I'm a fan of free tools where viable, so I often use Ahrefs Image Alt Text Generator, which uses a language model to analyse images and suggests alt text descriptions I can use in emails or blog posts.
I'm a big MidJourney fan, simply because of the detail and control you can have over image output.
That said, one of the big reasons I use TouchBasePro to send my emails is because of their email-builder's ability to generate images from prompts, which means I don't need to leave my email platform to create images.
Again, I use TouchBasePro for these components due to the extensive use of optimised content feeds, their recommender engine, and for calculating optimised email sending time.
That said, for subject lines, I'm a fan of subjectline.com which has a beta AI engine for testing subject line efficacy.
Most email platforms have great tools to ensure mobile responsiveness across various devices.
For easy use, and a tool any email sender can use, I recommend Stripo, which works really well to create emails in a drag-and-drop editor.
For advanced, coded emails, I've also seen Parcel come highly recommended within the industry.
The best way to deliver truly dynamic and personalised email experiences is by understanding your audience.
The best way to do this is through your data. I've seen that Seino has burst on to the scene with an amazing platform to analyse email metrics and Google Analytics data to create a truly dynamic email experience.
Again, Stripo is a winner here! Our team have completed multiple projects for clients using this tool and the level of ease-of-use, time to completion and ability to select execution formats makes this a choice I'd happily recommend.
There are quite a few up-and-coming tools in this space, but I've still found Zapier and Make to be some of the most useful in this space.
Because they're established in this space, you'll want to watch out for the introduction of their AI components and what they build in the AI space to bolster client experiences.
To analyze images and suggest or generate alt text descriptions
AI Alt Text is free to use and straight-to-the-point. It provides alt text descriptions even for simple images. It also has a WordPress plugin that simplifies my day-to-day work.
Using Generative Adversarial Networks (GANs) to generate images, graphics, or illustrations.
I love Canva’s built-in AI image generator where you can also select the style of your image (eg. anime, watercolor, 3D). The tool can generate both images and videos.
AI-powered A/B testing algorithms that can help optimize email campaigns - subject lines, content, images, and other elements.
I don’t often use AI for A/B testing. I rely on the tools available in the ESPs, like Mailchimp or GetResponse, although they’re not always based on AI.
How AI algorithms can automatically adapt email layouts and designs to ensure they look perfect on various devices and screen sizes.
Again, I don’t rely on AI when it comes to responsive email design. I prefer to use the ESPs’ features to preview and test email designs on different devices.
AI-powered data analytics tools and integrating AI with CRMs can deliver data that can be used to create dynamic email designs to improve the personalization game.
AI tools that help to convert design files in all formats to a coded email. What level of perfection do they deliver? Once the tool provides the code, how much time is required manually to make the email template client-ready?
I don’t need that.
AI-powered tools that help to do various types of integrations- integrating email in ESP, integrating different tools with the ESP, etc.
I usually use Zapier for all sorts of integrations, not necessarily email-related. They’ve recently introduced some features that are based on AI.
To analyze images and suggest or generate alt text descriptions
I haven't given it a shot yet. It would be awesome to see it integrated into ESP since we're not dealing with a massive amount of images.
Using Generative Adversarial Networks (GANs) to generate images, graphics, or illustrations.
At Powtoon, we've got this awesome AI solution tailored for presentations. While it's not directly tied to emails, any slide you create can easily be downloaded as an image or Gif. Our Script Writer feature assists users in crafting scenarios and finding slide inspiration. Meanwhile, AI Producer empowers users to generate animated videos effortlessly, just by providing a prompt.
AI-powered A/B testing algorithms that can help optimize email campaigns - subject lines, content, images, and other elements.
The ESP we utilize at Powtoon, Iterable, boasts internal functions that we primarily leverage for AB testing subject lines and preheader text.
I've recently heard of an interesting concept where you upload two different designs or texts to ChatGPT and solicit predictions for the results of an AB test. It's definitely something I'm eager to give a shot!
How AI algorithms can automatically adapt email layouts and designs to ensure they look perfect on various devices and screen sizes.
We've had an amazing experience with Iterable. I've heard that Stripo offers comprehensive support in this area. In my opinion, it would also be beneficial to add an accessibility checker to the email AI Builder.
AI-powered data analytics tools and integrating AI with CRMs can deliver data that can be used to create dynamic email designs to improve the personalization game.
AI tools that help to convert design files in all formats to a coded email. What level of perfection do they deliver? Once the tool provides the code, how much time is required manually to make the email template client-ready?
I recently had a conversation with the Kombai team; they specialize in converting Figma designs into code. Although I haven't given it a try yet, it sounds promising.
AI-powered tools that help to do various types of integrations- integrating email in ESP, integrating different tools with the ESP, etc.
