WEBVTT
00:00:04.799 --> 00:00:16.079
Welcome to the AI Accelerator Podcast, where we get past the hype and into what actually works, practical insights and real-world use cases you can apply in your career and your company.
00:00:16.320 --> 00:00:19.359
Today's guest isn't just talking about agentic AI.
00:00:19.519 --> 00:00:21.280
She's shipping it at scale.
00:00:21.600 --> 00:00:34.719
Priya Lakshmi Narayanin is the chief product officer at Recurly, which is the subscription platform behind brands like Paramount Plus and Twitch, serving over 76 million subscribers.
00:00:34.960 --> 00:00:46.399
She spent 25 years building product at Microsoft, PayPal, Meta, and Brex, and now leads Recurly Compass, which is an AI engine that doesn't just analyze the business, it takes action.
00:00:46.719 --> 00:00:48.479
Priya, welcome to the show.
00:00:49.280 --> 00:00:50.560
Thank you so much, Matt.
00:00:50.880 --> 00:00:55.439
It's great to be here and talking about AI and acceleration.
00:00:55.920 --> 00:00:58.479
I'm the chief product officer at Recurly.
00:00:58.640 --> 00:01:02.479
Recurly is a subscription management and recording billing platform.
00:01:02.719 --> 00:01:11.760
We help businesses, like you mentioned, household names, you know, Paramount, Twitch, et cetera, handle the complexities of the subscription business model.
00:01:12.000 --> 00:01:21.200
Our mission right now is to set subscription growth on autopilot, utilizing AI technology, agentic workflows, et cetera, for our merchants.
00:01:22.959 --> 00:01:27.760
So let's walk through your progression because your your resume is spectacular.
00:01:27.920 --> 00:01:33.760
So you've built Cloud and Microsoft and PayPal, Meta, these big names, BREX, right?
00:01:33.840 --> 00:01:35.680
And now you're CPO at Recurly.
00:01:35.840 --> 00:01:45.840
So walk us through the journey of how your career progressed and now and what pulled you into the subscription economy with a lot of things you know unfolding in the tech world today.
00:01:46.239 --> 00:01:47.120
Yeah, absolutely.
00:01:47.280 --> 00:01:50.640
Um believe it or not, I started as an intern at Microsoft.
00:01:50.799 --> 00:02:00.879
Uh, you know, those days they they would pamper us quite a bit with uh goodies and trips to Redmond, and I drank the Kool-Aid and didn't look back for 12 years.
00:02:01.040 --> 00:02:10.000
It was a fantastic ride at Microsoft where I started out as an engineer and then early days of Azure, I moved into product management.
00:02:10.319 --> 00:02:16.800
They wanted somebody with a technical background to help our customers sort of migrate onto the Azure platform.
00:02:16.960 --> 00:02:33.120
Um, and then you know, went to other enterprise companies after that, EMC built storage products at EMC, um, based on hybrid clouds connecting on-premise storage to you know other clouds like um AWS and Azure, etc.
00:02:33.520 --> 00:02:41.520
Um, and then uh went to PayPal for worked on FinTech there, and then Meta worked on commerce.
00:02:41.680 --> 00:02:47.280
So early days of COVID, commerce was just uh online commerce was going through the roof.
00:02:47.360 --> 00:02:56.800
So it was a fantastic experience at Meta building um trust and safety products for commerce, so ensuring that buyers on you know Facebook, Instagram, etc.
00:02:57.120 --> 00:03:03.199
Um, can truly trust that we are selling the right products, they're engaging with the right sellers.
00:03:03.439 --> 00:03:14.879
Um, and then after that, moved to BREX to work on um expense management and corporate credit card products, um, really got you know into it.
00:03:14.960 --> 00:03:17.120
Was I I would say early days of AI.
00:03:17.199 --> 00:03:20.000
Um, November 22 is when Chat GPT came out.
00:03:20.159 --> 00:03:22.240
Uh 22 is when I joined BRECS.
00:03:22.400 --> 00:03:34.080
And so a lot of experimentation at BREX, uh, working on you know uh assistance and how to do continuous finance uh with uh with AI workflows, etc.
00:03:34.400 --> 00:03:41.360
And now at Recurly, truly building agentic workflows uh for our merchants for revenue optimization.
00:03:42.000 --> 00:03:43.759
Wow, there yeah, there's so many pieces there.
00:03:43.840 --> 00:03:46.719
You have such a variety of product experiences are really exciting.
00:03:46.800 --> 00:03:51.599
There's been a lot of things we can unpack here in our uh in our podcast together, Priya.
00:03:51.759 --> 00:03:52.319
I'm excited.
00:03:52.400 --> 00:03:55.840
So let's talk about Recurly for uh what you're doing there.
00:03:56.080 --> 00:04:04.560
I know that um part of what Recurly does and what you do there, leading leading uh product is is uh managing the subscriber life cycle.
00:04:04.879 --> 00:04:05.039
Right?
00:04:05.199 --> 00:04:12.560
So, what is can you explain what that means to maybe people who are outside the tech space or maybe not not in the subscriber tech space?
00:04:12.960 --> 00:04:14.159
Yeah, absolutely.
00:04:14.319 --> 00:04:22.319
Um I think everybody is so knowing or unknowingly uh um aware of subscriptions or using subscriptions.
00:04:22.399 --> 00:04:34.319
Like if you hit subscribe and save because you wanted to, you knew you would you would need that product again in two, three months on Amazon and hit that button, then you know you are a subscriber, right?
00:04:34.480 --> 00:04:39.920
Um and you have people have many subscriptions like Netflix and other streaming platforms as well.
00:04:40.160 --> 00:05:08.480
Um, and so our goal is to help merchants sort of go through the entire consumer journey of acquiring consumers who then become subscribers, um, and then you know, activating them, retaining them, and growing their subscriber base with them for a good lifetime value, um, while they provide more products and more innovation on top of their existing platform.
00:05:08.720 --> 00:05:14.399
So that's sort of what we think about as the um subscriber lifecycle or journey.
00:05:14.480 --> 00:05:28.800
So they launch new products, they acquire subscribers, um, they activate them, they then retain, they want to retain those subscribers through more innovation in the product or service, and then we they scale and grow their business.
00:05:28.959 --> 00:05:38.720
And through this entire life cycle or journey, Ricurly wants to provide the right capabilities and services so that uh we are there every step along the way.
00:05:39.920 --> 00:05:40.879
Wow, that's fantastic.
