Introduction

AI is transforming the way startups do business. From developing a marketing strategy, to sales prospecting, artificial, artificial intelligence allows smaller teams and younger businesses to do more with fewer resources and capital.

According to HubSpot for Startup’s 2024 report on how AI is redefining startup GTM strategy, 80% of respondents believe AI will help them grow. Another report from Exploding Topics found that 82% of businesses are either already using or exploring the use of AI; however, the organization found that larger businesses are currently more likely to invest in AI tools. 

But for startups, AI can be an invaluable resource for scaling—that is, if you know how to use it appropriately.

A conversation with Kieran Flanagan, SVP of Marketing at HubSpot

Kieran Flanagan, SVP of Marketing at HubSpot, is well-versed in the benefits—and pitfalls—of current AI technology. 

In addition to his role at HubSpot, Flanagan is a Sequoia Scout for Sequoia Capital, an advisor to companies like Miro and Ramp, and a pre-seed and seed angel investor under his own company. He also co-hosts Marketing Against the Grain with HubSpot CMO, Kipp Bodnar,  a podcast that covers marketing trends and advice for growth, including how AI is impacting marketing. 

Whether as an advisor or an angel investor, Flanagan has watched many startups utilize AI to scale marketing or qualify leads in sales. Still, he’s also seen companies make big AI mistakes that affect the customer journey.

Flanagan sat down to discuss the state of AI in marketing today, plus how these tools can both positively and negatively impact startups in their growth journeys.

HSFS: What are the best ways that early-stage startups can scale their marketing with AI?

Kieran Flanagan: For early-stage startups, I think AI is pretty good at content and ad creative. It's pretty good at going back and forth on lightweight strategy because if you're a startup, you don't need a really complex strategy.

So it's a pretty good marketing partner for that. It's pretty good at email and anything text-based. It can at least give you—I wouldn't even call it a first draft—a pretty good draft that you can actually edit and make a little bit better. And, in some cases, maybe just even ship if you know how to use AI.

Which tools can smaller startups without a marketing team use to get started?

KF: For founders, I think one of the best ways is to try to have AI as your ghostwriter for social because, for the most part, if you're going to be a small, scrappy company, you want to be founder-led in terms of your content creation strategy. So you want the founder to have their own brand and use whatever social channels the company's prospects are on to have a brand built around that founder’s thoughts, point of view, and why their tool exists.

I think that's what I would use it for. I'd start to play around with the tools like Claude. You can start to build some templates and fine-tune the AI on examples of content that you want your style to be like. You can create a writing style based on any kind of content.

If you actually give it 10 to 20 posts and ask it to create a writing style from those 20 posts of someone you really like, it will create that template. Then, anytime you want to create a post, you just upload the writing style and say, “Can you create a piece of content, a first draft on this topic in this writing style,” and that's one of the ways you can play around with different writing styles to see what appeals to the customers that you're trying to appeal to.

How can startups utilize AI during the fundraising process?

KF: AI’s incredibly good for prospecting. If you actually give it a bunch of information about some of the funds you're trying to raise from, or even the person you're trying to raise from, it can actually create a pretty good first version of your outreach email. The more data you can give it, the better and more creative it can get with the prospecting email.

If I were doing anything text-based, I would be trying to use these tools. I think all you need to do is pay $20 a month or whatever it is for Claude. That's really it. 

I don't think you need any of these writing tools. All the writing tools just use the large language model (LLM) and then put the template on top of it. So, you might as well just create the templates for yourself.

What are some of the common templates or other elements you see people creating?

KF: I think a couple of things that make a lot of sense are understanding which post-type templates do best on social media. Basically, look at all of the posts in your category that are doing well, and then have Claude work with you to create a template from that post. That is a post template. Then, you have all these different post templates that perform really well on whatever chosen platform you have. 

And then the next one is the writing styles; you have a bunch of different ways that you can actually write, and if you know how to get the content that you like or the style that you like, you can actually work with Claude to create a writing style. Now, you have post templates for your platforms, and you can apply a writing style to those templates to bring them to life. 

You could argue, “Well, should you not have your own writing style?” 

I do believe you should have your own writing style, but they're actually a good starting point, so you can start to add to the template.

The AI says, “Give me the kind of core things that that this writing style should adhere to,” whether that's conversational, wisdom, punchy, witty, humorous, or whatever it may be. You can start to add your own things and edit to your style.

But it's a good starting point.

How do AI models work when setting goals for your content?

KF: You can segment them in the writing style. You could do a bunch of different things. But, what I do is key principles for the writing style, and then detailed guidelines. Within the detailed guidelines, you could have the call to action. For example, I've added into a writing style, “Don’t use emojis or hashtags.” So you can start to edit it to your tastes.

Have you noticed any brands doing interesting things with using AI in their marketing strategy?

KF: One of the companies I invested in recently allows your reps to call into an AI agent and train to be different versions of your customer. It's like a gym for sales reps that are still on ramp. That's pretty cool. I think that's a differentiated use case. 

There's a startup we're talking to that is able to provide everyone with an AI avatar that's like a sales engineer. So, it's trained on every single piece of knowledge around your product marketing material and your sales material, and it can join any call. Anytime the sales rep needs to have a question answered, it can just appear and answer the question. 

There are a ton of use cases that the startups that I'm either invested in or working with are doing that they're applying to their go-to-market. 

I think AI is very heavily weighted toward the customer interaction teams, like sales,  customer success, and support.

How can marketing leaders make a more AI-friendly culture for their teams?

