Bonus Episode: Leveraging AI While Leading with Vision with Nathan Snell

Nathan Snell is a founder, executive, and three-time entrepreneur best known for co-founding nCino, the global leader in cloud banking, and now Raleon, an AI retention platform for DTC brands. With 15+ years in fintech, marketing technology, and AI, Nathan has built products that don’t just scale companies, they transform entire markets, including a 10-figure exit.

With Raleon, Nathan is reimagining retention for Ecommerce. Instead of bloated teams or endless manual work, Raleon acts like a teammate, helping DTC brands and agencies handle retention 50% faster while driving more revenue. From campaign planning to segmentation, it automates the tactical grind so marketers can focus on strategy and growth.

Nathan’s story blends technical expertise with market-shaping vision. From scaling nCino into a public company, to investing as an active angel, to now tackling one of Ecommerce’s biggest pain points, retention, he’s seen how AI can accelerate results but still requires human pilots to go beyond “average” output.

Whether you’re building a lean DTC team, rethinking retention marketing, or trying to cut through the hype of AI, Nathan offers a grounded look at how to combine automation, brand taste, and strategy to drive the next era of Ecommerce growth.

In This Conversation We Discuss: 

• [00:22] Intro
• [00:46] Building expertise in workflow automation
• [01:46] Experimenting with LLMs in workflows
• [03:18] Comparing AI models for DTC marketing
• [04:01] Starting email AI with copywriting
• [05:08] Fine-tuning prompts for better outputs
• [06:40] Elevating outputs with better context setting
• [08:08] Analyzing past campaigns to guide outputs
• [09:50] Stay updated with new episodes
• [10:02] Automating segmentation and copy at once
• [11:57] Recognizing AI delivers average by default
• [13:38] Editing outputs instead of chasing perfect prompts
• [15:20] Connecting Klaviyo and Shopify for campaigns
• [17:41] Automating learning cycles across campaigns
• [19:14] Guiding systems instead of replacing teams

Resources:

• Subscribe to Honest Ecommerce on Youtube
• Automate DTC retention marketing with AI raleon.io/ 
• Follow Nathan Snell linkedin.com/in/nathansnell 

If you’re enjoying the show, we’d love it if you left Honest Ecommerce a review on Apple Podcasts. It makes a huge impact on the success of the podcast, and we love reading every one of your reviews!

Transcript

Nathan Snell

Anybody who's saying, “Oh, it's going to just completely automate everything that you're doing,’ you should just be skeptical about. And you'll never hear us say that because…”

Chase Clymer

It's the modern drop shipping mentality. It's like, “Oh, build an AI agent to sell the thing. Selling more AI agents, blah, blah, blah.” You have a Ferrari, you live on a beach.  

Nathan Snell

Yep. 

Chase Clymer

Nothing is that easy.

Chase Clymer

Honest Ecommerce is a weekly podcast where we interview direct-to-consumer brand founders and leaders to find out what it takes to start, grow, and scale an online business today.

Hey everybody, welcome back to another episode of Honest Ecommerce. Today, I'm bringing to show an extremely smart gentleman, the co-founder of Raleon, Nathan Snell. Welcome to the show. 

Nathan Snell

Thanks, Chase. Appreciate you inviting me here. 

Chase Clymer

Now I purposely, purposely buried the lead on what Raleon does. We'll get to that in a moment. But we can talk about your expertise in AI and specifically marketing and marketing automation. Quickly though, I guess, what's your background? How did you end up in this specifically? 

Nathan Snell

Yeah. So my background, as you might suspect, is all in workflow automation, AI, marketing analytics. Prior to really specifically, I was actually in the fintech space for about 15 years, doing  the same thing, just focused obviously in a different category overall.

And when I first got out of college, I was actually in the Ecomm space. So it's like, I won't date myself too much. But Magento was just the baby of a thing. So I wanted something that was a little bit different. I wanted to take our skills and expertise and myself and the team around this sort of stuff and bring it back into a space that's just fun from an entrepreneurial standpoint, the things that people are creating. 

Chase Clymer

Yeah, absolutely. And now your career has gone full circle and you're bringing these marketing automation workflows and using AI to helpback to Ecommerce. That's what we're going to talk about today in a nutshell, is how to use AI to run your brand's email marketing. So Nathan, where do we start with this? Do I just log into my favorite LLM and say, run my email? 