00:05:40.959 --> 00:05:42.800
So it's very much a B2B model, right?
00:05:42.879 --> 00:05:57.199
You're going after the larger uh enterprises that have uh subscription offerings, and you're helping them better manage the offerings, as you're saying, like reduce churn and and uh and uh and just you know maximize you know subscriber benefit, revenue, et cetera.
00:05:57.360 --> 00:06:12.079
Yeah, yeah, we predominantly work with merchants who are you know high volume of subscriptions, low order value, like 4099, 999, like the streaming media platforms, like I mentioned, um they accept consumer payments, so you know, direct-to-consumer services.
00:06:12.240 --> 00:06:18.240
Um, we also have quite a few B2B merchants on the platform, so sort of span the spectrum, if you will.
00:06:18.560 --> 00:06:19.040
That's fan.
00:06:19.199 --> 00:06:20.560
That's that's fascinating.
00:06:20.639 --> 00:06:33.680
Uh, not being a subscription expert myself, not having that type of experience, uh, question and curiosity I have is where in the world of AI or where where is AI making the biggest measurable difference right now?
00:06:33.759 --> 00:06:36.879
Is it more on like preventing the churn or is it winning people back?
00:06:37.040 --> 00:06:43.040
Is there another area like what what part of that life cycle is is where you're really uh making an impact?
00:06:43.439 --> 00:06:46.160
Yeah, I would say across across the board, right?
00:06:46.240 --> 00:07:02.959
So when you think of like a purchase funnel, it's like discovery, how do consumers you know discover your product when they consider it, like comparisons, so you so you make sure that they choose your product to buy um and then they make the purchase and then they stay on and retain.
00:07:03.040 --> 00:07:05.920
Um, that's sort of the best outcome.
00:07:06.079 --> 00:07:13.120
Um, so I think AI is making a difference in every step of that purchase funnel and the lifetime over the lifetime of the subscriber.
00:07:13.279 --> 00:07:22.319
Um, in Discovery, I think in a more and more AI native world, the storefront will no longer be a website that you specifically go to.
00:07:22.480 --> 00:07:25.360
Um, you know, it could be your conversational interface.
00:07:25.439 --> 00:07:31.199
It's you know Chat GPT or Claude, um, your favorite AI platform, right, of choice.
00:07:31.360 --> 00:07:40.720
Um, and you might go in and say, hey, what's the best way to watch you know the NFL playoffs this weekend or um you know the show called The Bureau on Paramount?
00:07:40.879 --> 00:07:53.920
And the agent will now act as on your behalf, like it'll go discover what's the cheapest way or the fastest way for you to get access to that content, and it'll provide a comparison engine, you know, platform A versus B, this is the price you would pay.
00:07:54.079 --> 00:07:59.600
They have a pay-per-view on Amazon, but on Paramount, you have to buy the subscription with a free trial.
00:07:59.680 --> 00:08:01.519
So it can give you all of these options.
00:08:01.680 --> 00:08:06.800
And so now discovery is through an AI storefront or AI channel, if you will.
00:08:07.120 --> 00:08:22.399
And then when you make the consideration, there's AI there as well, um, around you know, fast native checkouts, the payment methods you choose, um, you apply that, and then they once they purchase the product, they experience it, there's AI recommendations there.
00:08:22.560 --> 00:08:37.279
Um, and then you know, when you decide, okay, I've you know, I've watched this series, now I want to, you know, uh cancel my subscription, then you can uh imagine that you know Paramount would come back and say, hey, how about 50% off for the next three months?
00:08:37.440 --> 00:08:47.840
And those recommendations are also AI driven, and so so that we can know the churn propensity, like you said, and then help our merchants retain customers on the platform.
00:08:48.000 --> 00:08:50.320
So I would say sort of there's AI.
00:08:50.480 --> 00:08:53.519
I think the middle part of the funnel, there was AI for a long time.
00:08:53.679 --> 00:09:05.120
Um, now the discovery and you know, Justin's time, customization of retention offers, etc., more and more um, I would say generative AI is playing a part.
00:09:05.519 --> 00:09:07.440
Yeah, that's fascinating.
00:09:08.080 --> 00:09:18.960
There, there's uh, there's an aspect that you mentioned there about um uh you know about the uh well about discovery and people coming in, and then I guess it could happen anytime during the life cycle.
00:09:19.039 --> 00:09:22.080
I'd like to ask about uh uh failed payment revenue.
00:09:22.240 --> 00:09:23.440
So this is a big thing, right?
00:09:23.519 --> 00:09:31.919
If you're selling something, especially to millions of people, um, and I know I know that's that's part of um um uh what recurly assists with.
00:09:32.080 --> 00:09:45.440
And I think it was to the tune of like$1.6 billion, and maybe to date, or maybe it's an I don't know how what that is, but can you talk a little bit about failed payment revenue and how recurly helps with that particular side of subscriptions?
00:09:45.759 --> 00:09:47.039
Yeah, absolutely.
00:09:47.279 --> 00:09:50.639
Um we call it as intelligent retries.
00:09:50.799 --> 00:10:04.399
Um, so when a subscriber puts down you know a payment method um for you know when they purchase the subscription, and we then retry on the whatever if it's a monthly plan, monthly, or a quarterly plan, et cetera.
00:10:04.559 --> 00:10:09.759
Um, but credit card numbers change all the time, you know, fraud happens, it gets canceled.
00:10:09.919 --> 00:10:16.480
Um, a lot of times, you know, there are different ways, reasons because of which a transaction might not go through, right?
00:10:16.559 --> 00:10:20.159
It might be insufficient funds, it might be the reasons that I mentioned earlier.
00:10:20.399 --> 00:10:39.039
And so um what Rick curly helps with is we have a machine learning model that determines um how to intelligently retry the failed transactions, you know, what time of day, you know, when is payday for a particular uh subscriber of a certain age or industry they are in, etc.
00:10:39.360 --> 00:10:48.480
Um, and retry at that optimal time so that we increase the um success rate of the transaction and recover more revenue for our merchants.
00:10:48.639 --> 00:10:51.440
Um, there's also what we call as a dunning window.
00:10:51.600 --> 00:11:12.960
Um, a dunning just means that um, you know, what is a time period during which you will inform the consumer of hey, you have a payment due, you now the payment is overdue, you need to update, and so you can increasingly um you know raise the urgency for them that your subscription might get cancelled if you don't pay in now five days.
