AI and The Future of (Re)Commerce

Karri: Hello and welcome to the Recommerce podcast, the podcast where we talk all things Recommerce and circular businesses. This week, we are talking about AI and how it's going to affect both commerce in general and especially Recommerce. And we're going to go through both buyer and seller or the merchant side. Like every other week so far, I'm joined by Tuomo Laine, the co-founder and CEO of Twice Commerce. Welcome.

Tuomo: Great to be here. Thank you, Karri. We need to start inviting other people to the podcast also.

Karri: That's what I've been hearing. Yes, yes, we'll have to see what we can do about it. Yeah. All right. Let's start going through the buyer journey from the buyer's perspective and how is AI going to affect that? So how is e-commerce or or commerce in general going to look like in a couple of years that the AI and MCP and all of these new technologies are are swimming into the world of commerce.

Tuomo: Yeah, sure. I think it's it's probably easiest to start thinking about like from product discovery. How do people actually discover products? So, now I think the product discovery route might be that if you need a product, you you might already go directly to the brand's website or then you just simply Google it. And if you think about Googling stuff, you usually have, you know, websites as results, the online stores, but you also see those product cards there on the top. Um, now, AI driven discovery is to me is kind of instead of going to Google, you might go to, I don't know, Chat GPT. So you go to a uh uh chat client like uh where you can go and have a discussion around that hey, I might need to I want to buy a iPhone, where might I find that and or maybe you have a more specific need like in Recommerce that I'm looking for an old Jimi Hendrix vinyl from this and this era in this condition. So, what happens there is that instead of you searching for like an endpoint, like a appropriate website, you are kind of hiring an an AI agent to do the search for you. I think that's a big, big uh change in in the product discovery process. So you you will rather have a discussion with a AI uh uh AI agent that then goes out and tries to find that product from for you from uh different endpoints like websites and so on. And Chat GPT could be one of those. I'm pretty sure that at some point one will be able to just like attach their credit card to Chat GPT to allow it to actually do even purchases on your on your behalf. But there's also other uh players that I think will have a role here. So wallet apps, uh I'd see as a logical place to kind of for them to start introducing this capability. So companies like Klarna, PayPal, uh I've seen some stuff by Visa and Mastercard on on this stuff also that their own wallet apps might have some kind of product discovery uh tools. So I think that's one of the kind of first things. And this is not Recommerce specific, it will affect all of the commerce, but for Recommerce it just maybe makes finding secondhand goods a lot easier because kind of finding that secondhand good that like specific old Jimi Hendrix vinyl, it's a bit harder uh to do, I think in in in your traditional Google Googling and and and all of that.

Karri: So most likely the first step is going to be, I think there is already some capabilities in in Chat GPT that it's going to maybe go and look for those products and then kind of report those to you and then you are in control if you want to actually actually purchase those and probably the end goal or the next phase is that you just say what you want and maybe what you are willing to pay for that and then the agent actually goes and and and finds the product for you. And in in that world I'm kind of worried about like how are the sellers and how is the whole market going to look like because most likely there there could be some deals of course with some merchants with the with the uh AI agents and the platforms that are offering those that they are working working together but if you are for example saying that I always want to find the cheapest price on something that you have already been able to do and if that's fully automated like how what's going to happen to the stores that are near you that are maybe more convenient but at some point when you when you don't actually need to need to uh worry about that either when the shipping is is handled fast and everything and basically everything is going close to the margin zero from the seller's perspective.

Tuomo: Yeah, I think it's it's maybe a thing that relates a bit to this is that also is is kind of what's your purchase uh preferences outside of price also. So uh it's not necessarily always about like a specific price like we've used to do we've gotten used to kind of doing, I'd say active instant e-commerce purchasing. Like I designate 15 minutes to buy my jeans from online and I want want to find them quickly and I do the purchase decision quickly and why I want them shipped immediately. Um, I think what AI agents might open up is this more passive search also. So I kind of hire that agent and say that I'm looking for these jeans in this price range. I'm willing to, you know, spend two, three weeks even if like allowing the agent to kind of scan and crawl and see if anything pops up in a in a Recommerce marketplace for example. Uh and you can give preferences like soft preferences that I'm willing to actually go to a store if it's 10 miles from my home. Uh or I might prefer a mom and pop shop over a big retailer or whatever the preference is. Again, those kind of filters would be kind of hard in my opinion to set to like Google and try to find it in an active shopping session, but if you allow an agent to passively kind of uh find those, uh it might be possible and that allows also the seller and merchant side to kind then differentiate a bit. So it's a question like is there is it an active purchasing or do we have like agents passively on on the market out looking stuff that meet some some bit more softer criteria than just you know, price and immediate availability.

