished building this Jason AI app.
47:11.35
Jason Greenwood
There’s some content distribution and redistribution stuff ah that I want to build that doesn’t exist. not ah well It doesn’t exist in the market the way i want it to work.
47:21.81
Jason Greenwood
And so therefore I’m to, I’ve got like four more or five more apps already in mind that I want to build to make me more efficient in my daily processes that I have to do not only with client work, but also from content perspective. So I’ve got some really exciting things up my sleeve.
47:34.74
Natalija Pavic
That’s really, I can’t honestly can’t wait to see what you do next and what you develop next. So um thank you so much for joining me today and dropping knowledge bombs as always.
47:44.59
Jason Greenwood
No worries.
47:45.90
Natalija Pavic
And to everyone else, until next time.