00:11:13.120 --> 00:11:17.039
Um, and so that series of communication also Ricoli helps with.
00:11:17.200 --> 00:11:29.759
And then over that time period, we do you know multiple retry attempts, which is what is sort of together is intelligent retries plus you know dunning optimizations that we offer.
00:11:30.559 --> 00:11:31.440
That's fascinating.
00:11:31.600 --> 00:11:43.919
So on the on the product side, um like about how large is the product organization there that does product development and so forth, like how how big is how big is Recurry's product organization, like number of employees, or I'm just curious.
00:11:44.320 --> 00:11:50.799
Yeah, we um we totally around you know um 350 or so employees.
00:11:50.960 --> 00:11:59.120
Uh we are spread out geographically um in uh Medellin, Colombia, in Austin, in Denver, etc.
00:11:59.360 --> 00:12:02.240
We also have quite a few employees in the Bay Area.
00:12:02.399 --> 00:12:06.720
Um the product org is about you know 11 or 12 people.
00:12:06.960 --> 00:12:15.519
Um we have designers, uh, and then we also have product owners and uh program managers, um, etc.
00:12:15.840 --> 00:12:26.799
So together we are looking at also evolving how we build products and taking that product development lifecycle to a more AI-native one.
00:12:26.879 --> 00:12:44.399
So we're going through that right now with you know building a PRD skill in Cloud that generates the PRD that automatically is then fed into you know to code generate, and then that is reviewed by a senior principal engineer, it goes into evals and then auto-deploys.
00:12:44.480 --> 00:13:03.440
So that is sort of our dream cycle, um, where then the you know the product managers, designers, product owners, et cetera, can then be more customer focused, can look at you know um new business opportunities, how do we um test things more often and iterate more?
00:13:03.679 --> 00:13:05.440
So it's it's an exciting time.
00:13:05.759 --> 00:13:06.399
That's beautiful.
00:13:06.559 --> 00:13:11.200
You are you started to answer my next question, which is like what type of AI models do you use internally?
00:13:11.360 --> 00:13:16.559
Uh, are you starting to notice benefit um from that from a product development perspective?
00:13:16.639 --> 00:13:27.679
Um, and I guess related to that is like what type of uh data uh do you collect from the marketplace from your clients, et cetera, to sort of feed that product development lifecycle?
00:13:27.759 --> 00:13:31.279
Because I know AI is great at looking across a lot of data, especially complex data.
00:13:31.360 --> 00:13:35.440
I'm just curious, sort of big picture from a product development uh cycle.
00:13:35.679 --> 00:13:41.440
I know you mentioned you're using Claude, uh, you know, Claud Code, which is you know super powerful uh model.
00:13:41.759 --> 00:13:46.559
Um can you talk a little bit about like what type of data are you collecting in so you can improve the product?
00:13:46.639 --> 00:14:02.639
So sort of the learning engine that you're building in between humans and AI, obviously, it's not just all AI, but just like how do you um how do you approach the uh data collection and then um you know product molding the clay of the next generation of the product?
00:14:03.039 --> 00:14:04.240
Yeah, absolutely.
00:14:04.399 --> 00:14:25.200
Um, I mean, we of course track adoption and usage data on we have a merchant-facing product, so you know, admins who come in and create new subscription plants, uh, pricing, set pricing for their subscriptions, create coupons, create offers, et cetera, um, that then they launch.
00:14:25.279 --> 00:14:30.080
So we sort of have you, you know, usage data on how they're using our platform itself.
00:14:30.159 --> 00:14:36.799
And then we have um a lot of data around the actual payment that we capture for the merchants, right?
00:14:36.960 --> 00:14:48.559
Transaction data that you know we then use to determine um, you know, the retry timing and better revenue recovery for our merchants based on that as well.
00:14:48.799 --> 00:14:55.200
And so the the key thing is it's not just um one merchant, we have many merchants in a vertical.
00:14:55.279 --> 00:15:11.679
So based on you know, vertical specific, we could say digital media and streaming, um, you know, in addition to uh you know monthly plans, you can offer like weekend passes for certain events, and you will do better because you will get you will acquire net new subscribers.
00:15:11.759 --> 00:15:20.000
So we are able to provide insights like that to our merchants that then help them grow their revenue as well as acquire more subscribers.
00:15:20.159 --> 00:15:41.200
Um, and then on the churn side, um we also have uh a product called Recurly Engage, which um which allows us to on the subscriber side, allows our merchants to deploy um what we call as prompts uh based on actions that the subscribers are taking on their app or website or in the TV app, etc.
00:15:41.519 --> 00:15:51.840
So as they're about to cancel, we can track that and then present an offer based on you know how long the subscriber has stayed on the platform, how much do they use the product today.
00:15:52.080 --> 00:15:55.360
Um, and that then helps with churn mitigation.
00:15:55.919 --> 00:16:03.519
Wow, there's so you you're so uh I'm sure I'm sure your team is probably two, three, five times more effective than they would be without AI.
00:16:03.600 --> 00:16:07.519
Like without AI, it's just so much, everything you just shared, there's so much involved with that.
00:16:07.600 --> 00:16:11.200
And it's it's only so much the human mind can can cover, right?
00:16:11.440 --> 00:16:15.279
So um yeah, I mean, even the even designing an offer, right?
00:16:15.360 --> 00:16:20.480
Like I'm sure you must have seen some of those offers yourselves, you know, based on uh the apps you use.
00:16:20.639 --> 00:16:22.960
Like I use an app called Better Me.
00:16:23.039 --> 00:16:31.519
It's like a health and fitness app, um, where I, you know, in the everyday with small mini workouts, you could do whenever you can fit in the time.
00:16:31.759 --> 00:16:35.600
And I didn't, because I was traveling for work, I didn't used it for a little while.
00:16:35.759 --> 00:16:40.000
And it, you know, it sort of showed me a prompt saying, Hey, you haven't used it in a little while.
00:16:40.080 --> 00:16:42.320
Like, do you want you know X dollars off?
00:16:42.399 --> 00:16:48.799
Or um, so those prompts now we can help our merchants design those with the help of AI.
00:16:48.879 --> 00:17:02.799
So we've done an integration with Figma and our compass assistant, where you can, using natural language, say, Hey, I want you to increase the tone, you know, more friendly or casual, things like that, where you don't need a whole copy team to go do it.
00:17:02.879 --> 00:17:05.839
You can try different iterations, you can run tests.
