AI in the modern inbox: what's happening?

This isn’t a crisis. But it does require some adjustment. Here’s what’s happening across each platform, what it means in practice, and what we think is worth prioritising.

What’s changed: platform by platform

Apple Mail and iOS 18
Apple holds around 45% of the smartphone market, so changes to Apple Mail have an outsized effect on email marketing performance. iOS 18 introduced two features in particular that are worth understanding.

The first is automatic inbox categorisation. Apple Mail now sorts incoming emails into four tabs: Primary, Transactions, Updates, and Promotions. Marketing emails generally land in Promotions, which is not where subscribers spend most of their attention. This is similar to Gmail’s tab system, which many marketers have been navigating for years.

The second, and arguably more significant change, is AI-generated email summaries. Rather than displaying your preheader text, Apple Mail can show subscribers an AI-generated synopsis of your email content. The implication is straightforward: if the actual text of your email isn’t clear and substantive, the AI has little to work with, and the summary it generates may not represent your message well.

Apple also introduced digest bundling, which groups multiple emails from the same sender. Your most recent email sits at the front; older ones are pushed back. Campaign sequences that rely on building across multiple messages are affected by this.

On the positive side, Apple’s Branded Mail feature places a verified sender logo next to your emails in the inbox. It supports deliverability indirectly by reinforcing authentication and user trust signals, which can help your messages avoid spam filters and earn better inbox placement over time. It’s free to set up, improves visual recognition, and is one of the more straightforward wins available from this update.

Gmail and Gemini
Google has been integrating Gemini, its AI assistant, into Gmail. The most visible result for email marketers is Gmail AI Overviews: automatically generated summaries of emails and email threads, presented at the top of a conversation.

The practical consequence is that your email is now processed by AI before most subscribers engage with it. Gmail analyses the text content of your email, extracts what it considers the key points, and presents those as a quick overview. Emails that are built primarily from images, with minimal HTML text, give the system very little to analyse. In those cases, the summary may draw from footer copy or navigation links rather than your actual message.

Inbox placement is also becoming more dynamic. Gmail has long used engagement signals to determine relevance, and deeper AI integration makes those signals more consequential. Where & how your email appears in the inbox is increasingly a function of how subscribers have historically interacted with your content, not just whether it was delivered successfully.

It’s also worth noting that Gmail is moving towards natural language search, where subscribers can query their inbox conversationally. That’s a shift in how your emails need to be written, which we’ll discuss below.

Microsoft Outlook and Copilot
Microsoft has integrated Copilot into Outlook, giving subscribers a summarise button that generates an instant overview of any email. The same logic applies here as with Gmail: if your email’s core message isn’t present in readable text, the summary won’t reflect it accurately.

Outlook still uses a broadly chronological inbox for now, but that is expected to evolve as AI-driven prioritisation develops further.

What we’ve learned

AI reads text, not images
This is probably the most important practical insight from the past year. Many email templates lean heavily on imagery, with the key message, offer, or call to action embedded in a graphic rather than in HTML text. That approach still works for the human eye, but it leaves AI systems with almost nothing to interpret.

When we ran test emails through AI summarisation tools, image-heavy emails produced summaries that were either generic or drawn from the least relevant parts of the email, such as “View in browser” links or legal footer text. Text-rich emails produced summaries that accurately reflected the offer and intent.

Intent is what AI categorises, not style
AI systems categorise emails based on what they are for, not how they look or how creatively they’re written. A beautifully designed email with a vague or unfocused message is harder for AI to interpret than a plain, clearly structured one.

This also matters for search. As subscribers use natural language queries to find emails in their inbox, your content needs to contain the kinds of words and phrases someone would actually use to search for it. An email promoting a theatre production in London should contain phrases like “West End tickets” or “shows this summer” in its actual text, not just in its imagery.

Engagement signals are compounding
The AI inbox doesn’t just affect how your current email is received. It affects future ones too. When subscribers regularly engage with your emails, their inbox AI learns to treat your content as relevant. When they don’t, it learns the opposite. This creates a positive feedback loop for marketers with genuinely engaged lists, and a compounding problem for those sending to large, inactive audiences.

List hygiene has always mattered for deliverability. It now matters for inbox positioning too.

Accessibility and AI readability overlap
Email accessibility best practice, such as using semantic heading structure, writing descriptive link text, adding meaningful alt text to images, and keeping layouts logically ordered, turns out to also improve how well AI systems can interpret your emails. The practices that make emails usable for screen readers make them more legible to machine learning systems too.

Investing in accessible email design now pays off in multiple ways.

Five things worth doing now

1. Audit your templates for text content
   If a significant portion of your key messages live inside images, consider whether that content can be replicated or supported in HTML text. You don’t need to redesign everything, but your most important copy should be present as readable text.
2. Write the opening of your email as if it’s the only part AI will read
   State your main message or offer early and clearly, in plain language. Long warm-up paragraphs, vague teasers, or subject lines that rely on curiosity over clarity are harder for AI to summarise accurately.
3. Test how your emails are summarised
   Send a test to your Outlook inbox and use the Copilot summarise button. Alternatively, paste your email text into an AI tool and ask for a summary. Does the output reflect your actual message? If not, that’s useful information about how your content is structured.
4. Set up Branded Mail in Apple Mail besides the usual authentications and BIMI
   It’s a free trust signal that improves your visibility in one of the most widely used email clients. If you haven’t done this yet, it’s a straightforward place to start. Note:though there is no proven correlation between these elements and AI summaries in the inbox, they definitely help getting your emails in the inbox and increase trust with those who read your emails.
5. Revisit your engagement strategy alongside your list hygiene
   The AI inbox rewards relevance over volume. A smaller, more engaged list is likely to perform better in this environment than a large one with poor engagement rates.

The bigger picture

The changes happening in the inbox are, in some ways, an amplification of principles that good email marketing has always rested on. Relevant content, clear messaging, and genuine engagement with your audience have always mattered. What AI does is make those qualities more mechanistically important and more immediately measurable in terms of inbox placement and visibility.

For marketers who have been building their email programmes around real value and genuine audience relationships, this is more of a confirmation than a disruption. For those who have been relying on volume or visual impact to carry their results, the adjustment will be more significant.

Sources:

• Gemini in Gmail: dit verandert er in je inbox
• Branded Mail in Apple Business
• iOS18 and iPhone
• Apple’s iOS 18 email changes
• Apple’s doubling down on AI: what does iOS 18 mean for email marketing?
• Create emails for humans and AI
• Gmail AI Overviews
• The AI Inbox: what is it and what do you need to take into account?

Beata Linz

Brand – Digital – Email Marketing Consultant, Global CRM Director en International Speaker and Event Organiser

Tom Blijleven

Director of Marketing Operations bij Spotler Group