Nathan Snell

If only, if only. Although I guess maybe not if only because then that would make our job a little bit harder. No, I think when folks tend to think about AI and using things like LLMs for their email marketing, for marketing overall, but maybe we'll even just focus on the email side for this, oftentimes they approach from the same point of, “Okay, well, I've got all these things that I want to have happen, that I want done.” And they think too that they have to sort of transform their entire process overall. 

And I think contrary to say like last cycle of technology, I mean, oftentimes the easiest starting point is really kind of looking at your process as it is today and not sort of blowing the whole thing up, but instead going, “Okay, well, if I'm doing, you know, if I'm writing copy for my email, like, you know, how do I go about doing that in a more efficient way?” Right. 

“And then how do I go about making sure that copy is more on brand and matches our tone more?” So I'd say, “You know, the first point that we always think about, and we always even talk to brands about is just like, look at your process as it is today and pick. One part of it or two parts of it and begin to use and experiment with an LLM just as a simple starting point in that particular part of the process.”  

Chase Clymer

Absolutely. And then I'm just going to apologize in advance to all of our listeners.  I'm just going to default to saying ChatGPT when we're talking about AI, probably for the rest of this episode. But there are a bunch of them out there. That's just the one I use.  

Nathan Snell

Yeah. And I would say too, if it's helpful, we can. I don't know if in the show notes or at the end, but we actually just released a DTC AI benchmark where based on the things like the activities that founders and CMOs and stuff, the different brands do, we actually automatically grade  basically all of the core models. 

So whether that's ChatGPT,  Anthropic, Claude, Google's Gemini. So we run it through a pretty rigorous rubric. So we actually have a really good idea of which models even perform well for which task. Because different tasks work better for different ones. 

Chase Clymer

We are a brand. We want to use AI to power our email. We're going to start tackling smaller chunks of our process to take over parts of our email? Is it as easy as, I'm writing a campaign about XYZ and help me with my first draft? Or how should people start really thinking about this? 

Nathan Snell

So yeah, I mean, in terms of a starting point, we usually recommend folks start in two places. So one, and I mentioned before, is the copywriting side. And what we tend to find is that that's an area where it's easy to get started, but it's actually hard to master.

So, generally when people jump in on the copywriting piece, they see say a 25, maybe that's even too aggressive lift at the outset. And what we tend to find is when you refine that with agentic systems and tools, that honestly goes from say 15 to 20% kind of improvement to say 90%, right? 

Where people are really going, “Oh my gosh, this thing is so on brand. It's nailed the problems, the problem, the solution language, right? It's got our tone and voice, the words to avoid.” All that sort of stuff. So that's kind of why we say that's a good starting point.

Chase Clymer

When you're saying train it, if we're thinking about email marketing, and we're training it, it's delivering to this AI system, historical email campaigns or copy that's written in the brand voice. The brand guidelines probably would help tremendously as well. So just doing that upfront work will get you to that better result that you're talking about? 

Nathan Snell

It'll help. What we see within our platform, and we cover this much more broadly, which we can talk through some of the other parts of the process you can work through. But if you're doing it just using a straight ChatGPT, more context is going to be helpful. So brand voice is going to be really helpful for sure. Past emails are going to help there some.  

But there's also more fine tuning to that prompt overall. That's going to help in terms of just guiding it towards the kinds of things that it's thinking about. What we see is, it's actually, I find it kind of ironic where LLMs tend to do a pretty poor job without a lot of good prompt engineering and even multi-step prompts. They tend to do a pretty lousy job with B2C emails, with direct-to-consumer emails because  they're not often text-typing. It's more about the layout. It's more about the theme and the purpose of that email.

So oftentimes, what we actually see working really well is using multiple prompts. So you might tie it up with that brand context that you were talking about and an initial starting point. And then actually take the output of that and move it into another prompt that's more guided and everything else. So that's some of what we actually even do within our system for that particular part of the process. 

Chase Clymer

Yeah. I mean, it's truly with AI, it's made things  easier. But I won't say it's made things better because it's just. It's shit in, it's shit out. And I forget who told me this, but it's basically AI is an amalgamation of all of the stuff that exists. So it's going to be, at best, average. So if you aren't tailoring it towards a better solution with your prompt, you're going to get an average result. 

Nathan Snell

Yeah, that's true. It's definitely true. I think one of the things that we're seeing more and more is,  especially when it comes to the prompt side, I mean people talk about prompt engineering.

