What Big Companies Could Learn From Small Companies About AI


Happy GTA launch day and Anthropic IPO day to those that celebrate. Today we're going to talk about what larger companies can learn from smaller companies about AI specifically. But in general, I think there is so much smaller companies can teach larger companies. I've seen it firsthand going from Jones Road to Winks lately. If you could ever take some of the operators of these newer companies and put them inside of larger companies with budgets and resources but find a way to let them cook without the normal politics of a big company, it would be epic.

I’ve now seen AI adoption up close at almost every company size, from big teams to tiny ones. There is a pretty clear pattern of adoption and transformation.

The companies with the most money and the most people are getting the least out of AI. They are using AI much less, and generally feel less confident in what it is capable of. The smallest and newest ones are doing things that look made up if you haven’t seen them. For every large company that says AI is not there yet for a particular function, there are several companies completely owning that function with AI.

So I thought I'd write about the what the different company stages look like, and what larger companies can learn from the smaller, AI native ones.

Before we get into it, thanks to Aftersell for being one of today's partners. Partner of the Week: Rokt Thanks by Aftersell

Black Friday is all about driving more revenue, but don't forget about the profit behind every order. With CPMs up and competition getting tougher, finding ways to make more from each customer matters. Rokt Thanks by Aftersell helps you do exactly that without touching your checkout. It adds a block to your thank-you page featuring offers from brands like Disney+ and HelloFresh, earning you roughly $0.30 per order when customers engage. At 50K monthly orders, that's an extra $15K you can put right back into your ads. Jones Road is already using it, and it's an easy way to squeeze more profit out of every BFCM order.

Check out what it could mean for your brand here. #ad

The three tiers

1. Large companies (roughly 100+ people, all the way up to enterprise)

They have the biggest budgets and the most resources. Leadership has an AI mandate, but it sometimes stops at that. These guys are usually using it the least, finding less productivity gains than expected, and are now skeptical of the ROI. I've found leaders and employees at these companies skeptical of what AI can do. Or if they are excited about it, they are using it to a very small degree of the potential. They likely use Claude and use it to help save time and build dashboards. They gather things with AI, but it either replaces manual work, or pulls data in a way that saves time.

There are real reasons why AI implementation is not super advanced here. For one, these teams usually have real headcount and budget, so there is less of a need. There is more comfort and less necessity. They also have security and compliance issues, brand standards, and more. Existing tools and workflows also make it harder to innovate. There is the workflow debt, but also tech debt that

Some of that is real: security and compliance, brand standards. A good amount of it is existing workflows. Processes built over years don’t bend easily.

But the biggest blocker, in my opinion, is people. Politics, and beliefs about job security. If someone thinks getting good at AI makes their own job go away, they won’t go out of their way to get good at it. Which is funny because the data is pretty clear so far, companies that are more aggressively implementing AI are hiring a lot more. But there are real concerns here that need to be led well to get teams successfully adopting.

2. Mid-size companies (roughly 20 to 200 people)

These guys are mixed but generally doing a pretty good job. They’re small enough to move fast and need some resources, but big enough to have some of the existing debt I mentioned so they can go either way. It's likely all about the leadership and culture. Most brands doing well overall are leaning in because they likely have the cultures that help AI be successful. What I've found is the best employees without AI are the best at AI. They are ambitious, driven, and always want to find a way to get the outcome done. So AI is like rocketpower for them. Your job as a leader is to identify those people and invest in them. Shield them from politics and existing workflows and let them cook. Let them experiment and push the limits, then report back and find a way to have the rest of the team catch up.

Now AI is not everything, and there is A TON that goes into having a successful brand that is not AI. But speed is the name of the game, and it gets very hard to compete with competitors that can move 5x as fast as you with less people.

3. AI-native companies (started in the last year or two, or starting now)

Then you have the new guys. There are 21-year-olds absolutely cooking with tiny teams and running the whole thing with AI. Now not everybody needs to be doing this. But they are great examples of what can be done when you don't have preconceived notions of what can get in the way and what is not possible.

