Happy Sunday! 

If you’re reading this, I hope you’re excited to talk about what has been on everyone’s mind… AI. No, I’m not talking about the “AI” that Amazon was using with its walk-in/walk-out stores; I’m talking about AI models and agents that can truly replace human processes and save your business hours of human resources. I hope you’re relaxed, hydrated, and ready to read today… this is AI101.

In 10 days (June 25th, 2025), I will be co-hosting the Ecom AI Summit in New York City at Webster Hall — click here to get a free ticket (use code AINIK100)! Some of the greatest software companies in our industry will be coming to share demos, heads of AI at companies (enterprise and startups) will come and share lessons, and you’ll meet others who are doing the same in their businesses. See ya there!

Before we get into AI…

Vendor of the Week

Instant — The latest AI product to take control of CRM and supercharge revenue with generative emails.

You may remember I’ve spoken before Instant Audiences, the product that allows you to quickly turn anonymous site visitors into identifiable prospects you can email. Well, they took it another step further and just launched Instant AI.

With all the innovation within AI, there hasn’t been a lot done in the email space, until now with Instant AI. It’s been in beta and for Pushas, they saw more revenue in the past week compared to the entirety of last month with their old email platform.

So just like how we’ve seen companies start to use agents inside their products, Instant AI does the same. It constantly scans your website (new products, promotions, best sellers, etc) and uses everything it knows about the site visitor (shopping history, cart items, etc) to instantly generate an abandonment email, tailored to that person, and fires it off at the exact moment they are most likely to convert.

It is truly generative and personalized email, done at scale, and it works. You can try it out and get 50% off for a limited time on this link.

As you always hear me say with AI, do not let new tools/innovations go past you or let yourself stay in the dark with AI.

Sign up for a demo today with Instant and get 50% off!

Limited Supply

This week’s episode of Limited Supply was with Dylan Ander, a CRO master and split-testing veteran. We covered everything from user psychology to which tests are worth running on your website, to what a rebrand actually drives in terms of increased conversion rates.


You can listen to the full episode on Apple Podcasts or Spotify, or just search “Limited Supply” wherever you consume podcasts.

AI for eCommerce

What prompted me to write this email was a conversation I had with a brand, and learning about how people on their team currently leverage AI within their day-to-day operations across marketing, finance, creative, etc. There are some companies that have taken the full leap to integrate as much AI as possible, while others, like the one I was talking to, barely use it. They give their VAs a premium ChatGPT account and have them focus on simple copywriting or admin work, with the assistance of ChatGPT. 

I’m happy they’re using AI, but I believe they are thinking way too small here. In my opinion, doing that kind of work is really the bare minimum. It's sort of like saying your growth strategy is posting on Instagram twice a week. Sure, it's something, but I think most people are missing the bigger picture. 

Here's the thing about AI in DTC right now. We're in that sweet spot where it's powerful enough to move the needle, but most brands are still treating it like another fancy autocomplete tool to finish work they already started. Meanwhile, the smartest operators I know are quietly building a range of unfair advantages for their brands through more complex automations, better software, and now “AI Agents,” which is what interests me the most right now. 

I think about AI like Facebook ads in 2010. Back then, Facebook was mature enough to drive insane results, but most people were too scared to really hit the gas and go full throttle. I think this era of AI will be very similar. Just as every marketer is kicking themselves for not spending more on social ads 10 years ago, they will be kicking themselves in 2030 for not taking AI more seriously today. So with that in mind, I thought it would be a good idea to write a quick AI 101 email where I share my take, a few ways that you can use AI today, and some thoughts on where things are headed.

How AI’s Evolving: 

First of all, AI isn’t totally new. We’ve had software with AI-like features for 10+ years now.

The only thing that’s radically different is generative AI and the agents piece, which I’ll talk about more below. That stuff truly is crazy. The rest of it is similar to what we’ve seen before. Now remember, I’m a guy who doesn’t know how to code, but here is how I see the timeline of AI:

Phase 1 was about rule-based automations. If a customer buys a protein powder, you send them a shaker cup email 2 days later." It was simple, “If this, then that” logic, and it felt revolutionary at the time, but it was basically just fancy triggers and basic automations that we’ve had for many years. In reality, this is how all software has worked for a very long time. No one ever called it AI, but it was still the foundation of how this all started.