Karri: What about the transparency in that that world? I think that's something that most likely worries a lot lot of people but for me example when I'm using these chat agents and especially when I'm talking about something that maybe doesn't have facts like like that you are able to verify easily then kind of relying on the chat like what it's actually outputting me like if if it's saying in this instance that hey this is the best place to buy it like is it the best place for you as a customer or is it the best place for the AI agent or or what's the how do you see that that going forward?

Tuomo: Well yeah, it's uh probably two things. Like something that is like outside of Twice's uh scope but it's like whatever agent you would choose to use probably depends on your purchasing uh criteria. So it sounds that you would like to use a a purchasing agent that is extremely transparent that does that whereas someone else just wants the job get like done. They're less sensitive for some of the aspects. So I think that's what will set apart like hey how does it why do you use PayPal agent? Why do you use Klarna agent? Why do you use Chat GPT? So it is kind of just a preference uh thing then that for for some it's more mass buying, for some it's like specialized buying. Um so that's kind of my tendency that that's probably the reason why there will be different providers that then focus on different aspects of the purchasing experience. Now from the merchant side, um you touched upon the idea like hey what if there would be a store nearby that is better. This is kind of classic that that like if you're a completely offline store your goods, your inventory or your catalog is not indexed by anyone. Of course you're kind of living in a void. You you can only be found by someone walking in and browsing through your uh offering and maybe that could be your brand kind of that you have to come here to know what's here. But for majority of the smaller stores I think it's going to be a uh necessity to make sure that your catalog and your availability of of of of your inventory is constantly available not only in your own online store. So people will not find your online store and start browsing but rather to be kind of indexed on an individual product level. Kind of the same as in with Google now you have the product cards there on the top of the uh search results. In the same way I'd say kind of imagining that the AI agents actually go to the product card level kind of that they these are the things that they would be purchasing and maybe showcasing to you inside Chat GPT or similar. So so I would say that for the smaller merchants it becomes a necessity to to have a software that allows you to index your catalog and and inventory in real time for to be discoverable by AI agents, not only your website.

Karri: How how do you see that being different between like linear sales where you are selling or buying new items that should be like identical no matter where you buy them but then on the resale, Recommerce side where every item is unique. How do you see like the AI affecting this this process because in uh in many of the resale stores for example the uh camera store that sells the used old vintage cameras and so on. I think for example the pictures and and all of those are like super important though. So do you see that the AIs are gonna be experts in all of these areas and they can say like hey maybe in that picture there is something that even human cannot see. So is it is it gonna be like a better or worse experience or at least now I think it's mostly relying on like text input from the website so I I'm not sure if it's actually analyzing all the images on on websites at the moment but probably in the future.

