When AI Can’t Read What We Publish

Why interactive magazines create an unexpected blind spot — and what we’re building next

Over the past year, we’ve been closely observing how AI systems interact with industry content — not in theory, but in practice.

One issue keeps surfacing.

Many AI systems cannot meaningfully read interactive digital publications, even when those publications are excellent for human readers.

A common example is Flipsnack-based magazines.

From a human perspective, these publications work beautifully. From an AI perspective, however, they are largely opaque.

The practical problem (not a criticism)

AI systems do not “open” a Flipsnack magazine the way a human does. They can’t flip pages, navigate the player, or reliably extract the underlying text.

As a result, from an AI point of view:

• the content cannot be searched properly,
• cannot be compared or summarised across issues,
• and often cannot be referenced at all.

This isn’t a content-quality issue. It’s a delivery-format issue.

And it matters more now than it did even a year ago.

Why this suddenly matters

AI is increasingly involved in:

• discovery and search,
• summarisation and research,
• benchmarking and comparison,
• and early-stage credibility assessment.

In these contexts, what AI can’t read may as well not exist — regardless of how good the content is for humans.

That creates an unintended risk for publishers and industry bodies who are doing everything right for their human audience, but whose work is becoming partially invisible in an AI-mediated environment.

What we’re building (and what we’re not)

At Evan Tech, we’re developing a parallel publishing structure that is intentionally readable by AI, while remaining usable and clear for humans.

This is not intended to replace Flipsnack or other human-first publishing formats.

In fact, we believe the opposite approach is emerging as the wiser one:

Important content should increasingly exist in parallel – one format optimised for human engagement, – another structured so AI systems can reliably read, interpret, and reference it.

A different experience for humans too

An unexpected benefit of this approach is how humans interact with the content.

Instead of being limited to reading issue by issue, users can:

• search across multiple issues,
• find specific topics, themes, or terms,
• trace how ideas and positions evolve over time.

This becomes difficult once content is locked inside page-turning interfaces — even though those interfaces are excellent for linear reading.

Coming soon — and open to collaboration

This project is still evolving, and we’re sharing it early because we think it raises industry-wide questions, not just technical ones.

If you’re involved in:

• publishing,
• industry magazines,
• associations,
• or long-form content that matters beyond the moment it’s read,

this is a conversation worth having.

We’re not replacing tools. We’re trying to make important work legible to both humans and machines — at the same time.

More to come soon.