Chase Clymer

That's my favorite made up term.

Nathan Snell

It's true. I mean, honestly,  it is one of those terms where you're like, “Really? Is that really a thing?”

Chase Clymer

I know some people are better at it than I am, so I could give them.

Nathan Snell

Honestly, it's been a craft that we've obviously tuned over time. And what we see, even the better at part, I've been thinking about it a lot like Photoshop, or like a Figma, that sort of thing, where the tool's there, but if a designer jumps into Photoshop or Figma, what they're going to produce is going to be a lot better than what I can produce.

And it's funny because as easy as LLM seems, to your point, it's like garbage in, garbage out. So there is an element now that we're seeing more and more where it's like your ability to actually understand the right context that works well to tune the prompt and everything else. It's like having a designer that's really good at Photoshop versus Nathan jumping in who can probably put text on an image and that's about it. 

Chase Clymer

Absolutely. So it is a lot about the inputs. And obviously, there's the historical inputs of the brand voice and everything associated with that historical stuff like that. But it's also how you're asking for that result. So that's more of the campaign strategy. But another strategy that you and I talked about before, that I think that a lot of people overlook with AI is just, it can help you come up with just more emails to send. 

Nathan Snell

Yeah, absolutely. It's actually very. Again, especially with the right context and that feedback loop of what's working and what's not working. It does an amazing job at planning emails.

And even using it, if you've tied it up properly, for it to be able to understand your brand and the kinds of events and things that are important to your brand, like what's been working well for your brand. So it actually does a phenomenal job at being able to even analyze past performance. And again, it's not joining one single prompt, but when set up properly, it can really do a great job planning, not even just like one email, but 20, 30 emails kind of planning the month overall, or even suggesting kinds of particular emails for different slices of your audience. 

You know, I think we're really, you know, we're entering into a point in time now, I think where historically it's been very difficult to produce a lot of emails and the design side, I think it still has some of that, you know, but historically a lot of brands, you know, they focus on like the 30, 60, 90 day, right? 

It's just like a batch blast of stuff that goes out. And it's because it's been hard and expensive to say take 20% of that segment really well and give them a message they care about. But in reality, it's becoming easier and easier because of AI to be able to produce really high quality emails that are on message for a particular segment that as a result are going to perform better. 

Chase Clymer

Hey everybody, just a quick reminder. Please like this video and subscribe if you haven't. We're releasing interviews like this every week, so don't miss out. Now back to the interview.

Yeah. And I mean, if we just look at the larger slice of the pie and I'm sure a larger slice of my audience, we're talking about  roughly million dollar brands. Maybe a little smaller, maybe a little bigger. There's a lot of them out there. And this is a founder-led industry where a founder is taking time out of their day to produce that weekly email campaign. 

And maybe they're bringing in an AI assistant type thing to help them do it. But I've seen what your software can do. You can even do this without that software. You can take that struggle of hitting 4 campaigns a week  to hitting 30 emails a month, and it will take you less time. 

Nathan Snell

Yeah, absolutely. We certainly see that. There was a brand that just started using us. They're a 7-figure brand. And it was a couple of weeks ago. Very much the founder led to your point. And he jumped into the product.

And I think the hard thing too, oftentimes is AI is so overhyped. So like we try to very much like even just undersell it. So you have to like the product or like, “Look, like just try it out.” You use it yourself and you're going to see whether or not you like it and whether it's on brand and everything else. So he did. And literally the next day he was like, he was blown away. And it's because to your point, he went, “Oh my gosh, like usually it's hard for me to get kind of one email out a week.” And, you know, within a day he had four emails planned for the next week.

He had it over to his designer, like his designer designed them that evening. All the copies were in it. All the segmentation was set and everything else. So he literally went from the same time it was taking him to figure out one email per week. 

He's now getting four emails per week out to his team with all the segments that he needs fully targeted and everything else. And it's literally taking them no more time. And it's actually higher quality because it can, it's able to produce all the sort of thought and automated analysis that he didn't have time to do to begin with. And I think that's some of the magic that's starting to happen with this stuff.

Chase Clymer

And before we get into talking about the super cool things you can do with AI and particularly Raleon, I do want to take a step back though. It's like AI isn't the end-all be-all. What should founders be skeptical of when it comes to AI? Because there's just so much noise out there. 