00:17:06.000 --> 00:17:12.240
Um, so it's uh yeah, we we we we are now finding a lot of uses for how we can integrate.
00:17:12.640 --> 00:17:13.680
That's fascinating.
00:17:13.839 --> 00:17:16.240
What what's your favorite part, Priya, about your role?
00:17:16.319 --> 00:17:22.960
Because you have you had a number of leadership roles across different uh companies, and obviously um you're where you're at now for a reason.
00:17:23.039 --> 00:17:24.079
What's your favorite part?
00:17:24.240 --> 00:17:27.119
Is it is it working with the with the team internally?
00:17:27.200 --> 00:17:28.480
Is it designing the new products?
00:17:28.559 --> 00:17:29.920
Is it the creativity involved with that?
00:17:30.000 --> 00:17:32.240
Is it more of interacting with the clients, I guess?
00:17:32.400 --> 00:17:36.160
What what's what's your uh your favorite part about about leading product?
00:17:36.559 --> 00:17:52.000
I mean, definitely once you launch a product um and a customer uses it and and they their mind is blown, that's sort of my favorite part of getting the product actually in customers' hands and finding that it has changed somebody's life, right?
00:17:52.319 --> 00:18:04.960
Um, and then more recently, you know, normally, like if I have an idea, then I would sort of ping someone on my team, like, hey, let's get together a small pod to prototype this.
00:18:05.119 --> 00:18:12.880
Um, for example, like I wanted to see how price optimization recommendations can look like, like, what would that workflow be like?
00:18:13.119 --> 00:18:24.319
Um, and you know, I sort of just enlisted to our GitHub branch and had Claude Code and just asked Claude Claude to you know generate a workflow for me for price optimization.
00:18:24.400 --> 00:18:35.279
And I was able to send that prototype to my team, and it just felt so empowering after a long time, maybe because you know I started in engineering and um and yeah, that was fantastic.
00:18:35.680 --> 00:18:37.519
That that is that is very, very cool.
00:18:37.680 --> 00:18:41.039
It's so much uh so much fun um to have that power.
00:18:41.200 --> 00:18:57.279
I I have a number of um friends who are technical and non-technical who are doing so many amazing things now with AI, uh just with natural language and controlling it, working with working with code, GitHub, all these things, uh, even though they would never have worked with it before.
00:18:57.359 --> 00:19:04.799
And you and I have an engineering background, so we've seen that we sort of could play there, but we we didn't because we're in other roles.
00:19:04.960 --> 00:19:07.119
So um that's uh that's fascinating.
00:19:07.279 --> 00:19:11.839
So with AI as an industry, as we're as we're winding down, Priya, uh, you know, a couple questions.
00:19:12.000 --> 00:19:31.759
One is what do you see as sort of the future of AI with uh with the industry, whether it's with subscription-based um uh offerings or whether it's just in general, you know, AI and how it's affecting uh you know the e-commerce uh you know uh business today?
00:19:32.160 --> 00:19:46.319
Yeah, I mean I would say that um AI is sort of transforming a lot of the static customer funnels into more hyper-personalized um growth loops, if you will, for our merchants.
00:19:46.400 --> 00:19:51.599
So it's um and for people who are now builders, right?
00:19:51.680 --> 00:19:53.519
Everybody can be a builder now.
00:19:53.599 --> 00:20:02.960
Um it's sort of shifting from building traditional feature sets to like intelligent systems that can self-optimize over time, which is very exciting.
00:20:03.119 --> 00:20:13.359
Like even for us, you know, setting subscription growth on autopilot, it might find patterns that we might not have found as Recurly to offer as insight.
00:20:13.440 --> 00:20:16.319
And so I think the future is quite promising.
00:20:17.440 --> 00:20:18.880
That's that's fantastic.
00:20:19.279 --> 00:20:29.519
Last question as we're wrapping up for a business leader out there who could be a potential uh client of Recurly in the future, um, or maybe they're a current client.
00:20:29.680 --> 00:20:51.279
What what is your what is your advice to them about um how they should approach AI and how they should approach um you know um applying AI into their into their companies and so forth, whether it's with the vendor like recurly or whether it's with just a general perspective, what are you what's your advice for business leaders out there uh on the topic of AI today?
00:20:51.680 --> 00:20:58.400
Yeah, I would I would say that um don't think of adding AI into the product for just the sake of AI.
00:20:58.720 --> 00:21:02.640
Always think about what is the customer problem you're solving first.
00:21:02.799 --> 00:21:08.960
So in our case, we looked at it as, hey, retention is a problem we keep hearing from our merchants.
00:21:09.119 --> 00:21:16.480
How do we solve the retention problem for them through the data that we have access to, through workflows that we can provide them?
00:21:16.720 --> 00:21:28.720
And so if AI doesn't reduce friction or save time or deliver measurable ROI to your customers, then it's sort of just noise and slop, as may many of us have seen.
00:21:28.880 --> 00:21:35.519
Um, and then definitely look for the data mode that you might have in your business.
00:21:35.680 --> 00:21:53.039
So look at look deep into what data you have access to and then how you can translate that um to you know great outcomes for your customers, and that translation would, you know, now you can do that easily with uh a lot of the Gen AI capabilities.
00:21:53.279 --> 00:22:03.519
And so just again, like not doing AI for AI's sake, I would also say that um trust in all of this is a big thing.
00:22:03.680 --> 00:22:12.079
Um, and so you have to also balance you know automation with your um end customer trust and do they have controls?
00:22:12.240 --> 00:22:22.319
Um, some of our merchants ask us, I want to turn off compass as a whole because of you know some regulatory um um industry that I'm in.
00:22:22.400 --> 00:22:27.759
And so we we are building with that in mind, like auditability and trust controls.
00:22:27.839 --> 00:22:30.079
Um, and that is very important as well.
00:22:30.480 --> 00:22:32.000
Yeah, that's fascinating.
00:22:32.240 --> 00:22:41.599
Priya, thank you so much for all your insights, your your time, your your energy, and your attention, just for sharing your your passions and your experience with uh with our audience today.
00:22:41.759 --> 00:22:42.559
Really appreciate it.
00:22:42.640 --> 00:22:46.880
It's so so great to uh to uh to have you on the podcast today.
00:22:47.200 --> 00:22:48.720
Yeah, thank you so much, Matt.
00:22:48.799 --> 00:22:52.720
Um, I really enjoyed this conversation and thank you for having me here.
