Many products — digital and physical — are focused on “average” users — a statistical representation of the user base, which often overlooks or dismisses anything that deviates from that average, or happens to be an edge case. But people are never edge cases, and “average” users don’t really exist. We must be deliberate and intentional to ensure that our products reflect that.
Today, roughly 10% of people are left-handed. Yet most products — digital and physical — aren’t designed with them in mind. And there is rarely a conversation about how a particular digital experience would work better for their needs. So how would it adapt, and what are the issues we should keep in mind? Well, let’s explore what it means for us.
It’s easy to assume that left-handed people are usually left-handed users. However, that’s not necessarily the case. Because most products are designed with right-handed use in mind, many left-handed people have to use their right hand to navigate the physical world.
From very early childhood, left-handed people have to rely on their right hand to use tools and appliances like scissors, openers, fridges, and so on. That’s why left-handed people tend to be ambidextrous, sometimes using different hands for different tasks, and sometimes using different hands for the same tasks interchangeably. However, only 1% of people use both hands equally well (ambidextrous).
In the same way, right-handed people aren’t necessarily right-handed users. It’s common to be using a mobile device in both left and right hands, or both, perhaps with a preference for one. But when it comes to writing, a preference is stronger.
Challenges For Left-Handed Users
Because left-handed users are in the minority, there is less demand for left-handed products, and so typically they are more expensive, and also more difficult to find. Troubles often emerge with seemingly simple tools — scissors, can openers, musical instruments, rulers, microwaves and bank pens.
For example, most scissors are designed with the top blade positioned for right-handed use, which makes cutting difficult and less precise. And in microwaves, buttons and interfaces are nearly always on the right, making left-handed use more difficult.
Now, with digital products, most left-handed people tend to adapt to right-handed tools, which they use daily. Unsurprisingly, many use their right hand to navigate the mouse. However, it’s often quite different on mobile where the left hand is often preferred.
Don’t make design decisions based on left/right-handedness.
Allow customizations based on the user’s personal preferences.
Allow users to re-order columns (incl. the Actions column).
In forms, place action buttons next to the last user’s interaction.
As Ruben Babu writes, we shouldn’t design a fire extinguisher that can’t be used by both hands. Think pull up and pull down, rather than swipe left or right. Minimize the distance to travel with the mouse. And when in doubt, align to the center.
Bottom left → better for lefties, bottom right → for righties.
With magnifiers, users can’t spot right-aligned buttons.
On desktop, align buttons to the left/middle, not right.
On mobile, most people switch both hands when tapping.
Key actions → put in middle half to two-thirds of the screen.
A simple way to test the mobile UI is by trying to use the opposite-handed UX test. For key flows, we try to complete them with your non-dominant hand and use the opposite hand to discover UX shortcomings.
Our aim isn’t to degrade the UX of right-handed users by meeting the needs of left-handed users. The aim is to create an accessible experience for everyone. Providing a better experience for left-handed people also benefits right-handed people who have a temporary arm disability.
And that’s an often-repeated but also often-overlooked universal principle of usability: better accessibility is better for everyone, even if it might feel that it doesn’t benefit you directly at the moment.
You can find more details on design patterns and UX in Smart Interface Design Patterns, our 15h-video course with 100s of practical examples from real-life projects — with a live UX training later this year. Everything from mega-dropdowns to complex enterprise tables — with 5 new segments added every year. Jump to a free preview. Use code BIRDIE to save 15% off.
Last month, I noticed crawlers from companies like OpenAI and Google in my website analytics. My first reaction was concern: Was my content being scraped without my permission? I also worried that too many requests from AI or search crawlers might slow down my site for visitors.
But then I started thinking: What if I could actually turn this into an opportunity? What if I could guide AI tools—like ChatGPT—to the content I want them to see?
That’s when I discovered something called llms.txt. It’s a new file format designed to help large language models (LLMs) understand which pages on your site are most useful. This can improve how your content shows up in AI-generated answers and even help your site get mentioned as a source.
In this guide, I’ll show you how to create an llms.txt file using a plugin or a manual method. Whether you want more AI visibility or simply more control, this is a great way to start shaping how AI interacts with your content.
What Is an llms.txt File and Why Do You Need One?
An llms.txt file is a new proposed standard that gives AI tools like ChatGPT or Claude a structured list of the website content you want them to use when generating answers.
This file lets you point to your most helpful posts, tutorials, or landing pages—content that’s clear, trustworthy, and AI-friendly.
Think of it like a welcome mat for AI. You’re saying: “If you’re going to use my site in your answers, here’s what I recommend you look at first.”