Nathan Snell

There is. Yeah. I think sometimes we get over-indexed on how simple it feels. I think anybody who's saying, “Oh, it's going to just completely automate everything that you're doing, you should just be skeptical of that.” And you'll never hear us say that because.

Chase Clymer

It's the modern drop shipping  mentality. It's like building an AI agent to sell the thing. It's selling more AI agents. You have a Ferrari, live on a beach. Like, nothing is that easy. 

Nathan Snell

It's true. It's just not. And I think taste and seniority, I think, is becoming more and more valuable to this because it does help you do things faster. But I think, the example that you gave before,  it's really good at giving you average which means that like, if you don't know what above average looks like, or you don't have that sort of above average taste or sense, like it's not going to do a great job helping you get above that sort of average point. 

And I think that's where I can, you see, you know, you see brands that are using it that have phenomenal results. And it's because they've got sort of the experience, the taste, the seniority to figure out not how to just get to that sort of 50% mark quickly, but how to actually then take it to that 80, 90% mark. That's really positive for the brand and everything else. And that's not, it's just not automated. 

I think we get wrapped around the idea like, “Oh my gosh, 90% of our job is done.” I was like, “Well, it can get you there. And it can accelerate it. But again, it's not full autopilot here. There's still actually a lot of piloting from  people that are really great at what they do to get it to the results that they're going to want.”

Chase Clymer

Yeah. And I think that something that  I still do and I think a lot of marketers do is like, just use these AI platforms to just get me either from 0 to 1 or I'll get pretty close and then I'll pull it out and I'll tune it myself because I know this is close enough. I think a lot of people just keep trying to prompt it to give them a perfect result. It's almost there. Just take it and make the changes yourself. 

Nathan Snell

That's true. 

Chase Clymer

You still have that capacity to do it yourself. 

Nathan Snell

Yep. Agreed. And I think some of that is because most AI platforms today don't give you that live editing ability. We can get into it later. But that's part of why we put some of that into our product is because you're exactly right. It's like the goal is like, we're shooting for 100%, right? 

But in reality, it's going to get you to 80 or 90%. And then when you have that experience, you can go, “Oh, cool. Like, this title is great. I want to change this one word.” Instead of fighting the AI to hit that one word just right, you can just go.

Chase Clymer

Yeah, because it changes the headline and does exactly what you want and then changes the copy. The other stuff was great. Like, now you're copying and pasting things from different parts of the conversation and.

Nathan Snell

Yep. There is that frustrating challenge where like the conversation gets too long and then it's like, you've lost the thread. You've lost what we're doing here. Change it back. But I think the other place that I'd say too, and we're seeing this is design wise, I think design is still a bit overhyped. 

You know, we see good results from it and I think it'll get there. I mean, honestly, like we've got this quarterly cycle that we're seeing where new models are being dropped. So in six months, it'll probably be a lot better, but I think even the design side is probably not as far as people think. 

Chase Clymer

No. 

Nathan Snell

It's easy to spit out something for LinkedIn. But then when you're really drilling the details and you've got a product and you want to make sure that product looks exactly right and everything else, it's really tough to get it there. It just is so. 

Chase Clymer 

Yeah. I think that we're a little off when it comes to designing emails. And still, to this day, everybody, fully designed picture emails are bad. You need your text to not live in a picture for various reasons. Just don't do that.  

Nathan Snell

At least follow the 80-20 rule. At least follow the 80-20 rule. 

Chase Clymer

We've been pushing off talking about Raleon for too long. I met Nathan a little over a month ago. And he, in my opinion, has one of the best tools in the Ecommerce space that's using AI to solve a very specific problem the right way. And it's not a glorified ChatGPT wrapper. 

Nathan Snell

Appreciate it.  

Chase Clymer

But yeah, let's talk about how brands can use products like Raleon and have it run their email marketing and just make their lives easier. 

Nathan Snell

Yeah. I mean, honestly, you described what we do really well. It's that. So we think of it as like an AI retention team. So it's an extension of the internal team. And what we've found in talking with a lot of founders, and even agencies, is folks more and more are moving away from this point solution. I'm like, “Oh, I just want this thing that does this little piece like, say, copywriting.” Because at the end of the day, none of us have enough time anyway to grow our businesses. So it's sort of a, “Hey, how do we get the end-to-end job done?”