Tuomo: Yeah, I think there's there's multiple aspects to that. So first of all I think the it's probably important that your when you have a website like or it's not necessarily always that the agents would access the catalog by crawling your website and going through that. They might access it via like MCP layer or kind of getting even better access to the uh catalog rather than kind of scraping your website. Uh in that context things like images become a thing that like have you made your images uh to be such that they can be analyzed by AI which essentially would be like Vision API uh like AI Vision and well for example Vision API by Open AI can kind of go through that and analyze what do I see there. Uh or then at minimum providing kind of uh already as a on a catalog level or article level if you have images being able to provide like a AI agent friendly description of the image. So not saying that it's an iPhone X Y but actually describing a lot better like that it's an iPhone X with a slight scratch on the screen on the bottom right or something like that. Now in in in the context of Recommerce uh like you said in in linear commerce if you're buying new stuff all of those are comparable. So once you've found found it somewhere you can just basically do price comparisation and maybe shipping speed. And that's it. It's a uh it's the same product. In Recommerce I think the the role of the agent become can become a lot more valuable. So it's kind of helping you by saying that okay I've found three of these phones. All of them are a bit different. One of them have scratches on the on the glass. One of them has maybe reduced battery life and one of them uh is wrong color but pristine condition. So which which one would you prefer? Here's the images, here's the links and so on. So it's a lot more concierge driven relationship that you have with the agent because again you're hunting for that perfect resale find or secondhand find uh and and if you would do that manually in in terms of active buying you would spend a lot of your buying time just finding the products and analyzing it whereas an agent can remove a lot of time, a lot of friction from that purchasing experience and just offer you the kind of final details like concierge like here's the three things and all of them have pros and cons and uh which one would you prefer if any. So I see that uh agent purchasing will become even more important for Recommerce than maybe your linear commerce where it's more around discovery and price. In Recommerce it's everything like preferences and uh small details.

Karri: Yeah, that sounds interesting. So in with new items it's mostly about shipping price and maybe how easy it is to do returns or are they included in the price or something like that. But yeah in the in the used categories it's like every item is unique and and if the AI is able to help with that that that's probably going to be pretty valuable uh and interesting to see where where it kind of goes and I can easily see that it fits this kind of product discovery but if we go all the way to like buying buying. I think there might be that new items might be easier to just tell agent like hey find cheapest and fastest and just get it for me but on the on the used side you probably at least before you kind of start trusting it that it's actually like delivering exactly what you want you probably want to do more of that like end decisions after the agent has has found the products for you but how how do you see the buying experience evolving?

Tuomo: Yeah, I think in in buying there's these few things that you mentioned like the agent could do a checkout already. That's of course uh one thing that you allow you give it a credit card, you give it a budget and you allow them to do purchases maybe below a threshold or or similar. Like grocery shopping could be like that all right you I allow you to buy the stuff uh with 100 bucks and here's the recipes that I want to make this week. Uh and then errors if if they happen they're not too too dramatic. Uh So I yeah buying is one. Uh I think shipping is one and then then there's an interesting category uh probably a bit niche but what what's interesting like uh idea is that I've seen a few startups that for example focus in uh let's call it like price elasticity or negotiation capabilities. So they they kind of allow businesses to say on their website that all right this item costs $1,000 in the website but in practice if you would go in store there's always wiggle room. Like nobody like almost like with with uh I don't know if it's a global thing but it feels that with furniture there's always like the you know the bed costs $1,000 and then you're saying okay it's a bit expensive and the seller always have wiggle room to give you some like uh discounts there. Kind of the same with cars. So that translated into an agent relationship is that the buying agent that you've hired is not just looking at a uh static catalog but it's actually interacting with a seller agent there who has a bit bit of a wiggle room also. I'veseen some concepts on that. It's it's kind of interesting and sometimes they end up in a super odd place. Uh but it's kind of the would you allow a seller agent also to do some kind of that hey what if you buy this bundle or what if you do the like like these campaigns that if if you end up buying five you get the fourth one free. So they're kind of allowing this level of facilitation uh that would be interesting to see. A bit niche maybe but uh I I've seen a few concepts on that. I think that's interesting the that kind of could I tell my purchasing agent that I like this but go back and try to get me a discount. Uh other than that I think there's in in in in in buying experience uh deliveries are are of course one. I think there's like now if you think about when is the delivery happening I think that experience experience is usually like okay I have to agree a certain time slot. Uh and depending on the delivery method it might be like from 9 to 5 or then it's maybe a two hour slot or similar. You have a tracking code, you constantly get some updates on where the item is. I think many of those things can be transformed from a kind of a static notification to a more of a like a discussion almost with the customer uh support or kind of a concierge service where the agreeing of when when things get delivered or or knowing what's happening there might be a more of a agent relationship.