00:00:04.799 --> 00:00:16.079
Welcome to the AI Accelerator Podcast, where we get past the hype and into what actually works, practical insights and real-world use cases you can apply in your career and your company.
00:00:16.320 --> 00:00:19.359
Today's guest isn't just talking about agentic AI.
00:00:19.519 --> 00:00:21.280
She's shipping it at scale.
00:00:21.600 --> 00:00:34.719
Priya Lakshmi Narayanin is the chief product officer at Recurly, which is the subscription platform behind brands like Paramount Plus and Twitch, serving over 76 million subscribers.
00:00:34.960 --> 00:00:46.399
She spent 25 years building product at Microsoft, PayPal, Meta, and Brex, and now leads Recurly Compass, which is an AI engine that doesn't just analyze the business, it takes action.
00:00:46.719 --> 00:00:48.479
Priya, welcome to the show.
00:00:49.280 --> 00:00:50.560
Thank you so much, Matt.
00:00:50.880 --> 00:00:55.439
It's great to be here and talking about AI and acceleration.
00:00:55.920 --> 00:00:58.479
I'm the chief product officer at Recurly.
00:00:58.640 --> 00:01:02.479
Recurly is a subscription management and recording billing platform.
00:01:02.719 --> 00:01:11.760
We help businesses, like you mentioned, household names, you know, Paramount, Twitch, et cetera, handle the complexities of the subscription business model.
00:01:12.000 --> 00:01:21.200
Our mission right now is to set subscription growth on autopilot, utilizing AI technology, agentic workflows, et cetera, for our merchants.
00:01:22.959 --> 00:01:27.760
So let's walk through your progression because your your resume is spectacular.
00:01:27.920 --> 00:01:33.760
So you've built Cloud and Microsoft and PayPal, Meta, these big names, BREX, right?
00:01:33.840 --> 00:01:35.680
And now you're CPO at Recurly.
00:01:35.840 --> 00:01:45.840
So walk us through the journey of how your career progressed and now and what pulled you into the subscription economy with a lot of things you know unfolding in the tech world today.
00:01:46.239 --> 00:01:47.120
Yeah, absolutely.
00:01:47.280 --> 00:01:50.640
Um believe it or not, I started as an intern at Microsoft.
00:01:50.799 --> 00:02:00.879
Uh, you know, those days they they would pamper us quite a bit with uh goodies and trips to Redmond, and I drank the Kool-Aid and didn't look back for 12 years.
00:02:01.040 --> 00:02:10.000
It was a fantastic ride at Microsoft where I started out as an engineer and then early days of Azure, I moved into product management.
00:02:10.319 --> 00:02:16.800
They wanted somebody with a technical background to help our customers sort of migrate onto the Azure platform.
00:02:16.960 --> 00:02:33.120
Um, and then you know, went to other enterprise companies after that, EMC built storage products at EMC, um, based on hybrid clouds connecting on-premise storage to you know other clouds like um AWS and Azure, etc.
00:02:33.520 --> 00:02:41.520
Um, and then uh went to PayPal for worked on FinTech there, and then Meta worked on commerce.
00:02:41.680 --> 00:02:47.280
So early days of COVID, commerce was just uh online commerce was going through the roof.
00:02:47.360 --> 00:02:56.800
So it was a fantastic experience at Meta building um trust and safety products for commerce, so ensuring that buyers on you know Facebook, Instagram, etc.
00:02:57.120 --> 00:03:03.199
Um, can truly trust that we are selling the right products, they're engaging with the right sellers.
00:03:03.439 --> 00:03:14.879
Um, and then after that, moved to BREX to work on um expense management and corporate credit card products, um, really got you know into it.
00:03:14.960 --> 00:03:17.120
Was I I would say early days of AI.
00:03:17.199 --> 00:03:20.000
Um, November 22 is when Chat GPT came out.
00:03:20.159 --> 00:03:22.240
Uh 22 is when I joined BRECS.
00:03:22.400 --> 00:03:34.080
And so a lot of experimentation at BREX, uh, working on you know uh assistance and how to do continuous finance uh with uh with AI workflows, etc.
00:03:34.400 --> 00:03:41.360
And now at Recurly, truly building agentic workflows uh for our merchants for revenue optimization.
00:03:42.000 --> 00:03:43.759
Wow, there yeah, there's so many pieces there.
00:03:43.840 --> 00:03:46.719
You have such a variety of product experiences are really exciting.
00:03:46.800 --> 00:03:51.599
There's been a lot of things we can unpack here in our uh in our podcast together, Priya.
00:03:51.759 --> 00:03:52.319
I'm excited.
00:03:52.400 --> 00:03:55.840
So let's talk about Recurly for uh what you're doing there.
00:03:56.080 --> 00:04:04.560
I know that um part of what Recurly does and what you do there, leading leading uh product is is uh managing the subscriber life cycle.
00:04:04.879 --> 00:04:05.039
Right?
00:04:05.199 --> 00:04:12.560
So, what is can you explain what that means to maybe people who are outside the tech space or maybe not not in the subscriber tech space?
00:04:12.960 --> 00:04:14.159
Yeah, absolutely.
00:04:14.319 --> 00:04:22.319
Um I think everybody is so knowing or unknowingly uh um aware of subscriptions or using subscriptions.
00:04:22.399 --> 00:04:34.319
Like if you hit subscribe and save because you wanted to, you knew you would you would need that product again in two, three months on Amazon and hit that button, then you know you are a subscriber, right?
00:04:34.480 --> 00:04:39.920
Um and you have people have many subscriptions like Netflix and other streaming platforms as well.
00:04:40.160 --> 00:05:08.480
Um, and so our goal is to help merchants sort of go through the entire consumer journey of acquiring consumers who then become subscribers, um, and then you know, activating them, retaining them, and growing their subscriber base with them for a good lifetime value, um, while they provide more products and more innovation on top of their existing platform.
00:05:08.720 --> 00:05:14.399
So that's sort of what we think about as the um subscriber lifecycle or journey.
00:05:14.480 --> 00:05:28.800
So they launch new products, they acquire subscribers, um, they activate them, they then retain, they want to retain those subscribers through more innovation in the product or service, and then we they scale and grow their business.
00:05:28.959 --> 00:05:38.720
And through this entire life cycle or journey, Ricurly wants to provide the right capabilities and services so that uh we are there every step along the way.
00:05:39.920 --> 00:05:40.879
Wow, that's fantastic.