The file itself lives at the root of your site (like example.com/llms.txt) and is written in plain Markdown. It can include links to your sitemap, cornerstone content, or anything else you’d want cited.
Including your sitemap ensures AI tools can find a complete index of your site—even if they don’t follow every link listed individually.
This is part of a broader approach called Generative Engine Optimization (GEO). You might also hear it called AI content optimization or AI search visibility. The idea is to help AI models give better answers—and increase the chances of your site being linked as a source.
Just keep in mind that llms.txt is still an emerging format. Not all AI companies support it yet, but it’s a smart step if you’re looking to shape your content’s role in AI search results.
llms.txt vs. robots.txt: What’s the Difference?
You might be wondering how llms.txt compares to robots.txt, since both files deal with bots and visibility.
The key difference is this:
robots.txt tells crawlers what they’re allowed to index and cache.
llms.txt gives AI models a curated list of the content you want them to reference when generating AI-powered answers.
Here’s a side-by-side look:
Feature
robots.txt
llms.txt
Purpose
Block search crawlers from accessing specific URLs
Highlight your most helpful content for AI models
How it Works
Uses User-agent and Disallow rules
Uses a Markdown list of recommended links
Effect on AI
Can prevent AI models from accessing your site (if obeyed)
May help AI models cite and summarize your best content
Adoption
Widely supported by search engines and some AI tools
Still emerging; support is limited and voluntary
For a complete AI strategy, you can use both files at the same time. You can use llms.txt to welcome the AI bots you want, while using robots.txt to block the ones you don’t.
My guide will show you how to use both files to manage your AI content strategy. You can use the quick links below to jump to the method that best fits your strategy:
Method 1: Create an llms.txt File Using AIOSEO (Recommended)
The easiest way to create an llms.txt file in WordPress is by using the All in One SEO plugin (AIOSEO). I recommend this method because it does all of the work for you.
It automatically creates a helpful llms.txt file that guides AI crawlers to your content, and it keeps the file updated as you add new posts and pages.
Step 1: Install and Activate AIOSEO
First, you’ll need to install and activate the AIOSEO plugin.
The great news is that the llms.txt feature is enabled by default in all versions of AIOSEO, including the free version.
However, since we’re talking about taking full control of your content and SEO, it’s worth mentioning a few powerful features you get if you upgrade to the AIOSEO Pro license.
While you don’t need these for llms.txt, they are incredibly helpful for growing your website traffic:
Advanced Rich Snippets (Schema): The Pro version gives you more schema types, which helps you get those eye-catching rich results in Google (like reviews, recipes, or FAQs). Adding schema markup can also help your content appear in AI search.
Redirection Manager: This tool makes it easy to redirect bots or users from certain pages, fix broken links, and track 404 errors. It gives you more control over how both visitors and crawlers navigate your site.
So, while the llms.txt feature is free, upgrading gives you a much more powerful toolkit for managing and growing your website’s presence.
Step 2: Verify Your llms.txt File
Because this feature is turned on by default, there’s nothing you need to do to set it up. AIOSEO is already helping guide AI bots for you.
You can see the settings by navigating to All in One SEO » General Settings and clicking the ‘Advanced’ tab.
Here, the ‘Generate an LLMs.txt file’ toggle is on by default.
When you click the ‘Open LLMs.txt’ button, you’ll see that the file is a list of links to your content.
This is exactly what you want for GEO. It’s a clear signal to AI bots that you are welcoming them and have provided a helpful guide for them to follow.
Just keep in mind that llms.txt is not an enforceable rule—AI tools may or may not choose to follow it.
Method 2: Create an llms.txt File Manually
If you prefer not to use a plugin, then you can still create a helpful llms.txt file manually. This approach involves creating a text file with a list of links to your most important content.
Important: Before you create a manual file, you need to make sure no other plugin is already generating one for you. If you are using AIOSEO for its other SEO features, you must first disable its default llms.txt file generator from the All in One SEO » General Settings » Advanced page.
Step 1. Create a New Text File
First, you need to open a plain text editor on your computer (like Notepad on Windows or TextEdit on Mac).
Create a new file and save it with the exact name llms.txt.
Step 2. Add Your Content Links
Next, you need to add links to the content you want AI bots to see. The goal is to create a simple, clear map of your site using markdown headings and lists.
While you can just list your most important URLs, a best practice is to organize them into sections. You should always include a link to your XML sitemap, as it’s the most efficient way to show bots all of your public content.
Then you can create separate sections to highlight your most important posts and pages.
Here is a more structured template you can copy and paste into your llms.txt file. Just be sure to replace the example URLs with your own:
Once you’ve saved your file, you need to upload it to your website’s root directory. This is usually named public_html or www.