So that's what we focused on from an email marketing standpoint. So it's everything from campaign planning. So it's hooked into your Klaviyo, it's hooked into your Shopify, and it actually has our own AI segmentation and stats and everything about it. So when you do something, when you ask for something like, hey, find me my most overstocked product and come up with five campaign ideas that I can run next week that don't involve any promotions, you can literally give Raleon that and it will look it up for you. It'll see what past performance,  what past campaigns worked really well.  

It'll pick that particular product and give you the suggestion. It'll put together the campaigns overall. uses our AI segmentation to even suggest the segments to send it to. So it does a lot of this whole analysis process that takes hours to do the analysis, to figure out the plans. It writes the copy for you. It puts it in the same email layout for you. And then depending on the sophistication of the email, it'll actually generate the email design for you and load it into Klaviyo. 

Really, we're trying to build out that true end-to-end process that lets you go from an email a week or an email a month to four a week with effectively the same amount of time spent, but even better results because of everything that's built in. And it learns and it does the learning for you. 

Chase Clymer

Back to the start of the conversation, that whole workflow automation part of it, it's yes, you can do this with any AI you've access to roughly the same stuff, but you are going to be copying and pasting stuff from A to B, you're going to be spending more time doing those things that have been automated in this process. 

But also training the AI every time you open your browser or reset your browser to do the thing again is just going to be sunk costs with every time you're doing this. Whereas I think the investment versus the result that you see from it's there. It makes sense to me. I see the value. 

Nathan Snell

Yeah, absolutely. And I think what we're doing, like what we're doing on Raleon, in most AI companies, it's not to say that you can't sort of do it yourself, but it's the same way with anything, right? I mean, like, yes, you could build it yourself the same way that you could build a subscription app yourself, the same way that you could build a post purchase app yourself or, you know, or a pop-up app and so on. 

Or, you know, you can leverage sort of the benefit of folks that have been there before and really know what they're doing and are constantly sort of, you know, using, especially this cutting edge technology across many different surface areas and brands to go.

Here's how we can create sort of an automated learning cycle. That way, instead of, you know, founders having to figure out how do I do this for myself? What's the right prompts? How do I like to pull in this past detail and get it in properly so it understands it? 

You know, the system sort of constantly does that itself anyway, and constantly gets improved overall. And as a result, you know, you're able to kind of focus on what you really want, which isn't sending emails or creating emails to send, it's growing the business and you know, that's what we want folks to be able to do.

Chase Clymer

Absolutely. Yeah, that makes a lot of sense. Now, Nathan, is there anything I didn't ask you about today that you think would resonate with our audience? 

Nathan Snell

Yeah. I think one of the things that we hear often that we maybe didn't talk a lot about is there's a lot of folks that I think get scared that AI is going to replace them, whether that's on the agency side or whether that's on the brand side. And I think even a lot of brands at times  are like, “Oh my gosh, well, there's all this AI stuff. I'm just going to go replace my agency.” 

I'd be like, “Don’t burn all the bridges at the same time.” There's an element of, like we were saying towards the beginning, there still needs to be pilots for this stuff. When you think about what AI is doing, it's handling the tactical stuff. And more and more, it's going to handle the tactical elements of  finding the right segment, doing the analysis, writing the copy, and so on. But there's still a strategy that has to be present there. 

And I think agencies, you know, I mean, even senior marketers and things like that are still really valuable to guide that overall strategy. So I think the one thing I'd say is, you know, like, there be it? Will there be an impact?  You know, it is often a question of like, “Oh, these jobs are to be lost. Like, is there going to be sort of an impact there?”

Absolutely. Are we seeing that even with the brands that use us and the agencies that use us where they're hiring less because of us? 100 percent. But they still need that strategic person. They still need people on the team to guide the systems. That applies to Raleon and to whatever other additional AI tools that start to get rolled out across the board. So that's maybe one thing I'd add there. 

Chase Clymer

Now, if people listen to this episode and they're like, this sounds great. Where do they go to learn more? What do they do? 

Nathan Snell

Yeah, I mean, they can go to our website, Raleon.io. I'd say if they mention you, we're even happy to give them a little bit of discount benefit or run through a special onboarding. So I'd say come to the site or hit me up on LinkedIn or on Twitter. I respond to all that stuff well, too.

are the 3 core places that you can find us. 

Chase Clymer

Awesome. Now we'll make sure to link to those things in the show notes along with that list  of large language models that you guys have ranked for certain strengths and weaknesses. Nathan, I can't thank you enough for coming on the show today. 

Nathan Snell

Chase, man. Thanks for having me.