Karri: And that's probably going to be interesting when the AI assistants kind of like the new generation of Siri or Google assistants actually start to be powerful enough that they kind of know your schedule, they know when you're home and they can also do that automatically in the in the background if they have access to your calendar and and emails and all of that stuff. So they actually know really well like what's the good time to receive the product and maybe something that happens in the background and doesn't involve you at all.

Tuomo: Exactly. So there could be a lot of like automation on the on exactly like you said home assistants saying that like doing this this this kind of facilitation that bring it on Thursday at 10:00 a.m. and uh then your schedule changes and maybe it can handle some of that and when it cannot handle it maybe it will kind of notify you that hey don't forget that you have that delivery happening at 10. So I think that's that's kind of uh one of the uh things. Then when we go beyond purchasing and we go to maybe returns there um you usually return an item due to it not being the right item or it doesn't fit you or then it's damaged. So uh for those damaged uh items or similar there's probably a lot that uh where AI vision and and everything that can be kind of deduced from that can help both sides. So you taking an image of the item allows a lot of things. It first of all first of all it allows to declare whether it's the right item so that you're sending the right stuff. The seller knows that okay this is what we will be receiving back. They get a condition declaration kind of already from the image an idea that all right is it how damaged is it. Uh and and to kind of make a deduction that do they want it want it back or would they just like for example want to ship you a new one and ask you that do you want to keep the old one. I think there's been cases like with Amazon or or similar where the unit economics for them in some price points it's easier to just send you a new one and not take the old one back. Uh So I think that return process where uh AI vision can kind of create a declaration for an item. So what it is what is its condition and all of that can help the the reverse logistics process that whether you send it to the merchant or whether there's a third party in between that takes it into a marketplace or to recycling and all of that. I think there's a lot that can be removed a lot of friction that can be removed uh there for both sides.

Karri: That's very interesting. Yeah. Do is there any other aspects of the buyer journey or buyer side that you see that AI is going to have a major major uh effect. So I think we now covered like product discovery, buying experience, a little bit on the delivery and shipping.

Tuomo: There's been a few concepts that I've seen also. There's been like in in the social uh social commerce side of things so if you operate on a website there could be things where um if you need assistance in an online store environment I think there's these shopping assistants, virtual shopping assistants that uh are there that you can instead of uh doing this like filter based uh search you start by chatting with a like on website assistant. Uh and then there might be some like community aspects that might happen there that again more more they've been more on a concept level that I've seen in in other startups doing but there's been like community building almost like across websites that if you're interested in this stuff and kind of building building uh building communities and and and kind of facilitating those with AI but they they are still on a on a concept level. But I think there's a lot that when you end up on a website we'll probably start to see a lot of that like almost like when Intercom came and everywhere all of a sudden you started to have that chat or or or other providers. I think we're going to start to see similar uh concepts appear on the website from like shopping assistance and uh and kind of uh more personalized ways of finding the stuff that you need.

Karri: Yeah and actually one one thing that came came to my mind was from the Google's IO demo. Not sure if you saw that but you were able to kind of I think it's taking a photo of yourself and then actually try out all the clothes that are are available in that store like how they actually look look on on you and I think there were a lot of lot of positive feedback from some of the early testers on that. So that's especially something that's interesting in the in the uh clothing accessory space where of course you only know when it's actually on top of you what it looks like. It's it's if it's on a some model on on a website it always looks good.

Tuomo: Yeah the body types tend to be different but the the same I think IKEA has been doing similar stuff in in um in in furniture. So like even like tapping into AR VR uh where you can see kind of the items in the real space. Yeah. So uh whether it's AI or whether it's a uh I think the AI might be that it helps you to kind of generate the concept of how you like what kind of furniture this room might have or what paint to use on the wall and then you can make it even more powerful by just positioning them with VR or other like other technologies that allow you to then superimpose stuff on on on top of existing uh real world.

Karri: All right so that covers maybe the main points from the buyer buyer's side. So what about the seller or the merchant side?

