Some practical ways to use AI in marketing

As we wind down on 2025, a year where there’s been undoubted gains in and hype around AI, I thought it’d be useful to summarise some of the ways I’ve been using AI in my day-to-day.

This post is not designed to be conclusive: there’s many more ways you can use AI than those listed here. It’s not meant to cover all the available tools and it’s not meant to cover some of the advanced engineering-driven ways you can use AI.

Instead, it’s a collection of tips I hope you can use immediately.

Longform writing by talking

I’m starting here as this has been the biggest gamechanger for me.

Virtually every marketer has to do some form of long-form writing and using new, AI-driven dictation products like WisprFlow is epic.

Dictation products have been around at least as long as I’ve been using the internet. As someone who’s always done a lot of writing in their work, I’ve tried many dictation tools, remembering those such as Dragon NaturallySpeaking and, more recently, using Apple’s built-in dictation software.

They have always had a huge challenge in that too much of the formatting is manual. You cannot just speak and get reasonably well-edited writing. You spend far too long backtracking and directing the tool: “period here”, “new paragraph”, “scratch that, I mean, no, delete! Backspace!”

WisprFlow has changed the game for me as it automatically formats paragraphs and interprets control text like “scratch that” along with tidying up all the "ums" and "ahs" and other unnecessary text. It even detects where to put new paragraphs. Without direction.

It’s not perfect, but it’s massively better than anything that I have used in the past, and I’m sure you will only get better along with competing tools. If you have to do any form of long-form writing, I would strongly recommend checking this out.

Brainstorming

A no-brainer, and one that I imagine many already doing. Brainstorming has been another of the most effective uses of AI for me. I use ChatGPT as a sparring partner when brainstorming new blog posts, brainstorming social content for my LinkedIn, for the company LinkedIn, brainstorming ad ideas and so on.

A couple of tips here that I found have been useful for me:

1. I encourage ChatGPT to push back, not just to agree with everything I say. I regularly tell it that I don’t like specific suggestions it makes, regularly ask it to think outside the box or go wider and these tend to generate some of the more useful or interesting ideas.
2. In my experience, it’s most useful when you’ve already got a direction you want to go in, and you’re merely asking ChatGPT to flesh out the direction or provide a countervailing view.

Building lightweight tools

One of the great things about the recent AI wave has been the rise of prototyping tools like Lovable and Repl.it. Combined with Cursor, Claude and other models, if you have a little bit of technical know-how, these products can help you build genuine software. Whilst I’m currently sceptical that these tools are at a point where they can build out production-ready software at this time, there’s no doubt they can build complete lightweight products. I’ve done it myself.

Releasing free tools can be effective marketing and, whilst it isn’t the right fit for every company, the software you can build with AI is a great fit for these sorts of marketing initiatives because the tools you release do not necessarily need to be production-ready. Many tools will operate very simply, very transactionally, perhaps even run in the browser. They generally won’t require the same security rigour or scalability as your actual production software.

The example tool I released last week, that enables you to convert Markdown to fully-rendered HTML emails, is something that was written virtually entirely by AI under my direction and is an example of using AI to market Vero.

Content conversion

Something else that I imagine many are already doing: it’s now relatively easy to take long-form internal or external content and to use AI to generate short-form content, such as ads or social media copy, or content in a different format, such as a cartoon.

You can take a blog post, some internal documentation on strategy or a series of customer conversations and ask AI to convert these into something you could share on X or LinkedIn or other relevant social networks. The same applies to virtually any other channel.

Image design

A huge part of marketing is visual. Canva, Midjourney, Nano Banana: there’s been huge advancements in image and video generation using AI in the last year or so.

I’ve used these tools to generate cartoons, to generate a starting point in Canva that I’ve then taken and improved and you can run much further to create designs you can use in ads, as many are now doing.

Whilst these tools may not always be able to create a complete, finalised version of what you want, I am continually finding them useful for rapidly generating elements that I can then piece together to get the end result. And to do so a lot faster. Or even to do things I simply couldn’t have done before.

Something that would have taken a designer a considerable amount of time is now much more achievable for our full team, including those that are non-designers. The net result is that we are able to be more creative with what we’re putting out there.

Optimizing and critiquing

I have found optimizing works best when optimizing something for which there are broadly agreed principles on what “good” look like. As an example, optimizing content for SEO.

Other examples could be asking AI to optimise a social media post, subject line, ad copy, some email content, etc. AI is useful as it applies both technical analysis, e.g. how long a subject line is, with qualitative feedback on your approach.

Similarly, asking ChatGPT or another model to critique something you’ve produced and deliberately come up with alternative angles is a really useful strategy. This generally leads to an optimized result.

Multiplier: talking to AI

A multiplier to all the above is using dictation tools like WisprFlow to interact with all of these other tools.

I have found that when you are iterating on AI code or writing long-form content and asking for critiques or brainstorming, you generally have a lot of back-and-forth with the AI model. It can get very quickly slow you down, or you might find yourself asking if it’s in fact quicker just to do the work rather than type everything into ChatGPT.

Sometimes the tool (e.g. ChatGPT) has good enough dictation you can use that. But many of the others do not. That’s where WisprFlow comes in.

–

I hope there are a few ideas in here that you aren’t already applying and that you can use immediately.

Also, I wanted to note that I deliberately haven’t spoken about using AI to fully write your content, despite the proliferation of tools in this category like Copy.ai, Jasper.ai, etc. I believe generating content using AI is a race to the bottom. I’ve instead tried to focus on tools that massively multiply our team’s ability to write natively (i.e. without AI generating the raw draft).

By brainstorming, using a dictation application and optimizing, you can take your genuine ideas and copy and very quickly write something long-form that is original and unique to you.

I think the same is true of visual designs: combining AI-generated elements with elements you’ve created yourself yields something more original.