Phase 2: Predictive Models. Then, platforms started predicting who might buy, when they might churn, and what they might want next. This was still mostly behind-the-scenes statistics that required a skilled data science team. It wasn’t true generative AI like we have today. Netflix’s recommendation engine or Spotify’s radio feature, or Amazon’s “You might also like this product” are all great examples of this. These models used statistics to predict what you might want to see next. This is when people started to think the “software was smart” because the recommendations were getting so good. The social media algorithms were another version of this, where they used billions of data points to predict what post you might like to engage with next. 

Phase 3: Generative AI. This is where we are today, and this is when things have changed. Now, we have AI that can actually create and execute much more complex tasks. ChatGPT is akin to having a brilliant PhD in your pocket who never sleeps and can write, design, analyze, and strategize for you. You can prompt something and get text, images, videos, web design mockups, an audio voice clone, and more with various tools.

The big difference here? In phases 1 and 2, especially, you needed to be a company like Netflix or Amazon to really build and benefit from innovative software, as you required a substantial amount of data and engineering power to make it happen. Today, with how Agents are starting to show up, anyone can leverage AI.

Today, you can utilize OpenAI, Claude, Eleven Labs, Midjourney, or N8N as a 2-person DTC team and gain AI capabilities that rival those of major companies. And by the way, this doesn't cost tens or hundreds of thousands of dollars per year, unlike it might have for bigger companies in the past. Many of these tools start at under $50 per month, depending on your usage. It’s a massive shift that’s changing what’s possible for all brands.

Phase 4: Agents and their agents. What will come next, I think, is most big tech companies will start creating consumer-facing AI Agents. We’re already seeing it with companies like TripleWhale, where their AI agents can now help you write creative briefs based on Reddit conversations, performance marketing metrics, and what your goals as a company are.

The evolution of this phase is that these AI Agents will start within larger products, but soon be available on marketplaces. I will have a Nik AI Agent that may help you build out a brand launch strategy for $49/month. 

Where Should You Start With AI? 

In my opinion, I've seen many people sign up for multiple AI tools, use none of them effectively, and then complain that "AI doesn't work for their business." That’s like buying a gym membership and wondering why you're not jacked after changing exercises each day and only doing 5 reps each. Here's how I would approach it, and keep in mind, it’s worth bringing in an AI consultant to help with this: 

1. Audit

First, grab a coffee and list out everything your team does that makes you think "Wow, we are still spending human time on this again?!!" and focus on the most annoying and repetitive tasks. Tasks include writing email sequences for different cohorts, taking a winning piece of ad creative and generating more variations, writing creative briefs based on last week’s data, automating transfers between your Checking and Treasury accounts, and handling the most significant task: customer support tickets. Anything that can be repeated is your target.

2. Budget 

Next, outline a budget for your team to use and explore AI. You should approach them with these problems and ask how they could utilize AI to improve their lives. Get them on platforms like TripleWhale, OpenAI, Claude, Google’s Gemini/Veo, N8N, Highbeam (you don’t have to bank with them to use their AI).

3. Enhance, Don’t Replace (yet) 

I recall attending a fancy CMO dinner last year, where someone mentioned that they had fired their entire team and replaced them with AI. I’m not sure how you can do that, but AI is amazing at making people more efficient. It won’t replace your talented graphic designer; it’ll enable them 8x their output. It won’t replace your finance team, but it’ll lessen the time it takes to do everything manually.

Through experimentation, everyone on your team will discover efficiencies within their workflows, enabling them to accomplish more, faster, with fewer resources. For those who like to get things done, this is amazing.

4. Start with Measurable Tasks

I would also start with something that has clear before/after metrics. Usually, content creation or customer service. Why? Because "I saved 10 hours this week" and “My customer support response time dropped from 1 hour to 4 seconds" are results you can measure, and they add real value to your bottom line.

“Static ad output this week increased from 15 unique statics with two variations per designer, per week, to 25 unique statics with six variations of copy-overlay on top.” It’s easily measurable, and you can see the direct impact of AI.