00:05:40.959 --> 00:05:42.800
So it's very much a B2B model, right?
00:05:42.879 --> 00:05:57.199
You're going after the larger uh enterprises that have uh subscription offerings, and you're helping them better manage the offerings, as you're saying, like reduce churn and and uh and uh and just you know maximize you know subscriber benefit, revenue, et cetera.
00:05:57.360 --> 00:06:12.079
Yeah, yeah, we predominantly work with merchants who are you know high volume of subscriptions, low order value, like 4099, 999, like the streaming media platforms, like I mentioned, um they accept consumer payments, so you know, direct-to-consumer services.
00:06:12.240 --> 00:06:18.240
Um, we also have quite a few B2B merchants on the platform, so sort of span the spectrum, if you will.
00:06:18.560 --> 00:06:19.040
That's fan.
00:06:19.199 --> 00:06:20.560
That's that's fascinating.
00:06:20.639 --> 00:06:33.680
Uh, not being a subscription expert myself, not having that type of experience, uh, question and curiosity I have is where in the world of AI or where where is AI making the biggest measurable difference right now?
00:06:33.759 --> 00:06:36.879
Is it more on like preventing the churn or is it winning people back?
00:06:37.040 --> 00:06:43.040
Is there another area like what what part of that life cycle is is where you're really uh making an impact?
00:06:43.439 --> 00:06:46.160
Yeah, I would say across across the board, right?
00:06:46.240 --> 00:07:02.959
So when you think of like a purchase funnel, it's like discovery, how do consumers you know discover your product when they consider it, like comparisons, so you so you make sure that they choose your product to buy um and then they make the purchase and then they stay on and retain.
00:07:03.040 --> 00:07:05.920
Um, that's sort of the best outcome.
00:07:06.079 --> 00:07:13.120
Um, so I think AI is making a difference in every step of that purchase funnel and the lifetime over the lifetime of the subscriber.
00:07:13.279 --> 00:07:22.319
Um, in Discovery, I think in a more and more AI native world, the storefront will no longer be a website that you specifically go to.
00:07:22.480 --> 00:07:25.360
Um, you know, it could be your conversational interface.
00:07:25.439 --> 00:07:31.199
It's you know Chat GPT or Claude, um, your favorite AI platform, right, of choice.
00:07:31.360 --> 00:07:40.720
Um, and you might go in and say, hey, what's the best way to watch you know the NFL playoffs this weekend or um you know the show called The Bureau on Paramount?
00:07:40.879 --> 00:07:53.920
And the agent will now act as on your behalf, like it'll go discover what's the cheapest way or the fastest way for you to get access to that content, and it'll provide a comparison engine, you know, platform A versus B, this is the price you would pay.
00:07:54.079 --> 00:07:59.600
They have a pay-per-view on Amazon, but on Paramount, you have to buy the subscription with a free trial.
00:07:59.680 --> 00:08:01.519
So it can give you all of these options.
00:08:01.680 --> 00:08:06.800
And so now discovery is through an AI storefront or AI channel, if you will.
00:08:07.120 --> 00:08:22.399
And then when you make the consideration, there's AI there as well, um, around you know, fast native checkouts, the payment methods you choose, um, you apply that, and then they once they purchase the product, they experience it, there's AI recommendations there.
00:08:22.560 --> 00:08:37.279
Um, and then you know, when you decide, okay, I've you know, I've watched this series, now I want to, you know, uh cancel my subscription, then you can uh imagine that you know Paramount would come back and say, hey, how about 50% off for the next three months?
00:08:37.440 --> 00:08:47.840
And those recommendations are also AI driven, and so so that we can know the churn propensity, like you said, and then help our merchants retain customers on the platform.
00:08:48.000 --> 00:08:50.320
So I would say sort of there's AI.
00:08:50.480 --> 00:08:53.519
I think the middle part of the funnel, there was AI for a long time.
00:08:53.679 --> 00:09:05.120
Um, now the discovery and you know, Justin's time, customization of retention offers, etc., more and more um, I would say generative AI is playing a part.
00:09:05.519 --> 00:09:07.440
Yeah, that's fascinating.
00:09:08.080 --> 00:09:18.960
There, there's uh, there's an aspect that you mentioned there about um uh you know about the uh well about discovery and people coming in, and then I guess it could happen anytime during the life cycle.
00:09:19.039 --> 00:09:22.080
I'd like to ask about uh uh failed payment revenue.
00:09:22.240 --> 00:09:23.440
So this is a big thing, right?
00:09:23.519 --> 00:09:31.919
If you're selling something, especially to millions of people, um, and I know I know that's that's part of um um uh what recurly assists with.
00:09:32.080 --> 00:09:45.440
And I think it was to the tune of like$1.6 billion, and maybe to date, or maybe it's an I don't know how what that is, but can you talk a little bit about failed payment revenue and how recurly helps with that particular side of subscriptions?
00:09:45.759 --> 00:09:47.039
Yeah, absolutely.
00:09:47.279 --> 00:09:50.639
Um we call it as intelligent retries.
00:09:50.799 --> 00:10:04.399
Um, so when a subscriber puts down you know a payment method um for you know when they purchase the subscription, and we then retry on the whatever if it's a monthly plan, monthly, or a quarterly plan, et cetera.
00:10:04.559 --> 00:10:09.759
Um, but credit card numbers change all the time, you know, fraud happens, it gets canceled.
00:10:09.919 --> 00:10:16.480
Um, a lot of times, you know, there are different ways, reasons because of which a transaction might not go through, right?
00:10:16.559 --> 00:10:20.159
It might be insufficient funds, it might be the reasons that I mentioned earlier.
00:10:20.399 --> 00:10:39.039
And so um what Rick curly helps with is we have a machine learning model that determines um how to intelligently retry the failed transactions, you know, what time of day, you know, when is payday for a particular uh subscriber of a certain age or industry they are in, etc.
00:10:39.360 --> 00:10:48.480
Um, and retry at that optimal time so that we increase the um success rate of the transaction and recover more revenue for our merchants.
00:10:48.639 --> 00:10:51.440
Um, there's also what we call as a dunning window.
00:10:51.600 --> 00:11:12.960
Um, a dunning just means that um, you know, what is a time period during which you will inform the consumer of hey, you have a payment due, you now the payment is overdue, you need to update, and so you can increasingly um you know raise the urgency for them that your subscription might get cancelled if you don't pay in now five days.