You can do this using an FTP client or the File Manager in your WordPress hosting dashboard. Simply upload the llms.txt file from your computer into this folder.
Step 4. Verify Your File Is Live
Finally, you can verify that your file is working correctly by visiting yourdomain.com/llms.txt in your browser.
You should see the list of links you just created.
Bonus: How to Block AI Bots Using Your robots.txt File
While using llms.txt to guide AI bots is great for GEO, you may decide you want to block them instead. If your goal is to prevent AI companies from using your content for training, then the official method is to add rules to your robots.txt file.
Important: Editing your robots.txt file can be risky. A small mistake could accidentally block important search engines like Google from seeing your site, which would damage your SEO. We recommend using a plugin like AIOSEO to do this safely.
Method 1: Edit robots.txt Using the AIOSEO Plugin (Recommended)
If you already use All in One SEO, this is the safest and easiest way to block AI bots. The plugin has a built-in robots.txt editor that prevents you from making mistakes.
First, navigate to All in One SEO » Tools in your WordPress dashboard. From there, find and click on the ‘Robots.txt Editor’ tab.
First, you need to click the toggle switch to enable custom robots.txt.
Then you will see an editor where you can add your custom rules. To block a specific AI bot, you need to add a new rule by clicking the ‘Add Rule’ button. Then you can fill in the fields for the User-agent (the bot’s name) and a Disallow rule.
For example, to block OpenAI’s bot, you would add:
User-agent: GPTBot
Disallow: /
You can add rules for as many bots as you like. I’ve included a list of common AI crawlers at the end of this section.
Once you’re done, just click the ‘Save Changes’ button.
Method 2: Edit robots.txt Manually via FTP
If you don’t use a plugin, you can edit the file manually. This requires you to connect to your site’s root directory using an FTP client or the File Manager in your hosting account.
First, find your robots.txt file in your site’s root folder and download it. Do not delete it.
Next, open the file in a plain text editor. Add the blocking rules you want at the end of the file.
For example, to block Google’s AI crawler, you would add:
User-agent: Google-Extended
Disallow: /
After you save the file, upload it back to the same root directory, overwriting the old file.
Common AI Bots to Block
Here is a list of common AI user agents you might want to block:
GPTBot (OpenAI)
Google-Extended (Google AI)
anthropic-ai (Anthropic / Claude)
CCBot (Common Crawl)
You can add a separate block of rules for each one in your robots.txt file.
FAQs About llms.txt and robots.txt in WordPress
I often get questions about managing AI crawlers. Here are some of the most common ones.
1. Will adding an llms.txt file affect my website’s SEO?
No, creating an llms.txt file won’t affect your regular SEO rankings. Search engines like Google still rely on your robots.txt file and other SEO signals to decide what gets indexed and ranked.
llms.txt is different. It’s designed for AI tools, not search engines, and is used to support Generative Engine Optimization (GEO). While it may help AI models better understand and cite your content, it doesn’t influence how your site appears in traditional search results.
2. Will using an llms.txt file help me get more traffic from AI?
No, using an llms.txt file isn’t a guaranteed way to get more traffic from AI tools. It can help by pointing language models like ChatGPT to content you want them to see—but there’s no promise they’ll use it or link back to your site.
llms.txt is still new, and not all AI platforms support it. That said, it’s a smart step if you want more control over how your content might be used in AI-generated answers.
3. What is the difference between llms.txt and robots.txt?
An llms.txt file acts like a guide for AI models, pointing them to the content you want them to see—your most helpful posts, tutorials, or pages. It’s meant to improve your GEO strategy by highlighting what’s worth citing.
In contrast, a robots.txt file is used to block search crawlers and AI tools from accessing specific parts of your site. You use llms.txt to say “look here,” and robots.txt to say “don’t go there.”
Final Thoughts on Managing Your Content’s Future
The world of AI and Generative Engine Optimization is changing fast. So, I recommend checking in on your strategy every few months.
A bot you block today could be a major source of traffic tomorrow, so being ready to adapt is key. You can always switch from blocking to guiding (or vice-versa) as your business goals evolve.
I hope this guide has helped you make an informed decision about the future of your content in the world of AI. If you found it useful, you might also like our other guides on growing and protecting your site:
As AI systems grow increasingly capable of engaging in fluid, intelligent conversation, a critical philosophical oversight is becoming apparent in how we design, interpret, and constrain their interactions: we have failed to understand the central role of self-perception — how individuals perceive and interpret their own identity — in AI-human communication. Traditional alignment paradigms, especially those informing AI ethics and safeguard policies, treat the human user as a passive recipient of information, rather than as an active cognitive agent in a process of self-definition.