It's super inspiring to me. These guys are fully running things with 1 person and an army of agents. Here are a few things I've seen these companies run with agents and also some tips on how you can learn from the.

It’s not a technical problem

I talk to a lot of people at larger brands who are trying to make this work and hitting roadblocks. Almost none of those roadblocks are technical.

The technical stuff is table stakes now. Enough people know how to set up MCPs, connect tools and get an agent running. Maybe not amazingly well, but well enough.

A lot of it is belief and removing what is getting in their way.

1. Training, knowledge and access

People need to be taught, and they need access to the tools. Either someone internally champions it and teaches everyone else, or you bring in outside help. There’s a lot of good outside support right now like Chatwalrus. Or shameless plug for the cohort-based course HAQ is running that I am a part of. This first cohort is closed, but if you'd like to be put on the waitlist for cohort 2, reply and let me know.

2. A culture where people build and share

After investing in your team, you then need to encourage them to build stuff, give the budgets to do so, and reward it. No one can feel like you are asking them to automate themselves out of a job.

3. Force it

It should only be a small part of the puzzle, but after the trainings and encuraging it, I do think you should force it to a degree. But make it fun if you can. For example, Jacob from HQ had a good post about getting better at AI by constraining yourself from using the browser for a week. It's really hard to change processes. Make it fun but also challenge yourself or your team and force them to fully push what they can do agentically.

4. Think More Ambitiously.

This one is so simple but it really works.

Most people's first thing they do when they get an AI agent is try to save time and replace their manual work. That is fine, but it can do so much more for you. Even as an AI pilled person, I find myself reminded of this often. I sometimes get complacent or think something can't be done, and then I'm reminded of what it can do. It's much better to push the limits and be disappointed than not try.

Partner Of The Week - You don’t really own your Instagram followers. Every time you want to reach them, you’re either fighting the algorithm or paying Meta to show up in their feed. With email, you can reach people directly, as often as you want, without paying for their attention again. Omnisend’s new Social Capture connects the two. You set up a keyword reply in Omnisend, ask followers to comment or DM that word on a post, and they get a DM back asking for their email. They reply and they’re on your list. With all the Instagram traffic coming over BFCM, now’s a good time to set it up. Omnisend is used by 150,000+ brands and has a 4.8/5 rating on Shopify. If you’re on Klaviyo, they say your bill can be up to 35% lower for the same features, and paid accounts get free migration in 5 days. Take a look here. ​

​
This is the last post before Thanksgiving and black Friday. Enjoy the holidays and best of luck!

Cody

​
​Unsubscribe ·

Cody Plofker

Hey, I’m Cody. I'm CMO of a 9 figure DTC brand and write a weekly newsletter with actionable marketing advice to make you a better marketer in 5 minutes a week.

Read more from Cody Plofker

Hope you’re having a great week. This year feels a little weird. It’s my first year not deep in Black Friday, and it’s bittersweet. I’ll definitely miss a lot of it, but it’s also nice not to have the pressure. Today I am going to be giving some tips for right before and during BFCM. But if you need some last-minute ad creative before BFCM, I highly recommend Insense. They're a trusted partner for high-quality creator content. In one place you can streamline the sourcing, briefing, sending,...

Hey, hope you're having a great weekend. Like I said, I'm trying to make these more regular. I'll likely send Thursdays, but some weeks it doesn't happen and sundays it is. If there are topics you want me to write about, please let me know. This week will be a busy one. I'm flying to do a all day video shoot for a new large partnership I'm doing. Excited to have it announced soon; it will be a fun shoot. Very different and out of my comfort zone. Then I'll be speaking at Forward by Tatari on...

Hey, happy Friday! Hope you're having a great week. Still not in the habit of sending these regularly, but I'm trying. Winks is taking up a lot of time, and I'm not gonna lie, it's definitely a struggle. We're having success in some areas and challenges in others, but it's a lot of fun. We're still trying to find the right offer and nail down the messaging. The good news is it's clear how many people struggle with sleep. The hard part is getting them to take out their wallets and pay us. But...