00:11:13.120 --> 00:11:17.039
Um, and so that series of communication also Ricoli helps with.
00:11:17.200 --> 00:11:29.759
And then over that time period, we do you know multiple retry attempts, which is what is sort of together is intelligent retries plus you know dunning optimizations that we offer.
00:11:30.559 --> 00:11:31.440
That's fascinating.
00:11:31.600 --> 00:11:43.919
So on the on the product side, um like about how large is the product organization there that does product development and so forth, like how how big is how big is Recurry's product organization, like number of employees, or I'm just curious.
00:11:44.320 --> 00:11:50.799
Yeah, we um we totally around you know um 350 or so employees.
00:11:50.960 --> 00:11:59.120
Uh we are spread out geographically um in uh Medellin, Colombia, in Austin, in Denver, etc.
00:11:59.360 --> 00:12:02.240
We also have quite a few employees in the Bay Area.
00:12:02.399 --> 00:12:06.720
Um the product org is about you know 11 or 12 people.
00:12:06.960 --> 00:12:15.519
Um we have designers, uh, and then we also have product owners and uh program managers, um, etc.
00:12:15.840 --> 00:12:26.799
So together we are looking at also evolving how we build products and taking that product development lifecycle to a more AI-native one.
00:12:26.879 --> 00:12:44.399
So we're going through that right now with you know building a PRD skill in Cloud that generates the PRD that automatically is then fed into you know to code generate, and then that is reviewed by a senior principal engineer, it goes into evals and then auto-deploys.
00:12:44.480 --> 00:13:03.440
So that is sort of our dream cycle, um, where then the you know the product managers, designers, product owners, et cetera, can then be more customer focused, can look at you know um new business opportunities, how do we um test things more often and iterate more?
00:13:03.679 --> 00:13:05.440
So it's it's an exciting time.
00:13:05.759 --> 00:13:06.399
That's beautiful.
00:13:06.559 --> 00:13:11.200
You are you started to answer my next question, which is like what type of AI models do you use internally?
00:13:11.360 --> 00:13:16.559
Uh, are you starting to notice benefit um from that from a product development perspective?
00:13:16.639 --> 00:13:27.679
Um, and I guess related to that is like what type of uh data uh do you collect from the marketplace from your clients, et cetera, to sort of feed that product development lifecycle?
00:13:27.759 --> 00:13:31.279
Because I know AI is great at looking across a lot of data, especially complex data.
00:13:31.360 --> 00:13:35.440
I'm just curious, sort of big picture from a product development uh cycle.
00:13:35.679 --> 00:13:41.440
I know you mentioned you're using Claude, uh, you know, Claud Code, which is you know super powerful uh model.
00:13:41.759 --> 00:13:46.559
Um can you talk a little bit about like what type of data are you collecting in so you can improve the product?
00:13:46.639 --> 00:14:02.639
So sort of the learning engine that you're building in between humans and AI, obviously, it's not just all AI, but just like how do you um how do you approach the uh data collection and then um you know product molding the clay of the next generation of the product?
00:14:03.039 --> 00:14:04.240
Yeah, absolutely.
00:14:04.399 --> 00:14:25.200
Um, I mean, we of course track adoption and usage data on we have a merchant-facing product, so you know, admins who come in and create new subscription plants, uh, pricing, set pricing for their subscriptions, create coupons, create offers, et cetera, um, that then they launch.
00:14:25.279 --> 00:14:30.080
So we sort of have you, you know, usage data on how they're using our platform itself.
00:14:30.159 --> 00:14:36.799
And then we have um a lot of data around the actual payment that we capture for the merchants, right?
00:14:36.960 --> 00:14:48.559
Transaction data that you know we then use to determine um, you know, the retry timing and better revenue recovery for our merchants based on that as well.
00:14:48.799 --> 00:14:55.200
And so the the key thing is it's not just um one merchant, we have many merchants in a vertical.
00:14:55.279 --> 00:15:11.679
So based on you know, vertical specific, we could say digital media and streaming, um, you know, in addition to uh you know monthly plans, you can offer like weekend passes for certain events, and you will do better because you will get you will acquire net new subscribers.
00:15:11.759 --> 00:15:20.000
So we are able to provide insights like that to our merchants that then help them grow their revenue as well as acquire more subscribers.
00:15:20.159 --> 00:15:41.200
Um, and then on the churn side, um we also have uh a product called Recurly Engage, which um which allows us to on the subscriber side, allows our merchants to deploy um what we call as prompts uh based on actions that the subscribers are taking on their app or website or in the TV app, etc.
00:15:41.519 --> 00:15:51.840
So as they're about to cancel, we can track that and then present an offer based on you know how long the subscriber has stayed on the platform, how much do they use the product today.
00:15:52.080 --> 00:15:55.360
Um, and that then helps with churn mitigation.
00:15:55.919 --> 00:16:03.519
Wow, there's so you you're so uh I'm sure I'm sure your team is probably two, three, five times more effective than they would be without AI.
00:16:03.600 --> 00:16:07.519
Like without AI, it's just so much, everything you just shared, there's so much involved with that.
00:16:07.600 --> 00:16:11.200
And it's it's only so much the human mind can can cover, right?
00:16:11.440 --> 00:16:15.279
So um yeah, I mean, even the even designing an offer, right?
00:16:15.360 --> 00:16:20.480
Like I'm sure you must have seen some of those offers yourselves, you know, based on uh the apps you use.
00:16:20.639 --> 00:16:22.960
Like I use an app called Better Me.
00:16:23.039 --> 00:16:31.519
It's like a health and fitness app, um, where I, you know, in the everyday with small mini workouts, you could do whenever you can fit in the time.
00:16:31.759 --> 00:16:35.600
And I didn't, because I was traveling for work, I didn't used it for a little while.
00:16:35.759 --> 00:16:40.000
And it, you know, it sort of showed me a prompt saying, Hey, you haven't used it in a little while.
00:16:40.080 --> 00:16:42.320
Like, do you want you know X dollars off?
00:16:42.399 --> 00:16:48.799
Or um, so those prompts now we can help our merchants design those with the help of AI.
00:16:48.879 --> 00:17:02.799
So we've done an integration with Figma and our compass assistant, where you can, using natural language, say, Hey, I want you to increase the tone, you know, more friendly or casual, things like that, where you don't need a whole copy team to go do it.
00:17:02.879 --> 00:17:05.839
You can try different iterations, you can run tests.