Tuomo: Yeah I think with the with the seller we touched upon the idea of of making products discoverable. So making sure that your catalog and inventory is indexed uh in a way that is accessible for AI agents. I think that's the that's the important stuff. That's kind of the basics that you want to make sure that if someone's asking Chat GPT or similar where might I rent a bike in uh Los Angeles that you you are one of the contenders to be showcased there. So that's kind of the basics and I think that's where it's more of a qualification towards the software that you select to use for your commerce that they're capable of of building your catalog in a way that uh can enable that. Now on the other side of things there's a lot of like operational things that probably where AI helps you a lot. There's the basics that everyone saw uh as as kind of AI became more mainstream and and and most of the commerce tools are have already or pretty much every software tool is embedding these. It's the kind of generative capabilities. So instead of you having to figure out product descriptions for every item you might be able to kind of just generate them uh by AI. So if you want to have a nice uh product description for the surfboard and then you have some brand tonality there's a lot of tools that help you just generate that rather than you having to kind of type all of all of that out or and then you can maybe fine tune the last last mile there. So I think that's the kind of basics that when you think about registering items and generating the kind of data there like what is it what what variants does it have and the product descriptions and so on. A lot of that can be uh automated uh with AI. So that's just like like time saving and also again at the same time enables maybe a bit better purchasing experience for the end user. Uh other than that I think there's other operational aspects where for example inventory registration not only enriching the content but actually in in specifically for Recommerce where we talked about that that where vision capabilities can help a lot that if you're handling product returns or maybe your whole business is a secondhand store. So if you register in the inventory it can be as easy as just snapping a picture of it. So even if you wouldn't have like a brand catalog against which that image would be analyzed the general language models are pretty good at identifying what is this item and giving it a condition grading that can be used in your own inventory. So like item registration is as simple as snapping a snapping a photo of that item as it comes in. So that's to me like a again it's kind of generative but there's also the the element of vision vision capabilities there that allows the prompt to be born almost like yes please register with this data schema uh something for this this image that we have here. So I think that's that's a definitely a a core capability that even small store owners should have better data more enriched data and they can create that data more easily and then if the technology partner that they're working with is on on par with the modern standards then that data is discoverable and indexed by AI agents in the same way as websites are indexed and crawled by you know Google crawlers.

Karri: What what phase are we currently there? So if you are seller and you have a software that is enabling AI LLMs to kind of index that is it is it for discovery basically and then they are guiding you to your website for the purchase or is it already possible to kind of make the whole uh integration that it's actually all the way to the to the purchase or is that something that's coming up in in near future?

Tuomo: I think we're currently like on mainstream level where we're we're at the moment where your catalog and website can be discoverable. Uh the for example Chat GPT can kind of talk about that inside the chat client and then direct the user to do the purchasing themselves. Uh I would like I think there should probably are already kind of first like providers that allow AI agent driven shopping. I should check the latest Y Combinator batch but I'm pretty sure that there's should be at least two or three of these I would imagine. Uh but when does it become mainstream I would estimate like a year maximum two where companies like PayPal, Klarna or similar would introduce this or Chat GPT would introduce these capabilities like commerce capabilities inside their own existing UX. So I think that that would be the time when when things start to become more more mainstream and uh more merchants will organically start to ask the questions from commerce software providers like us that hey is this so that I'm now discoverable in Klarna app.

Karri: How how do you see when that starts to happen is it something that's going to take over or is is it that you basically have a new sales channel that is just through the AI now. So you already have bunch of different sales channels so is this just another one or is this actually going to replace most of the existing like are you going to have an online store at at some point?