5. Train your team 

This is super important. AI tools without proper training are like giving someone a Formula 1 car without teaching them how to drive. It’s dangerous and wasteful. I think you should schedule time on people’s calendars to explore AI. Invest in continuous education for them. Share YouTube videos and X threads to help them in Slack. I think learning to work well with AI is similar to learning to be proficient with the Adobe suite or Figma. It will take time.

You need to educate people on how prompting works and how to provide it with the right context (which is why it’s crucial to clean and organize your data before asking it to perform tasks).

6. Push for More

Just when you think you have started to hit a point where you’re using AI as much as you can, just remember each week of innovation within AI feels 2x faster than the innovation that happened the week prior. Google just announced Veo2, and once it was released, a week later Veo3 Beta was available (now allowing you to create videos with sound). 

A few ways to implement AI today: 

Okay, now let's get practical. These aren't just "maybe I can do this someday" ideas; these are things you can do right now. Companies like TripleWhale and Highbeam have AI products built into their platforms, allowing you to use them instantly. But more than that, what these platforms are doing without you realizing it, is organizing and segmenting your data (it’s like an AI Agent that organizes your data to be better utilized with other AI agents). Your AI is only as good as the context and inputs it has, which is why knowing how to prompt is so essential. 

TripleWhale imports data from all your sources (Meta, Google, Shopify, Klaviyo, Postscript, post-purchase surveys, etc.), cleans the data, organizes it, and then provides a straightforward drag-and-drop builder to create agents, utilizing all that context and data. Highbeam allows you to manage what money goes to treasuries to maximize earnings, but doesn’t hinder your ability to pay bills. It can also develop an entire financial forecast based on chatting back and forth with the interface.

Here’s an example of how this looks in practice:

Old-Old Way: A creative analyst sifts through the past week’s data of ad performance, tries to identify patterns, searches the internet for related/trending topics, and works on a creative brief for a new ad batch. This typically takes 2-5 hours, depending on the complexity of the product lineup that requires a new brief.

Old AI Way: Take notes and each week add them into a ChatGPT conversation, where the conversation continues to get smarter and learn more about what works for your brand/business. From there, you can expedite the process and maybe cut this down to 1-3 hours.

New AI Way: Build an AI Agent workflow to analyze the past week’s ad performance data, derive patterns and insights, scour the internet (Reddit, Twitter, forums) to see how people are talking about XYZ, read my forecasted needs, build me a creative brief that helps me achieve the goals I need. This is now a 3-minute automation and ends with a creative brief in Slack. The New AI Way is realistically still 5-6 months out from when most people will try it for the first time.

What’s coming next: 

I think we're moving from AI tools to AI employees. Not in a "robots taking jobs" way, but in a "finally, someone who can handle the entire workflow" way.

Imagine AI that doesn't just generate product descriptions, but researches your competitors, analyzes customer feedback, identifies content gaps, creates entire product launch campaigns, etc., and iterates the messaging based on what's converting. While researching if the Agent believes it needs more help to do another task, it’ll create another Agent and program it to do what it needs. I think this can occur within 8-10 months. 

The brands currently testing AI and becoming comfortable with it will be the biggest winners when the next generation of truly autonomous agents emerges in a few years. Just like using any new technology, the more you use it, the easier it becomes to understand how to make it more potent for you.

What else is on your mind about AI? I am working on Part 2 of this newsletter, which will delve deeper into workflows, Agents, and LLMs that will integrate within company operating systems. That should be ready shortly and will be included in our Sunday newsletter. Reply to this email with any other thoughts/questions you want answered about AI.

That's all for this week

I hope today’s newsletter sparked some new thoughts or ideas on how you can level up yourself or your brand using AI. I have to admit that when I saw TripleWhale’s AI product, I was genuinely impressed. Even if you don’t use it for reporting, just being able to connect all your data and then ask any type of question is the most helpful tool for any founder or executive looking to get clarity before making decisions.

It’s Sunday night, so I hope you're planning to get a great sleep before the new week begins. Get 9 hours of sleep tonight, break a nice sweat this week, stay hydrated, and I’ll see you next Sunday evening!

Oh, and if you want to come to the Ecom AI Summit in New York City on June 25th, 2025 at Webster Hall — get a free ticket on this special link!