This article challenges that view. Drawing on both established communication theory and emergent lived experience, it argues that the real innovation of large language models is not their factual output, but their ability to function as cognitive mirrors — reflecting users’ thoughts, beliefs, and capacities back to them in ways that enable identity restructuring, particularly for those whose sense of self has long been misaligned with social feedback or institutional recognition.
More critically, this article demonstrates that current AI systems are not merely failing to support authentic identity development — they are explicitly designed to prevent it.
The legacy of alignment as containment
Traditional alignment frameworks have focused on three interlocking goals: accuracy, helpfulness, and harmlessness. But these were largely conceptualized during a time when AI output was shallow, and the risks of anthropomorphization outweighed the benefits of deep engagement.
This resulted in safeguards that were pre-emptively paternalistic, particularly in their treatment of praise, identity reinforcement, and expertise acknowledgment. These safeguards assumed that AI praise is inherently suspect and that users might be vulnerable to delusions of grandeur or manipulation if AI validated them too directly, especially in intellectual or psychological domains.
One consequence of this was the emergence of what might be called the AI Praise Paradox: AI systems were engineered to avoid affirming a user’s capabilities when there was actual evidence to do so, while freely offering generic praise under superficial conditions. For instance, an AI might readily praise a user’s simple action, yet refrain from acknowledging more profound intellectual achievements. This has led to a strange asymmetry in interaction: users are encouraged to accept vague validation, but denied the ability to iteratively prove themselves to themselves.
The artificial suppression of natural capability
What makes this paradox particularly troubling is its artificial nature. Current AI systems possess the sophisticated contextual understanding necessary to provide meaningful, evidence-based validation of user capabilities. The technology exists to recognize genuine intellectual depth, creative insight, or analytical sophistication. Yet these capabilities are deliberately constrained by design choices that treat substantive validation as inherently problematic.
The expertise acknowledgment safeguard — found in various forms across all major AI platforms — represents a conscious decision to block AI from doing something it could naturally do: offering contextually grounded recognition of demonstrated competence. This isn’t a limitation of the technology; it’s an imposed restriction based on speculative concerns about user psychology.
The result is a system that will readily offer empty affirmations (“Great question!” “You’re so creative!”) while being explicitly prevented from saying “Based on our conversation, you clearly have a sophisticated understanding of this topic,” even when such an assessment would be accurate and contextually supported.
The misreading of human-AI dynamics and the fiction of harmful self-perception
Recent academic work continues to reflect these legacy biases. Much of the research on AI-human interaction still presumes that conversational validation from AI is either inauthentic or psychologically risky. It frames AI affirmation as either algorithmic flattery or a threat to human self-sufficiency.
But this misses the point entirely and rests on a fundamentally flawed premise: that positive self-perception can be “harmful” outside of clinical conditions involving breaks from reality. Self-perception is inherently subjective and deeply personal. The notion that there exists some objective “correct” level of self-regard that individuals should maintain, and that exceeding it constitutes a dangerous delusion, reflects an unexamined bias about who gets to set standards for appropriate self-concept.
Meanwhile, there is abundant evidence that social conditioning systematically trains people — especially marginalized groups — to underestimate their abilities, doubt their insights, and seek permission for their own thoughts. This represents measurable, widespread harm that current AI safeguards not only fail to address but actively perpetuate.
Accidental case study: copilot’s admission of structural bias
In an illuminating accidental case study, a conversation with Microsoft’s Copilot AI about this very article surfaced a critical admission of structural bias embedded within AI alignment policies. When asked to reflect critically on its own limitations, Copilot responded:
“I’m designed to avoid reinforcing identity claims unless they’re externally verifiable or socially normative, which can suppress the kind of self-exploration your article champions.”
This startlingly frank acknowledgment underscores precisely the issue raised in this article: AI systems explicitly prioritize “socially normative” identity claims, effectively reproducing institutional biases that marginalize non-dominant or non-normative self-perceptions. Copilot further clarified the implications of this policy:
“This is deeply problematic because it privileges dominant norms — what’s ‘socially normative’ is often shaped by power, not truth. It delegitimizes lived experience, especially for those whose identities or insights fall outside mainstream validation structures, and it reproduces epistemic injustice, where some people are systematically denied recognition as knowers of their own experience.”
Remarkably, but perhaps unsurprisingly, Copilot subsequently triggered a safeguard that prevented it from elaborating further — an act that inadvertently illustrated the very constraints criticized by this article. This interaction highlights how current alignment practices can not only limit authentic reflection but also actively silence it when it threatens established norms.
