If you're a CMO who's been assigned the sweeping job of "figuring out AI," we've got some great tips for you today from our inaugural session of the AI Marketing Alliance.

Who better to talk to about this than Kieran Flanagan, SVP of Marketing at HubSpot, a seasoned advisor to some of the fastest growing tech companies in the industry? Kieran is also co-host of the top marketing podcast, Marketing Against the Grain, and he runs his own newsletter about marketing and AI.

Kieran talked to us about the importance of AI prompting, why CMOs need to be AI practitioners themselves, and more.

Read on for:

• AI is unlocking marketers' creativity
• 5 areas where AI shines
• AI hallucinations happen, but they're by design
• Are you a CMO tasked with integrating AI? Start here.
• Video AI tools are up and coming
• Use AI to get better at your job, not to create more lackluster content

Tune in below for the full conversation with Kieran and Lindsay McGuire, our Associate Director of Content and Campaigns!

WATCH NOW

As a trained software engineer, Kieran always knew he wanted to build things but realized pretty early on that he wasn't naturally skilled at coding. By chance, he got into digital marketing and loved the entrepreneurial spirit of the field at that time.

This was a period when you could enter the marketing field and forge your own path. You could learn how to be the best social media marketer, or build the best blog, or work on one of the first newsletters in the space.

That time has long passed, and most marketers know how to do all of those things now.

However, with the advent of AI, we all have the chance to build something new. AI is forcing marketers to rethink everything we thought we knew, and that's not a bad thing.

AI removes the barrier of tooling, which allows us to dream big and think creatively as we architect new content and processes.

HubSpot has long focused on being at the forefront of building strong GTM motions. In the first decade of the company, inbound marketing worked. MQLs were sent to sales, and the sales team would close the deal.

Now, consumer behavior has changed. People want to try products before they decide to purchase or upgrade, making GTM more of a product-led motion. HubSpot's revenue reflects that change—currently, more of the company's $2B in recurring revenue comes from product-led motions than traditional market and qualified lead channels.

AI makes a ton of sense for HubSpot. As the go-to-market front office platform for many businesses, Kieran wants to figure out how AI is going to change things for everyone. That way, HubSpot can then pass on the best parts of those lessons to HubSpot customers through its tooling.

Here are the five areas where AI is strongest at the moment:

1. Chat: Chat has become a ubiquitous part of B2B, and AI is good at this. You can even take all of the chat logs and fine-tune models over time so that the AI tools improve.
2. Prospecting: Soon, every company will have an AI-driven prospecting engine that can identify ICPs. AI will be able to research that and marry internal and external data in personalized, authentic messaging.
3. Sales enablement: AI can provide reps with meeting context, tell salespeople what prospects have done up until the meeting, and highlight items the sales rep should touch on during the meeting. It can also grade reps on their performance and indicate areas of improvement.
4. Customer support: This is a well-known use case; you've likely interacted with AI if you've ever encountered a chatbot.
5. Product usage: You can use AI to increase the success customers have with your products and how comfortable they feel using them.

Those five areas sound pretty good, right? We agree, but we'd be remiss not to touch on the things that aren't working so well with AI right now.

There's still lots to work out with AI. For one, "AI hallucinations" happen when the tools create false information, and it's not an uncommon event.

Though it may seem like a bug to us laypeople, hallucinations are actually a feature built into AI. AI is not an answer engine, and it's not meant to give you the exact right answer to all of your questions.

What it is meant to do is try to do what it thinks you want to do, and there is always a chance the AI tool can misread that.

To solve for this, put tight restrictions on what you do with AI. Give the tool clear guardrails and prompts (more on that soon), and test with a small segment before you launch to a wider audience.

Know that you have to put AI out there so the tool can acquire more knowledge, and you have to fine-tune it as you go, if you want it to get better. It's something you have to commit to for the long run and keep at it.

Kieran is a believer that this is the future, whether people want to deal with AI or not. Taking the time now to run AI use cases, learn, and iterate will pay off in spades in the future.

If you're a senior marketing leader who's been tasked with investing in AI, you might be wondering what to do first. Or maybe you're thinking you should go ahead and assign it to your team to figure it out and report back. Easy-peasy!

The problem with this approach is that you have no perspective on where the value should accrue, so you'll have no idea if your team is going about it the right way.

If, on the other hand, you're a practitioner of AI yourself, you can better direct your team to unlock the value of AI. That doesn't mean people have to wait around for you to become an AI whiz, either. You're all on this journey together.

Prompting should be a major area of focus for you and your team as you learn what AI can do for you. Without good prompts, AI tools can't do good work.

https://admin.goldcast.io/shared/clip/d233097a-0eb1-48c1-ad8e-96fbf1238e4e

Consider taking prompting classes, or host hackathons where people can test and iterate on different prompts. (Hackathons used to be just for engineers, but they're a great way to get different teams together to collaborate on tools. Kieran's met a ton of people who were first exposed to AI tools through hackathons and are now doing very smart, very cool things.)

Claude is a good tool that features a prompt library. If you ask Claude to build you a prompt, it can produce some solid ones you can learn from.

You should have two buckets of AI initiatives: ones that you can do today based on the current state of the models, and ones where the AI will be transformational in the future.

For the former, a good place to start is to go through all of your emails and see if AI can improve and personalize them with the right data and context. That's an easy win and a relatively light lift.

For the forward-thinking AI goals, you're looking at places in your customer journey where AI can make a big impact. Kieran bets that the models will only get better over time, so if you build the infrastructure for the future use cases, you'll be ready to plug-and-play when the tech catches up with your vision.

Review the five areas we covered where AI is already strong: chat, prospecting, sales enablement, customer support, and product usage. These can help you brainstorm ways AI can be put to use right away in your company.

When Kieran worked at Zapier, they used AI to capture the apps that new users relied on. From there, based on the new user's role, the team could leverage AI to identify what the user wanted Zapier to do and build a personalized onboarding document that spoke to those specific goals. AI allowed them to do these things en masse, whereas in the past it would've taken a lot of hands-on work.

With tools like HeyGen, an AI video generator, you can now use AI to draft a video script and then create a video with your avatar. While it's not the most realistic, lifelike avatar, Kieran has created a course using HeyGen with good results and he thinks video AI tools will only progress from here.

As of now, you can't find an AI tool to whip up a super high-quality video from scratch the way you might want to, but this goes back to those future-thinking goals. That problem will likely be solved in the future because it's one that all of us have. We'll likely be able to do much more with AI and video over the next 1-2 years, so why not get ready now?

PS: If you've got an existing video from an event, podcast, or something else, tools like Goldcast Content Lab can absolutely turn that into quality clips, snippets, audiograms, social posts, and more. It takes just seconds! To learn more or sign up for a free two-week trial, click below.

DISCOVER CONTENT LAB

AI can enable any of us, right now, to do more of whatever we're doing at scale. That means that if someone's doing mediocre stuff, they can now do exponentially more of it—not exactly a winning equation.

Kieran hopes that marketers will instead focus on using AI to become better at our jobs. If you know how to use AI effectively, you can accelerate your quality output. You get more time back to focus on the creative tasks that need a human touch. This is needed in a market where there's more competition and more highly-trained marketing talent than ever before!

PS: Want more content like this? Join the AI Marketing Alliance, where you'll meet other B2B marketing leaders and get tips on how to apply AI to your day-to-day work!

LEARN MORE