The example I have is about delving deeper into the ESP functionalities. We use ChatGPT to analyze and refine the JSON code utilized in our journeys. This journey is designed to generate new member attributes or profile fields based on backend events. Essentially, it empowers marketers to create specific fields as needed.
None. Never had a need to generate alt tags at such a scale that automation would be required.
Midjourney, but so far only for limited editorial purposes. Maybe if a tool could be relied upon to generate on-brand imagery, that could be useful. Although some AI platforms can be fed images to learn a style, I don't believe any platform is anywhere near good enough to then generate usable images consistently. So, for now, manual graphic design is the only realistic option.
None. We use ESP split-testing tools combined with traditional mailing reports via the platform or SQL.
We tried ChatGPT extensively for this purpose, and published the results – which were junk. It would be interesting to see what Github Copilot can produce, but again as a matter of research rather than a business need.
ChatGPT can be useful for this. As we work in various email platforms, ChatGPT can be a quick point of reference for proprietary scripting languages like AMPscript. It's far from perfect but it can usually at least point us in the right direction.
None. Hand-coding all the way
Tool: DALL-E
Use: I like using DALL-E for recognizing images and generating alt text descriptions. It's great for creating detailed and appropriate alt text for accessibility or boosting SEO. DALL-E uses deep learning to understand and describe visual content really well.
Tool: Midjourney
Use: Midjourney is my go-to for creating visual images using Generative Adversarial Networks. It's perfect for generating unique images, graphics, or illustrations. Midjourney is fantastic for creative projects that need unique visual assets.
Tool: GetResponse
Use: I use GetResponse for A/B testing in email campaigns. The email marketing platform is great for testing different subject lines to see what resonates best with the audience. As an employee of GetResponse, I know this might sound promotional, but it's truly a solid choice for improving email marketing performance.
Tool: AI Email Generator
Use: I use GetResponse's AI Email Generator to quickly create responsive email content. The tool uses AI to generate unique, tailor-made email content, which fits perfectly on various devices and screen sizes. It's incredibly handy for quickly creating the first draft of a newsletter, saving a lot of time and effort.
Tool: GetResponse
Use: I rely on GetResponse for creating dynamic emails. Their platform allows you to personalize emails using dynamic content blocks. You can easily tailor different parts of your email to suit various audience segments based on their interests or behaviors. It's a fantastic way to enhance personalization, boost engagement, and improve conversion rates.
Tool: Stripo
Use: Stripo is my choice for converting design files into coded emails. It does a pretty solid job, though sometimes a bit of manual tweaking is needed to make the email template client-ready. The time needed for adjustments depends on the design complexity and how perfect the conversion is.
Tool: Zapier
Use: I like using Zapier for integrating email-related services with other tools and platforms. It's great for connecting ESPs with different applications, ensuring smooth workflows and communication between various tools.
We use GrooveAI almost exclusively. GrooveAI has various protocols built in exclusive to GrooveAI. Our AI tools were created for digital marketers, so we have all the tools that benefit digital marketers and content creators. We have seamless email protocols and more.
With Groove AI we have all the marketing tools any marketer would need all under one roof.
Dark horse surprise - clarifai.com
I've used Clarifai's APIs for a number of years – way before the ChatGPT headline explosion.
I think visual communication is too important to delegate — customers process visuals before language, so don't hand over "first impressions" to save a few pennies.
No experience. Phrasee just got $100M reasons for marketers to take a look though.
No experience.
SocialSignal.ai re-thinks the data layer of personalization. 1st-party ESP/CRM data is generally too sparse to support meaningful segmentation.
No experience, I'm a bit sanguine about the idea.
AI to help developers with integrations, sure. Standouts: GitHub Copilot, Google Gemini. AI to create integrations — again, color me dubious. AI to run integration — nah, does not pencil. Integration is about finding the simplest way to bridge complex things, not about putting more complexity in between them to fake simplicity.
Although I have seen very impressive results from MidJourney, I have not worked with this myself. What I like about what I have seen there is that it is possible to have a certain character in the form of a person to return in following image generations. So you could build an identity around something you have returned in all your images. That’s a nice feature distinguishing MidJourney from others. What I often use myself is Microsoft Copilot, and the Autofill-functionality in Photoshop. These two together give me a great toolkit for generating awesome images. And with Photoshop I can also modify images coming from Copilot to different image sizes, as that’s not working as well as I would like in Copilot itself. Copilot generates mostly square images. When I need a rectangle shape for example I paste the Copilot-square in the rectangle shape in Photoshop and then have Photoshop AI fill the blank spaces of the image so that it eventually turns into exactly what I need.
I have used Copilot, ChatGPT and custom GPT’s for generating email content variations. They come up with pretty decent suggestions, and especially if you write your prompt very well it will get you nice results. For example when you ask it to use the Cialdini’s scarcity principle, it will do exactly that. However, I often find that writing a very specific prompt takes about as much time as coming up with the results yourself. Also when you ask for campaign or content ideas, it comes up with a lot of very common suggestions and best practices. It gets you going quickly and saves time in the brainstorming phase. But I am yet to be blown away by an amazing idea that has never been done before, coming from AI.