Tuomo: Yeah uh I think you're going to have always like have an online store. It's well let's think about it through like there's never been like I don't think there's that many true replacing happening like overnight. It's always a new channel but the share of the channel just grows over time at the cost of some other channel. It might kind of grow the size of the pie in general. But if we start like first people went to like everything was in store. Everything was walk in. Then we started having mail catalogs and like mail orders. That started to grow. And then internet kind of took mail orders and mail catalogs to be digital. And then well it was again a new channel. People were still ordering with with mail catalogs still going to the uh uh physical stores. Uh maybe the overall purchasing also increased a bit but probably the e-commerce started to eat out the mail catalog uh more and then the the walk ins a bit less. Uh and then like social media buying and Klarna and these kind of wallet apps and such again new channels or marketplaces. New channels they take a share of it. What would be the prominent channel depends super uh depends a lot on the actual like brand and product. Like groceries there was like practical limitations on buying groceries online for a long time whether we can kind of have the logistics working there. Also uh if we now think about physical retail stores uh there's been this trend that slowly some physical stores they move from like warehouses and purchasing towards like showrooms. You go there you test the stuff they become a little bit more clean there's less of of of like clutter in terms of like just having items out there. Uh and then you go there figure out what you want but then when you do the order it arrives to your home in two days. So that might be the similar thing in in in that happens to like websites that all right the website instead of just being an ordering thing maybe they start to focus more on this showroom capabilities like we discussed earlier like how would this furniture look in my home or how would these uh clothes look on on when when I wear it and then the purchasing and the browsing of the overall catalog becomes a bit less secondary. So maybe the actual purchase then happens with an AI agent that is embedded in the in the online store. So that's how I kind of perceive it. So it probably wouldn't replace but it becomes a relevant enough channel and that will grow over time and the technology that you need in order to kind of because it will not be like one agent channel it will be probably hundreds. Uh so you just want to make sure that you're kind of agent friendly.

Karri: Okay. So that's the main thing near future or already kind of happening that merchant should be looking for in their software is like maybe first just creating the catalog super easily and and using AI there and I think that's that's something that you are already able to do inside some of the commerce software but also just by using Chat GPT and and putting the image there and say like can you do a description and most likely all of these let's say basic or simple capabilities are are coming to the commerce platforms soon and I know that they are already on on Twice. And then the next step is to have it discoverable so kind of do the distribution to the buyer side of this AI AI agents and and so on. But how how is AI then going to influence like the components of of this um merchant world where you are for example working with inventory or or pricing the items and and kind of maybe not optimizing but making sure that the operations are are working properly and why not of course optimize the business business for you as a seller.

Tuomo: Yeah. I think it's important well and I I have to disclaim that I'm not even even like the professional in this but it's like it's important to differentiate for example in inventory we could talk about like forecasting demand forecasting and supply forecasting or dynamic pricing like uh based on weather or whatever. Uh what's AI and what's just like like machine learning and and and data models like under like now we tend to throw the term AI on top of everything everything intelligent okay it's AI. Uh so there have been you know companies like Relex and and and I know there's a lot of dynamic pricing companies like Price and similar that already before AI hype did like a lot of uh data driven optimization of dynamic pricing and demand forecasting and so on. Um maybe the role of AI there might be that for less professional merchants or that kind of smaller operations that do not have the resources or knowhow or time to do like very complex data analysis on forecasting um these commerce platforms can embed these forecasting capabilities but deliver the results in a human readable uh format kind of via AI chat agent driven relationship which might be more understandable for a uh less data driven user.

Karri: So providing like recommendation and tips and tricks that you might want to implement or.

Tuomo: Exactly. So rather than giving like a huge Excel and and and whatnot maybe it can give a more simple that hey you should probably double your uh I don't know this gene model for Black Friday because seems to be the one that is is now hyped. Uh I think it could be simple stuff where you know the the the capability is the same but the way how it's delivered and packaged with an AI agent makes it more accessible for for the uh less resource than merchants with that has less resources to kind of analyze complex stuff.

Karri: Do you see that this is also like a journey that starts by AI kind of giving you these recommendations that you end up implementing yourself and maybe once you start trusting it or just in in in some time when you are kind of overall maybe trusting the AI then this all of this stuff can happen in the in the background automatically.

