How AI Predicts Buyer Behavior to Maximize Profit with Simeon Lukov

Recently Shopify connected with open AAI and once you start chatting and searching for a product you may create direct purchase from open AI without knowing the merchants uh or the brand. So that might be a new thing in in e-commerce of course with some budgets limitations and some human in the loop that's the the next thing in commerce that is coming. Hi, my name is Dmitri Bonichi and I'm a content creator, agency owner, and AI enthusiast. You're listening to the AI agents podcast brought to you by Jot Form and featuring our very own CEO and founder, Idkin Tank. This is the show where artificial intelligence meets innovation, productivity, and the tools shaping the future of work. Enjoy the show. Hello and welcome back to another episode of the AI Agents Podcast. In this one, we are featuring Simeon Lukov, the founder and CEO of Dynamic

Pricing AI. How you doing, Simeon? >> I'm doing great. How are you doing? >> I'm doing awesome and living the dream. That's uh that's what I always say. Um and it is the truth. Maybe it's not the truth. We don't know. Um but no, in all seriousness, uh Simeon, thank you so much for making the time today. Hopefully, you're living the dream building out your company uh right now. How did you first and foremost get into AI and how did you get into this um kind of realm of AI specifically? Um well um I used to be in the e-commerce business uh uh back then in uh and I saw that um I need to do something something better in terms of tooling and uh from e-commerce uh I did my first and second startup and the second one is dynamic pricing for uh e-commerce.

ers to change prices and this is how I get into AI. Uh also my background is more technical. Uh I've been uh CTO and um I'm interested in math. That's my education algorithmic and math and it's also related to AI in a way. >> Yeah, totally totally fair. um what uh what about um your company, you know, kind of uh is is unique? Tell us a little bit about what you guys do and and uh how it stands out from the the the crowd. >> Sure. Sure. So uh dynamic pricing AI is um right now focusing um high high volume merchants uh in the Shopify ecosystem and u is bringing new capabilities for first first of all price testing rapid price testing then for u revenue and profit uh optimization and then uh we bring uh lots of pricing rules and widgets for uh e-commerce

merchants. >> Okay, very cool. Uh tell us a little bit of what that looks like for for companies. Talk more about it. >> Um what do you mean in terms of >> Yeah, just like dive dive into it. Just dive into it a little bit more. I mean, obviously it's a it's a good breakdown, but like um you know, what does it look like um for people who are using it? like what uh yeah, >> you know, obviously you have you have different companies that that use your product. Like give give some cool examples of of how it works and how you help people out. >> Yeah, some some case studies. Um well, people are usually doing three things. First of all, they're they're trying uh to sell on a high margin. So, uh, for example, you have new collections, um, of, um, some products

and you're not sure what's the exact price that, uh, the market can pay for that one. So, you can try um let's say you you're selling a jeans uh you put the jeans for $35, but then you see that you have um good demand for that one and you try higher price point like $40 $45. That's that's the first application and um this is how you get uh more profit. The other one is um to check uh for discount. So um sometimes people undercut um while putting too much discount uh on a particular products while they don't need it. So they might try 15 20 25 30% discount and realize that 15% are bringing same amount of orders like 30%. So no need to undercut. [clears throat] That's that's the second one. And the third uh application u is um to test the willingness to

pay and the perceived value uh for for a particular product. So you test higher prices and lower prices just to see where you are what's the the optimal price positioning for for your products. >> Okay. Interesting. Um and you know I'm taking a look at your website right now. It's It's interesting to me you've um how do you guys kind of structure it from uh excuse [clears throat] me um like a pricing standpoint. How do how do they like what do they pay for and and I guess like how have people kind of seen the ROI come through because obviously some products are like just sign up and you you pay but you don't have to like disclose total pricing but how what's the like model work like? >> Sure. Sure. So first of all uh we have this um uh free um free tier

that you can check all the capabilities of the system and um okay >> once you get uh rewards on uh once you see revenue coming you can switch to a first paid plan uh that is 79 uh currently per month and uh you are mostly uh binded by the number of orders that um uh merchants the Shopify guys are doing per month. Uh so it's of course the first plan uh allows limited number of forders. Then you have 449 and 999 which brings uh more AI models uh uh more capabilities uh AI recommendations and um some agentic workflows. >> Very cool. Um, you know, I I want to talk to you a little bit more about agentic workflows. What What do those look like uh for your company, too, because obviously we're all about AI agents here. A uh agent workflows are the biggest thing.