This is a feature that I expect a lot from. At this time though, it is not at the level where it should be. Several techniques are not being used by AI and it’s often not optimizing for Outlook. It can be a result of a prompt that I made that’s not specific enough of course, but for now I don’t trust it. Yet. I expect AI to learn quickly and maybe be able to use everything the way it’s intended in the near future. For now, just like with the A/B-testing, it can get you a nice quick start, but you have to dot all the I’s yourself if you want to have the best possible result.
Tools: GPT-4 Vision Preview and GPT-4o
Experience: We use these tools in our AI Hub to generate email campaigns based on task-specific references. By developing a vocabulary representing each email as a list of content modules with defined structures, we utilize GPT-4 Vision Preview and GPT-4o to describe email images or PDFs accurately. This process creates detailed descriptions that facilitate email creation using our content library.
Future possibilities: The tools’ accuracy allows us to build high-quality emails from designs in Canva, Figma, or other tools. We also plan to address accessibility issues by generating alt text and titles for images in emails. Additionally, we have used this method to analyze thousands of emails from public resources to learn the most popular email structures and compose our library of modules.
Challenges: Integrating with various design tools and maintaining an updated vocabulary can be challenging. It’s crucial to have a rich library of supported modules to match the designs it reads and to find the right way to generate prompts that accurately explain the banners in the emails.
Overall, GPT-4 Vision Preview and GPT-4o significantly enhance our email campaign generation process and pave the way for improved accessibility.
Tools: DALL·E by OpenAI, Midjourney, Leonardo, Adobe
Experience: We utilize AI to address three main tasks in image creation for email marketing: Banner Creation, Creating Creatives, and Image Enhancement.
Banner Creation: Email banners are critical for engaging recipients. Effective banners need to be visually appealing and relevant. However, image generation tools often struggle with text placement within banners. To overcome this, we use a predefined structure and generate each part separately, employing a “divide and conquer” strategy for complex tasks. This ensures that key elements such as headlines, promotional texts, CTAs, and images are seamlessly incorporated and aligned with different content types and campaign goals.
Creating creatives: Beyond standard banners, AI helps produce a variety of creative assets like full-body images, collages, and themed graphics. These assets can be used across different sections of an email to enhance engagement and storytelling. This extends AI’s capabilities to generate promotional images, informational graphics, or interactive elements designed to captivate users.
Image enhancement: AI-powered tools improve the quality and effectiveness of existing images in email campaigns. Enhancements may include optimizing resolution, adjusting colors, and refining image details to ensure they are visually striking and on-brand. Another use is placing images into the context of the email, ensuring they align with the overall design and messaging.
Challenges: While AI-generated visuals offer significant benefits, there are several challenges. Ensuring that the generated images align precisely with the intended campaign theme can be difficult. Additionally, the cost of using these advanced AI tools can be high, and there are often limitations on the number of requests you can make. Although the technology is developing rapidly, it is still too early to rely on it for fully automating this process.
Experience: We use OpenAI to optimize email campaigns by testing subject lines, content, CTAs, and overall email structure and strategy.
Creative generation and direction: OpenAI excels at generating different variations and aiding brainstorming sessions. We can control the testing direction by applying principles like Cialdini’s persuasion techniques or experimenting with different tones of voice, with or without emojis. This approach allows us to reuse successful elements in future emails efficiently.
Expert predictions: OpenAI serves as an expert by predicting test results based on our audience knowledge and previous campaigns with a high degree of accuracy. This predictive capability helps us accelerate testing and increase the number of successful outcomes.
Some tasks, in my opinion, are not yet suitable for delegation to AI. Especially tasks that can be resolved once, controlled, and reused repeatedly, but have a high risk of errors.
I believe such tasks include email coding, responsiveness, and brand consistency. Due to the numerous undocumented peculiarities of each email client and the difficulty in controlling the results, it’s better to manually define the modules and their structure, test them across all email clients, observe rendering specifics, and maintain all branding details.
The key goal of using AI is to send the right email to the right people at the right time. Today, it’s very difficult to delegate these tasks to public GenAI methods due to the complexity of control, high costs, and speed. Instead, we actively use various internal ML approaches for recommendations in emails, finding the best send times, and personalizing content.
This is a crucial task that we’ve addressed successfully. Using AI tools, we can convert email designs into HTML emails effectively, ensuring they meet all design and functionality requirements. I described the experience in the first section.
AI significantly helps optimize disparate data into a unified CDP (Customer Data Platform). Tasks like matching parameters between database fields, querying across systems, and maintaining data formats and accuracy are greatly facilitated by AI. This ensures seamless data integration and consistency, improving overall data management and utility.