Tuomo: Yeah I think that that's an interesting aspect and probably like less visible to the merchant but it's maybe in the interesting geeky topic to uh discuss is like what does like agent automation look like under the hood. So if we've gotten used to kind of the idea that we do automated workflows where I do like we we have even this like visual drag and drop tools almost like Zapier or similar where all right if this happens then do this if this happens then do this. Now the reasoning capabilities of of many of the AI providers the agent providers are pretty good or even great in some cases. So instead of you having to map out all of these workflows you could hire a kind of AI agent to do the reasoning what should they do. Then they just need access to these various configuration endpoints. Uh and that's kind of the role again where the commerce platforms would need to make sure that like all of the APIs to actually trigger things like change the price or create a new catalog item or new bundle uh product for this campaign that all of those actions are available for those reasoning agents. Uh and that probably means like having MCP layers or model context protocols and and so on on top of your API so that the AI agent can kind of make sense of what's available for them to play with. Uh so so that's kind of for the merchant like at the end of the day it is more of like you describe that all right what do you recommend the reasoning part of the thing and then they would need to kind of be able to access the data to do the reasoning well and then it's would say that and I would create a bundle of these two products with this price. And then if it seems like a good idea for you you can actually do it yourself. Uh and then if it's constantly delivers good results then you might say that hey well why do I need to be here in between let's just like do your recommendations automatically like autoplay. Uh but under the hood there's kind of the the difference is that is does the agent has access only to the data or does it also have access to kind of the service layer where they can trigger these things.

Karri: So are you describing how we are actually losing the jobs to AI here?

Tuomo: No I I wouldn't say that because I think there's still like uh well if if we think about like self driving cars the technology has been somewhat there for a while but somehow there's this kind of trust gap that you want to you want to kind of fine tune the last mile or similar. I think it's going to uh also like the world of commerce or like any business you're trying to find that edge somewhere. And AI tends to be bad at finding the edge in in my opinion. It's it's a good thing to give you edge if no one else is utilizing it. You just get so much better productivity than anyone else that you get so much more done than anyone else that that's your edge. But then when it becomes more mainstream and it's embedded into these processes that's not anymore an edge and that maybe the kind of everyone gets some of the reasoning and and and uh those like so then that becomes the standard and then finding the edge finding the alpha might be again more of your human innovation that do you uh you kind of fine tune the last mile you change the price a bit or the product description you you like you you write it to be a little bit more edgy than the then what the agent would would do or you do some kind of uh multi channel hack that would be hard to then just model into one central AI. So I think there's always a role but the productivity definitely will be a lot higher individuals can achieve a lot more uh teams can be smaller people can be focusing on other aspects uh more so yeah so I I wouldn't be too pessimistic but usually when productivity goes up it means that uh yeah less people might be needed but then what the focus of the of the people is also shifts a bit.

Karri: Yeah and I think historically when these tools come that make you more productive it doesn't usually mean less work it actually means more work but now you're just able to able to also do do more work more more easily. And uh you mentioned like the self driving cars I think it's very interesting example where it seems like the standards for like AI or computer made decisions is so much higher than for human like if there is one accident caused by an AI it's like oh can we even trust this technology at all but at the same time there is like thousands of accidents caused by humans so probably similar dynamic also in the commerce space that when you maybe try it out and it makes one error you might be like oh wait let's let's take a couple couple steps back and like let me do actually the decisions but probably at some point the accidents when it's only AI driving the cars maybe the accidents actually actually disappear.

Tuomo: Yeah definitely and uh yeah and but this is probably then like there's bunch of uh like studies and such that adaptation of technology that you have early adapters and then the mass and it takes time to kind of emotionally and and in all of the ways to kind of adapt to that. But I think maybe for practically for any merchant listening I think it's we're we're probably still in an era and will be for a while where just adapting AI workflows augmenting your operation with AI workflows being mindful of it almost like digitalization early days making sure that you kind of are making wise bets on on the software that you use and being future proof for a long time it can give you the edge just by just by the fact that you have that better productivity in your workflows. Uh and then that edge might be the thing that for the next years that allows you to kind of increase your market share locally or globally. But then when when it becomes more mainstream and and and people catch up or other merge competition catches up what has happened you've kind of grabbed a larger market share and then you start to kind of the fine tuning but now these disruption moments are the ones where you can kind of take a bigger share of the pie probably just by adopting things a bit faster than your competition.