Uh what does that term mean to you mean to you, especially at dynamic pricing AI? Um well for us um agentic flow and um agents um mean something that um could support merchants uh in uh taking decisions um sensing and uh analyzing um lots amount of uh data especially their business matrix and KPIs performances of uh their portfolio so that um that that's easily sensed by um by models by by the agent uh and out of that um agent recommend um the best next step uh for uh for merchants. So they have uh pre-built pricing campaigns uh and uh this is this is the flow. uh they just um make review of the campaign and they they can launch something pre-built from an AI agents especially for for their uh for their shop. >> H very cool. And what um what are you guys uh

recently what have you guys recently released that you um are really excited about um productwise like uh updates? Um well uh we have um uh first of all this uh identic workflow uh and then uh and then we have um new type of price testing uh which is called switchback. Uh this is when two prices uh alternate between uh each other but uh this is sequential model. So um this is um uh counterintuitive uh it's uh it's not new but uh all the things that are doing um are very very new new stuff in terms of price testing and this switchback um allows um CRO guys the guys that are doing optimizations to see and to feel what we are actually doing with uh with the rest of the tools. >> Okay. Interesting. Nice. Um that's a cool update. And uh tell me a little

bit about um kind of the the integration um piece of the thing. I know you um you have it listed on your website. It's for Shopify, um Magneto, Salesforce. Tell tell me about like kind of the ease of use of of the integrations and and and whatnot. I I know for a lot of people that's that's a big deal how easy it is to get it all set up. >> Sure. Sure. Yes. So um definitely uh Shopify is um our our target and the integration is uh quite easy because they have uh extensive documentation uh they have um nice uh API uh for most of the things uh and uh the other ecosystem that uh we are participating not that strong are uh Magento, Salesforce cloud commerce and uh Presto shop which is um another opensource um commerce uh popular in especially in Europe. So

uh in terms of integrations um everything is uh I may say quite easy with uh with Shopify and um a bit more challenging with uh with the other platforms um being u open source or having some legacy old stuff of communicating like like the one with uh Salesforce but it's still working and um while merchants are using those platforms uh and we're going after big merchants uh we need to to do integrations. That's one thing. Uh uh the other integrations uh are with um similar apps uh complimentary apps uh apps that uh are very popular in uh some spaces and um this is this is going more case by case. >> Okay. Gotcha. Very cool. Um, what do you think is uh your favorite um uh thing that is going on in the company right now? Whether it be um you know just like uh

growth recently or um just yet or like a interaction you have with a client. What's kind of uh been really getting you excited about everything going on at Dynamic Pricing? Um the most exciting thing is uh that uh we get lots of installations for the uh past three months being on the Shopify marketplace. Um I see uh the enthusiasm in uh our team with uh every install because we are tracking uh what's going on and um uh we have more than 200 installs uh for that time. Um, of course, uh, our task is to convert those, uh, installs, uh, um, in a funnel and make, um, customers uh, out of that uh, installs. That is, um, the challenging part, but uh, we are still happy. Uh, we have um, bunch of uh, five stars reviews and um, paying customers. So this is um those are

the last cool things that uh my team and me are very excited. >> Very cool. No, that's uh that's awesome. And uh what in the the world of AI uh as well outside of just your own company has uh been top of mind for you recently with what you're seeing in the industry? >> Um sure. So recently we we started experimenting uh a lot with uh Gemini models uh for uh our AI agents and um those are the most uh exciting team uh things that that are happening uh it's not um it's not only the models but the infrastructure uh around the frameworks uh that appear on the market and um uh so uh I'm really excited uh about um Google uh identic development kit uh as a framework from one side uh and uh all the capabilities um from that framework uh I see

uh similar frameworks are popping up uh like uh Microsoft uh release uh similar framework but um yeah um I really like uh how Google have structured the the framework work uh from one side uh from the other side um uh they have great documentation uh and they uh they have published uh a book uh for um designing properly an agent. So um uh I was uh at the point uh where uh I was discussing this um designing u designing an agent and um all the learnings um that are uh written in in the open source book uh by Google uh and all the practices that uh developers uh can follow in order to to to have like meaningful uh agents uh and um uh things that uh let's say um senior developer are are mostly familiar with. Uh in the beginning it it was just