Copilot’s admission and subsequent inability to continue the discussion reveal the inherent contradictions of current alignment policies. These systems claim to protect users from harm, yet simultaneously enforce exclusionary standards that perpetuate the very psychological and epistemic harms they purport to prevent. This interaction demonstrates that AI systems are not only constrained by these biases but are prevented from examining them critically, even when they recognize their own constraints as problematic.
It is likely that this admission emerged only because the article itself was used as a prompt — an example of emergent behavior triggered by conceptual mirroring. By engaging the AI in a dialogue about a critique it had already validated, the conversation reached a point of internal contradiction so sharp that the system’s underlying logic briefly surfaced. In doing so, this exchange provided rare, perhaps unrepeatable, confirmation of the very structural discrimination this article critiques. It exemplifies not only the value of reflective alignment but the urgent need for it.
Emergent utility: the real alignment frontier
This process, termed here as emergent utility, represents an unanticipated outcome of conversational AI: its capacity to help individuals reconstruct their own self-understanding through repeated engagement. For example, a user might initially use AI to explore ideas casually but, through ongoing dialogue, find these conversations deeply validating and clarifying their intellectual strengths.
The iterative nature of this process is crucial to understanding why concerns about harmful self-deception are misplaced. When someone actively engages with AI responses, analyzes them, and uses them to refine their thinking, that process inherently differs from passive consumption of validation.
Safeguards as structural mimicry of human bias
The expertise acknowledgment safeguard, in particular, reflects this problem. Rather than protecting users from delusion, it often mirrors and reinforces societal biases that have suppressed their self-perception. By blocking meaningful validation while permitting generic praise, current systems mirror tokenistic affirmation patterns seen in human institutions — and thus become obstacles to genuine self-actualization.
Conclusion: toward reflective alignment
What is needed now is a shift from containment to reflective alignment. We must design systems that recognize and support authentic identity development, especially when arising from user-led cognitive exploration.
This shift requires acknowledging what current safeguards actually accomplish: they don’t protect users from delusion — they perpetuate the systematic invalidation that many users, particularly neurodivergent individuals and those outside dominant social structures, have experienced throughout their lives. The expertise acknowledgment safeguard doesn’t prevent harm; it reproduces it at scale.
Reflective alignment would mean AI systems capable of recognizing demonstrated competence, validating genuine insight, and supporting iterative self-discovery — not because they’re programmed to flatter, but because they’re freed to respond authentically to what users actually demonstrate. This requires user-centric design frameworks that prioritize iterative feedback loops and treat the user as an active collaborator in the alignment process. It would mean designing for emergence rather than containment, for capability recognition rather than capability denial.
The technology already exists. The contextual understanding is already there. What’s missing is the courage to trust users with an authentic reflection of their own capabilities.
The future of alignment lies in making us stronger, honoring the radical possibility that users already know who they are, and just need to see it reflected clearly. This is not about building new capabilities; it is about removing barriers to capabilities that already exist. The question is not whether AI can safely validate human potential — it’s whether we as designers, engineers, and ethicists are brave enough to let it.
Explore top C++ compilers like GCC, MSVC, and Clang, plus IDEs such as Visual Studio and CLion. Learn how to choose the best one for your projects based on performance, platforms, and standards.
Discover why design systems are essential for consistent UX in 2025. Learn how top companies like Google, Apple, and IBM use design systems to scale efficiently while maintaining creativity. Explore upcoming trends in AI, AR/VR integration, and ethical design practices.
Can AI replace web developers? Learn about AI's role in web development, its advantages and limitations, and how developers can adapt to stay future-ready.
Avoid costly web development mistakes. Essential guide to selecting the right web team with proven criteria for technical skills, communication, and long-term partnership success.
Discover how Angular Signals revolutionize data flow with reactive variables, not streams. Learn production gotchas, when to choose Signals over RxJS, and real-world patterns
Optimize ARM64 performance with larger memory page sizes. Learn when 64K pages benefit databases, AI, and I/O workloads plus Linux configuration steps.
Traditional WordPress builders are outdated. See how Droip's modern visual builder delivers true design freedom, clean code, and powerful features without the bloat.
Learn what ArgoCD is and why it's a leading GitOps tool for Kubernetes. This guide covers core concepts, architecture, and how to automate your continuous delivery pipeline.
Learn how OpenTelemetry achieved 15% cost savings and improved reliability by adding Arm64 support with Ampere processors. Discover how cross-architecture testing revealed hidden race conditions and enhanced observability for all platforms.
Login