KF: I think you have to hire the right people. If you're the one trying to force marketers to be curious about AI, it could be that you've just hired the wrong people. Because if you have really good marketers, they should be deeply curious about technology that changes the way they work.

Now, if you want to do something to support that, I think hackathons are the best thing you can do, which is to give the team half a day or a day. Just let them run a hackathon like a developer team would normally run a hackathon, getting them into pods and building cool things. That gets them playing around with the tools. It's really fun, and then you have some sort of competition where you upvote the best tool. Give them specific problems that you would want to solve in the company’s marketing. It’s as a prompt, so they can try to figure out how to build things that would actually get used within the company. 

That, to me, is the number one way I've seen companies foster a better culture of AI curiosity within the team. 

Then, you just have to provide the team with tools and make sure that they actually can play around with whatever LLM you want and that they have the right access to do cool things with it.

Have you seen any companies make major mistakes when using AI?

KF: Most companies started with AI with a “do” AI strategy, which basically involved having everyone figure out how to do something with AI. I think where we are as an industry, we are trying to figure out the ROI of all this AI work, and they're moving toward more of a deliberate strategy where they're like, “Okay, what are the big bets we want to make around AI within the company. How can we measure the return on investment in that bet?”

I think the biggest miss for companies is that they stay in the “do” phase. They have a ton of things being built with AI and a ton of fun projects, but no one's really sure if they're adding any value at all. No one's really sure how to measure the return on them.

The other mistake that I've seen a lot is all the teams just want to add some sort of AI chat to the part of the customer journey that they own. So, you end up with several different chat elements across your customer journey. They don't talk to each other. They don't pass context across chat. They're all different experiences. It's really messy for the customer. 

What those teams are thinking about is, “Oh, this is really cool. Let's add AI into our little bit of the customer journey.” But for the customer, it's a totally different experience with all these weird AI things plugged into whatever part of the customer journey that each team owns. You basically are just shipping your order design with AI intertwined in it. It makes the customer's life worse, not better.

An example of that was chatbots. Back in the day, everyone got really excited by the early versions of chatbots. We were going to use chatbots for customer success, and every single person who used a chatbot was like, “This is the worst experience of my life. I just want to talk to an actual person because this can't solve anything for me.” They still exist everywhere. I go places, and I have to talk to this chatbot. It can answer two questions and then just says, “Oh, I can't help you anymore.”

So, you still have to ship things that are as good or better than the current version of whatever you have.

Are there other ways, positive or negative, that you’ve seen AI impacting the customer experience?

KF: The negative is when you have a weird, decentralized AI experience. You have all these teams in the “do” phase trying to add AI into all their parts of the customer journey, and it shows up to the customer in a really weird way.

I believe the way we're going to make things better for the customer is that AI will probably collapse more channels into singular teams. I suspect marketing is going to own much more of the customer journey than it ever has. 

Today, what happens is this team does something, and then they have to hand it over to this team because this team has a specialty in that part of the journey. Then you have to hand it over to that team, and that team has the specialty about that part, and you end up shipping your teamwork. You end up shipping your team design, and that experience becomes very cumbersome.

For example, you go to a website and chat with the person there. Then, you go into the freemium product and chat with the person in there. Then, you go through some onboarding flow, and then you upgrade, and then you talk to a business development representative (BDR) who wants to book a meeting with you, and then you get passed to a sales rep, and the sales rep closes you. Then, you become a customer, and you go back to chat because you want to talk to support. It's a different team. Then, you ask, “Can I talk to someone who’s in customer success?” And it’s a different team.

AI will collapse all that because there will be an AI avatar that's like your concierge, and that AI avatar will follow you throughout the journey. It doesn't need to live in separate teams. The reason all of this lives in separate teams is that those separate teams have a specialty of knowledge. The marketer knows this part of the journey. The support person knows this part of the journey. The BDR knows this part of the journey. AI knows all of that, so we can have a singular journey through this consistent experience. And AI can context switch, so you don't need to have a separate experience.

Because the customer journey starts in marketing, I actually think marketing can own the majority of a modern B2B digital experience—not the back end, where you have to call someone and actually talk to the sales rep or the customer success rep—but the digital experience can and will be owned by marketing.

Are there any AI tools or features you are particularly excited about right now?

KF: In terms of features, Claude’s Artifacts was probably the best feature that has launched in a long time, so that has been my number one fun feature from AI. 

In terms of tools, we are looking at a bunch of different go-to-market tools that I really do think can transform the way that we think about a modern B2B go-to-market. 

Again, coming back to what I said, you can collapse everything into just a singular experience and have AI do the first interaction, qualify it, and send you to the correct place. So, humans really never have to talk to anyone until it's qualified and they have great context about that call. They understand how to show up and do a great call.

AI can do the whole digital part. That's the thing that I'm most interested in right now. 

There are some really great tools in that space, and I’ve invested in a lot of them. But there are so many that I don’t want to call out certain ones over others.

Conclusion

As we can see, AI tools are powerful assets for early-stage startups—that is if you know how to use them properly. As Flanagan made clear, sticking to the “do” phase with AI with no clear goals or direction is a huge mistake. 

But by nailing down your brand’s personal needs, you can tailor more helpful prompts to train AI to work more effectively for you. For instance, Flanagan and HubSpot CMO Kipp Bodnar have a free curated collection of AI prompts in a user-friendly library, so your startup can start scaling smartly with AI.

Whether you tap AI for marketing, customer support, financials, or even engineering, use these best practices for the best results, not just for your own experience but for the customers’, too.