hype like um you you can uh you can have a model and you can build agent but uh but after that um uh you need to think uh how to manage uh the memory, how you can store the sessions um how you can how you can use uh um rack uh for different purposes so that um uh the agent is not spending uh millions of doing uh a task. All that uh things have their patterns and um uh I I see how the industry is uh moving uh more from uh from the hype to a real uh applications slowly but surely. >> Uh what are your long-term goals for this company? Like what's the what's the ethos of the company? what what is the what do you want companies to get out of this uh overall? >> Oh well our next steps uh um are

to try dynamic pricing in other industries while we have uh uh request for pricing um train tickets uh sports events um drive roadside uh assistance where we have uh an application Um and currently we are exhausting the e-commerce uh space for dynamic pricing but um uh recent change in the market uh uh in terms of taxes uh in terms of tariffs in terms of um unstable uh market uh situation. People are trying to find tools that are quickly uh operate with their prices uh in uh in a way that they have uh still have revenue profit or uh mixed both. >> H yeah no that makes sense. Um, I uh I think what what I'm most excited about when I just take a look at your um your company is um the there's so many different and this is what I'm curious about. There's so

many different tools right now that are coming out from an AI standpoint to help out retail and and e-commerce and um I think yours definitely stands out as a interesting individual one. But what would you say is maybe some of the stuff that you'd like to add right in order to to help out um your companies even more? Is there anything in the works that you're you're working on that you think specifically as a feature you you'd like to add soon? Don't want to break any confidentiality or anything, but um something you can talk about. >> Sure. Sure. Yes. So um we talk about um AI recommendations and uh LLM how they uh help us um for the um for merchants and for uh AI recommendations. But uh the uh the actual think and the actual work uh is um is done by uh framework

which is um kind of promising. uh this is fast reinforcement learning uh called uh multi-arm bandit and contextual bandit framework. So um you know the the LLM could not predict uh the um the best uh next price LLMs are predicting tokens. Uh that's why we need uh we need to have uh our um own proprietary models uh for dynamic pricing uh that are getting uh reward signals from um from merchants in terms of sales and penalties. So based on the um rewards and penalties from from the market uh we can train uh this fast reinforcement learning AI uh and on [snorts] top of that use uh the LLMs not the opposite. So um in terms of um uh the models the model side and how those models are are learning. This is um quite special because um they they start learning from day one. They

don't use uh historical data to to train themselves but um start operating uh and gets rewards and penalties from from the environment uh from from the market. Uh and those models um can take uh can take different context uh for uh for the objectives that uh you want to achieve. for example um uh if this is a weekend or weekday uh or what time of the day is or how is your uh competitors doing. So having uh all the uh context the engine uh can real time decide what's the best price for that scenario for that context and um that's the thing that uh make us different from u uh everyone else. >> Absolutely. No, that's that's that's really cool. um what are some of the main like um industries that you found have have kind of latched on to this? Obviously, you know, the

e-commerce is kind of a broad term like is there any specific types of products that are commonly being sold in different industries? Is it more like mechanical? Like what what does it kind of look like that you're noticing? >> Yeah, sure. So those are industries that are selling not expensive products but >> um more more of a uh emotional purchases. Let's let's say products up to uh $2 $300 that uh uh people don't think too much, don't check uh so much. uh those are products that they need and uh they like the current price. No matter uh if the jeans are 36 or 38 or $39 once you need the product uh you make a purchase. But for um for the customers for for the uh for merchants those uh price difference uh could have uh huge sense. Uh once they sell hundreds or thousands

of products every dollar counts and uh at the end of the day they get nice additional revenue. >> H nice. You know, I um I'm also kind of curious as well about something that um I think a lot of people would be in in this market. You know, obviously um you mentioned like more lower ticket items are are kind of working into this. Why why and how have you seen has there been any like higher ticket pricing products like any um uh I guess more expensive products adopted if not your product any other types of like improvement products in the e-commerce space with AI because I'm trying to get more of a grasp overall on what it kind of looks like. I've seen Shopify build out its own AI. I'm just trying to understand more of that because we really haven't interviewed a lot of