00:17:06.000 --> 00:17:12.240
Um, so it's uh yeah, we we we we are now finding a lot of uses for how we can integrate.
00:17:12.640 --> 00:17:13.680
That's fascinating.
00:17:13.839 --> 00:17:16.240
What what's your favorite part, Priya, about your role?
00:17:16.319 --> 00:17:22.960
Because you have you had a number of leadership roles across different uh companies, and obviously um you're where you're at now for a reason.
00:17:23.039 --> 00:17:24.079
What's your favorite part?
00:17:24.240 --> 00:17:27.119
Is it is it working with the with the team internally?
00:17:27.200 --> 00:17:28.480
Is it designing the new products?
00:17:28.559 --> 00:17:29.920
Is it the creativity involved with that?
00:17:30.000 --> 00:17:32.240
Is it more of interacting with the clients, I guess?
00:17:32.400 --> 00:17:36.160
What what's what's your uh your favorite part about about leading product?
00:17:36.559 --> 00:17:52.000
I mean, definitely once you launch a product um and a customer uses it and and they their mind is blown, that's sort of my favorite part of getting the product actually in customers' hands and finding that it has changed somebody's life, right?
00:17:52.319 --> 00:18:04.960
Um, and then more recently, you know, normally, like if I have an idea, then I would sort of ping someone on my team, like, hey, let's get together a small pod to prototype this.
00:18:05.119 --> 00:18:12.880
Um, for example, like I wanted to see how price optimization recommendations can look like, like, what would that workflow be like?
00:18:13.119 --> 00:18:24.319
Um, and you know, I sort of just enlisted to our GitHub branch and had Claude Code and just asked Claude Claude to you know generate a workflow for me for price optimization.
00:18:24.400 --> 00:18:35.279
And I was able to send that prototype to my team, and it just felt so empowering after a long time, maybe because you know I started in engineering and um and yeah, that was fantastic.
00:18:35.680 --> 00:18:37.519
That that is that is very, very cool.
00:18:37.680 --> 00:18:41.039
It's so much uh so much fun um to have that power.
00:18:41.200 --> 00:18:57.279
I I have a number of um friends who are technical and non-technical who are doing so many amazing things now with AI, uh just with natural language and controlling it, working with working with code, GitHub, all these things, uh, even though they would never have worked with it before.
00:18:57.359 --> 00:19:04.799
And you and I have an engineering background, so we've seen that we sort of could play there, but we we didn't because we're in other roles.
00:19:04.960 --> 00:19:07.119
So um that's uh that's fascinating.
00:19:07.279 --> 00:19:11.839
So with AI as an industry, as we're as we're winding down, Priya, uh, you know, a couple questions.
00:19:12.000 --> 00:19:31.759
One is what do you see as sort of the future of AI with uh with the industry, whether it's with subscription-based um uh offerings or whether it's just in general, you know, AI and how it's affecting uh you know the e-commerce uh you know uh business today?
00:19:32.160 --> 00:19:46.319
Yeah, I mean I would say that um AI is sort of transforming a lot of the static customer funnels into more hyper-personalized um growth loops, if you will, for our merchants.
00:19:46.400 --> 00:19:51.599
So it's um and for people who are now builders, right?
00:19:51.680 --> 00:19:53.519
Everybody can be a builder now.
00:19:53.599 --> 00:20:02.960
Um it's sort of shifting from building traditional feature sets to like intelligent systems that can self-optimize over time, which is very exciting.
00:20:03.119 --> 00:20:13.359
Like even for us, you know, setting subscription growth on autopilot, it might find patterns that we might not have found as Recurly to offer as insight.
00:20:13.440 --> 00:20:16.319
And so I think the future is quite promising.
00:20:17.440 --> 00:20:18.880
That's that's fantastic.
00:20:19.279 --> 00:20:29.519
Last question as we're wrapping up for a business leader out there who could be a potential uh client of Recurly in the future, um, or maybe they're a current client.
00:20:29.680 --> 00:20:51.279
What what is your what is your advice to them about um how they should approach AI and how they should approach um you know um applying AI into their into their companies and so forth, whether it's with the vendor like recurly or whether it's with just a general perspective, what are you what's your advice for business leaders out there uh on the topic of AI today?
00:20:51.680 --> 00:20:58.400
Yeah, I would I would say that um don't think of adding AI into the product for just the sake of AI.
00:20:58.720 --> 00:21:02.640
Always think about what is the customer problem you're solving first.
00:21:02.799 --> 00:21:08.960
So in our case, we looked at it as, hey, retention is a problem we keep hearing from our merchants.
00:21:09.119 --> 00:21:16.480
How do we solve the retention problem for them through the data that we have access to, through workflows that we can provide them?
00:21:16.720 --> 00:21:28.720
And so if AI doesn't reduce friction or save time or deliver measurable ROI to your customers, then it's sort of just noise and slop, as may many of us have seen.
00:21:28.880 --> 00:21:35.519
Um, and then definitely look for the data mode that you might have in your business.
00:21:35.680 --> 00:21:53.039
So look at look deep into what data you have access to and then how you can translate that um to you know great outcomes for your customers, and that translation would, you know, now you can do that easily with uh a lot of the Gen AI capabilities.
00:21:53.279 --> 00:22:03.519
And so just again, like not doing AI for AI's sake, I would also say that um trust in all of this is a big thing.
00:22:03.680 --> 00:22:12.079
Um, and so you have to also balance you know automation with your um end customer trust and do they have controls?
00:22:12.240 --> 00:22:22.319
Um, some of our merchants ask us, I want to turn off compass as a whole because of you know some regulatory um um industry that I'm in.
00:22:22.400 --> 00:22:27.759
And so we we are building with that in mind, like auditability and trust controls.
00:22:27.839 --> 00:22:30.079
Um, and that is very important as well.
00:22:30.480 --> 00:22:32.000
Yeah, that's fascinating.
00:22:32.240 --> 00:22:41.599
Priya, thank you so much for all your insights, your your time, your your energy, and your attention, just for sharing your your passions and your experience with uh with our audience today.
00:22:41.759 --> 00:22:42.559
Really appreciate it.
00:22:42.640 --> 00:22:46.880
It's so so great to uh to uh to have you on the podcast today.
00:22:47.200 --> 00:22:48.720
Yeah, thank you so much, Matt.
00:22:48.799 --> 00:22:52.720
Um, I really enjoyed this conversation and thank you for having me here.