Karri: Sounds very similar to maybe like internet advertising when there is a new channel it's usually like works really well but then when people start going in there it kind of becomes similar to the other other more mature channels and you don't have that alpha alpha there anymore and and then you are looking for the next one and and if you are adapting these and and able to grow your business you are maybe in a little bit better position when the when the next wave comes in so sounds like a pretty similar strategy to actually adapt the AI.

Tuomo: Yeah definitely.

Karri: All right so we covered a lot of different things for the merchants sellers so far so are there any other main use cases that you see that AI is going to influence in the in the near future?

Tuomo: Yeah I think probably we would have to do like a day's worth of a podcast if we would go through all of them but definitely I think there it's harder like AI is a horizontal technology like like digitalization so it will affect everything and it there will be blurred lines of what's really AI and what's like machine learning or data like big data or or what but there's categories like fraud and and and payments and such where AI can help a lot. Um similarly you know there's like uh there can be like operational kind of governance related things where AI might help you in like tax optimization or similar and um so like it's probably if you can come up with a job to be done there's probably AI can can help you there. So I think it's it's it's important to just like or trying to kind of simplify what enables most of those changes is the under like being having the capabilities where your processes and the core data resources which tend to be customers your product catalog your inventory and your orders that those are managed by some system that is built in a way that these resources can be accessed by agents or or kind of even operated by agents and then it's like your decision when you want to go into that but like and and that each one of these resources in their underlying things like inventory that all right can I just register items by an image and all of these things so if you're operating on those probably it's easier for you to implement the AI capabilities across any job that you might come up with.

Karri: Okay. And maybe at the end I know that there is the new Twice platform coming out soon and and it has been designed to work in this new modern AI era so maybe just like a quick overview of like what is possible with the new Twice platform and how is it working in the in the AI world and helping helping the merchants.

Tuomo: Sure it's uh it's going to be super fun to just showcase it rather than to talk about it but I think it's it's many of the things that we discussed here that have been built into the fundamentals so first of all if we think about the core resources the inventory catalog uh customers and order management the approach that we've taken is that first of all like that we we don't force workflows to you but rather it's quite flexible data models behind it so you can store quite a lot of different data for each one of the assets and which means also that when accessed by AI agents or even created from an image by an AI agent a lot of information can be stored which like unstructured information also which isn't like uh modeled into some taxonomy. Um then like a geeky thing we've uh we've utilized some open source taxonomies behind uh everything so when for example one use case that you can do with the new platform is that if you get a product return uh you can just snap a picture of it and we've even made it so that it doesn't have to be with the device you can use any device available so in a retail environment you usually might have one iPad shared by the store personnel so with our system you can just like uh sync it with the with the session take a picture of the of the product and it creates immediately registers an inventory item for it for it uh and then does all of the things that we discussed here that it checks what it is it taxonomically categorizes it creates attributes for it does condition grading pricing all of the things that you kind of from a single image you have immediately a product that you could put into a shop in shop secondhand corner and at the same time being listed online and even uh uh listed in in any third party channels that you might be using like uh eBay or similar. The core thing that we've made is that all of these assets are AI agent friendly which means that this uh database schema there behind it is we put a lot of effort in thinking how how to make it so that we're not limiting you any any any of the merchants but we're rather enabling we're kind of we have sensible defaults that enables to create the a sensible default but you can take it as far as you want. Uh those are and then well one cool thing which is which I kind of like that in any view you are in our product you can kind of see what's the API endpoint or even the MCP kind of the agent endpoint for you to uh create the resource edit the resource or or even delete it so how do I access this like what I have in my screen right now how do I access the this via programmatically or via an agent. That's uh that's I think uh kind of trying to bridge the gap that you understand that anything that you do graphically there or kind of in the graphical user interface you could be doing programmatically which means that an agent could also do it because we've we've made sure that the APIs are uh AI friendly.

Karri: Okay. Sounds like a lot of lot of new capabilities coming to Twice real soon.

Tuomo: Yeah definitely.

Karri: But hey thank you Tuomo once once again for uh really nice insights and let's see if you're going to be the guest on the next episode as well.

Tuomo: I kind of I don't I don't know whether I should say hope not or hope so.