people in the e-commerce space before. So I don't know if you have any other cool landscape knowledge that I'm I'm kind of missing. >> Um well the yes there are there are many different tactics um also on the front end side cuz um you uh as a user on um e-commerce store you might see lots of uh widgets that may decide uh to purchase particular product. Uh so um we're building uh widgets that um uh can sit next to the price and uh and tell you uh this is a trending products the price might go up so uh you need to make a purchase within the next uh hours. uh or uh on the opposite side. Um we may put a batch uh under the price saying that uh there is discount cuz uh this product uh is out of season and the price will

decrease. So that uh you may try and check uh if it's expensive product on another day if there is discount uh or not. Um or it might be uh super quick flash sale for 1 to 5 minutes that you have like um um mystery discount especially for you in that u particular five minutes. You need to decide uh if you want to purchase that trending item on a lower price or to move on. Um and um of course there is uh personalization that is taking uh into account your uh purchase history um what kind of products uh you're purchasing and um there are always special prices uh for your segment as well. >> Yeah. you know, I I haven't uh I had heard a little bit about um the whole kind of I mean, and we've seen this for a while, right? Like with what's

going on and um the world of Amazon and the algorithm and whatnot. I've se I've heard a lot about like companies trying to take advantage of uh the AI to essentially give more personalization on the experience um for what they're selling on their website, which is is pretty intriguing to me. um where do you think some of the the more interesting innovations are going to come out of the e-commerce um uh AI space here soon? Like what other uh because that to me is a big is a big improvement. And then I I really like your competition monitoring. I think that's pretty cool on your website or on your product. um what new things that maybe you wouldn't be implementing but other bigger companies um might be trying to do uh in general or kind of in your industry that are trying to uh really

AI um AI um AI enhance uh the uh you know experience for shoppers. Um first of all uh from um from the AI perspective uh that that might be new generation of uh chat bots that are uh doing uh >> better better and more complex uh tasks. Um not not only to help you purchase a product but uh after that showing where these products uh are uh and some after sales um uh chats uh with the merchants. Uh you may refer to uh documentation via chat or um some additional service and um uh you have um different distribution right now. Uh once uh you are talking with the chat then uh you might be hooked with um WhatsApp or email or another uh messaging. So definitely hot space uh for um chat and uh assistance uh in general. Um the other thing that I uh

that I see popping up uh is um for um building building um product listing pages. um all the automation that uh that that you need to to insert uh images, descriptions, uh features of the product. It's getting uh better and better. Uh and uh the other thing the last thing uh might be um the back end uh the back end of the cloud uh cloud provider uh where they they allow um easily migration from one platform to another platform via AI. Uh that that's cool also for agencies, integrators. Uh right now switching from platform to to another one is getting uh easier. >> Nice. Very cool. Okay. And um you know in general obviously there are people who have done this manually, right? Um in some respects and and little parts here and there. Um you know obviously a lot of the chat bots answering

questions and stuff for people. Um, I think e-commerce and customer support kind of go hand in hand. Um, you know, for uh the recent year we've kind of seen AI chat bots just become kind of universal on websites. What do you think about um the impact of of those on to websites, right? Like how how do you feel like the impact has kind of changed things for from a customer service perspective for e-commerce stores? You know, you install it on your website, it answers question for you. You know, those ones I'm trying to I'm saying, right? Oh. This is uh this is the the the current um state like u chatting with um u with with the chatbot. Um most probably the the thing will uh shift more on the direction that um you do purchases from um Chad GP or from um Gemini directly

cuz you're searching and you get um you get uh several proposals directly in the chat window. So um recently Shopify connected with open AAI and uh once you start chatting and searching for a product uh you may um you may create direct purchase from um open AI uh without knowing uh the merchants uh or the brand. Um so um that that might be uh new thing uh in in e-commerce. Uh the other thing uh would be uh agent to agent uh purchases. So um imagine that uh you are getting some supplements and you need to to restock uh and you have your agents that is taking care about your entire health. planning uh your fitness u uh your your meals and your supplements. So those purchases uh might go via uh via agents. Um, of course with some budgets limitations uh and uh some human

