31:18 I want to go back to the first half
31:21 of what you're mentioning as well,
31:23 because I think that's so important.
31:24 It's been the biggest topic I've been speaking on
31:27 that mostly your AI strategy
31:30 is more of a data model problem.
31:32 So can you organize the data?
31:35 Can you build the semantic layer?
31:37 Can you build the context graph
31:38 where AI agents can actually make sense
31:40 of the data, connect the dots?
31:42 And I think it sounds like you've seen
31:45 a tremendous amount of value from that investment.
31:49 Tyler, I'm curious if you're seeing the same thing
31:51 on the go-to-market tech stack side as well,
31:56 and seeing that investment unlock productivity
31:59 on the GTMT.
32:01 Massively, actually, yeah.
32:02 And I think one thing I'll mention is,
32:04 Lalitha touched on observability,
32:07 and I think that's from my seat.
32:09 Also, when I talk to partners,
32:10 we have some partners that are fully AI-pilled,
32:13 and some of which were the other side of the scale.
32:16 And an answer to a question can never be,
32:19 that's what the agent told us, right?
32:22 We have to be able to show our work,
32:24 and yes, we want to move fast,
32:25 and they love the outcomes,
32:26 but sometimes they want to be able to tick and tie
32:30 if they really need to,
32:31 if it really comes down to it.
32:32 And so that observability is incredibly important as well.
32:36 What's happening, how do we drive growth,
32:38 and how is it impacting us?
32:39 One example I'll give is we just rebuilt this year,
32:42 just launched an entirely new MarTech stack
32:46 that allows us to really massively increase
32:50 the level of personalization
32:53 to drive much faster cycle times,
32:56 due to integrations with Snowflake,
32:58 and into our platform.
33:00 So all the data modeling,
33:01 and all the access to data
33:02 that Lolita is talking about for the risk models,
33:04 we're also feeding into,
33:05 how do we drive certain behaviors
33:07 and outcomes for our partners
33:09 across all of the brands that we have, right?
33:12 So we have more than a dozen brands that are live today.
33:17 At the end of the day though,
33:18 the number of actions is not that different,
33:21 at least in the first couple of weeks
33:22 of having a credit card, right?
33:24 I need to use my card,
33:25 I want to add it to my digital wallet,
33:27 drive my first transaction,
33:28 how do I earn and redeem rewards the first time?
33:31 But I go back to we have to make that feel,
33:33 look and feel just like it is
33:35 as if it's coming from these 100 year old brands,
33:39 where these customers have deep relationships.
33:43 And doing that at scale,
33:44 it's been a massive challenge in our history, right?
33:47 As we add new partners,
33:48 it was almost entirely manual to scale these things.
33:51 Now we can do that in a fully AI enabled platform
33:55 with access to the data,
33:57 one click to drive a whole new design implementation
34:01 in our Martech stack,
34:03 and then actually get the comms out to customers
34:06 and see the behavior change or not.
34:08 Sometimes it doesn't, right?
34:10 We wish it always did,
34:11 we wish we always got it right, but we don't.
34:13 And that's okay because that cycle time is so fast
34:17 in this new environment
34:19 that we can learn and improve these things,
34:21 but also it allows us to scale brands
34:25 and add more brands without scaling headcount
34:27 in a linear way, right?
34:28 So from a design team, from an analytics team,
34:32 from a Martech enablement and ops team,
34:35 we haven't really had to scale that headcount
34:40 as we've added new brands, especially over the past year.
34:43 So that's been a huge improvement and unlock for us
34:46 while still driving the outcomes that we want, right?
34:48 And seeing the outcomes that we need for our partners.
34:52 Well, and I think sometimes people
34:53 don't really realize the magnitude
34:56 of improving the cycle time,
34:57 improving the conversion rate of these things,
34:59 just a few points.
35:00 And if you can leverage AI to do that,
35:04 I mean, the investment beyond pays for itself
35:07 once you've done that.