in the loop uh but uh that that's the the next thing uh in commerce that is coming. >> Okay, very cool. Um you know I uh just have to ask because I think it's a decent like thing top of everybody's mind. What do you think that these improvements are going to kind of have an actual impact on on jobs? Right. Right. And I mean like job amount. Obviously people are curious about that with AI. It's kind of the most top of- mind question everybody has right now. Where where do you stand? Do you feel like it's just going to make everyone's jobs easier, but we'll keep the kind of same workforce? Do you think it'll remove people from the workforce? Where do you kind of stand on this issue? Um so I I think uh the hype of um AI taking over jobs um probably

is declining because um some of some of the godfathers of AI appears and uh said that um well uh there will be a time that agent will take care about many things but uh this time has not come yet. Uh so um after that I saw um many job offers uh popping up and um currently I think that uh uh people are uh are not in those balloon that AI will uh will replace them soon cuz um still there are hallucinations. Um still there are things that uh human needs to definitely check uh in um different industries like um legal even in the marketing. Um there should always be someone checking uh the the performance of um of AI and um and the agents. uh and um uh in terms of um workforce um I think there there will be replacement uh but uh um

that that might lead to to conflicts uh lots of conflicts so I don't think uh um I I don't think that we should go in that directions because Um, imagine you you fire um so many people what they're going to do. Uh there might be like crisis. So um yeah, I think that people were exciting uh initially but uh right now everything is um go to normal and people see more agents and AI as a tools instead of replacement. Considering that a lot of these companies like yourself are using open AI and claude and all these different models, how do you feel about how a lot of you know companies are getting uh being put in a situation where they're backbone of their company is based off of like another company's component. Obviously, we all already see this with like everybody's already on AWS, right?

So, that's like kind of step one and then now we're kind of like double stacking it almost with another set of tools, right? Like whether it be cloud, how do you kind of feel about that whole situation? And do you feel like with the way that they've lost money, you know, because they they haven't really been gaining money that maybe their pricing model will uh shift at all in the future um those types of companies? >> Yeah. Um you're absolutely right. So from um from from business uh perspective um this um locking with um AI uh and LLM model vendors uh might not be super healthy uh cuz um at the end of the day so many new apps uh appear uh and there is no uh big uh differentiation between functionalities and the capabilities of apps. The only differentiator uh is the distribution. How

fast uh you will show something that is wrapping up a model uh on the market. And um that's that's why we prefer to um uh to put the investment in uh research and development of uh additional models that are uh independent of LLMs and uh can operate especially in our dynamic pricing vertical um without the the need to use LLM. them uh for the core part of the job. But um to do like uh reinforcement learning uh and um machine learning uh in order to be kind of uh independent and have um advantage uh and not locking with uh with particular vendor uh and yeah LLM model >> absolely Absolutely. No, that makes sense. Um, and kind of last but not least, would love to hear your final uh thoughts on what is your favorite um AI tool um that you're using right now that

you uh use on a consistent basis. >> Sure. Um so um from uh open AI uh it's uh charging and uh the last uh the last models they they have and um from Google uh this is um Gemini 2.5 uh models uh and um I really like how uh Gemini is um uh getting back and Google are getting back in the game. Uh and the one thing that I particularly like for for their models is that they are very very fast. Uh they they they deliver the response uh super quickly uh quicker than than the rest on the market and uh this started to make sense. >> Yeah. No, absolutely. That that's fair. I really um I'm a big fan of there's like there's one there's one like my sleeper favorite thing is um uh Gemini uh honestly like 2.5 pros uh deep research mode

is is a lot um it's really good for like setting up um research uh knowledge bases for like um agents and stuff because it it has like unlimited deep research which is something that a lot of people don't talk about. you can just like spam uh Gemini 2.5 Pro deep research and I love Claude 4.5 Sonnet for writing. It's by far and away the single best writing uh tool that I've seen so far. So, big fan of that. Um all right, so with that being said, I hope you guys uh like this one. I just wanted to say last thing uh Sime if you could please let me know um and the the you know everyone where they can go to check out your product. Um, first of all, dynamic pricing uh AI and then Shopify marketplace dynamic pricing AI optimization. >> Awesome. Awesome. Well,

thank you so much, Timmy, and I appreciate your time. And, uh, everyone, please make sure to leave a like, comment, and subscribe. Make sure to leave a review on Apple Podcast and Spotify so that everyone here can see how great uh, dynamic pricing AI is. Thanks for watching. >> Thank you so much. Bye-bye.