AI Agents, Accessibility, and an Uncertain Future for the Web

AI Agents, Accessibility, and an Uncertain Future for the Web

“One of the more interesting conversations I had at the W3C’s Technical Plenary and Advisory Committee (TPAC) meeting in Dublin last month concerned web accessibility and AI agents.”

I have spent a bit of time thinking about how agents interact with websites on behalf of people, particularly the questions around identity, delegated authority, and user control. But what I hadn’t done was think about it from the perspective of accessibility.

AI agents are increasingly using some of the same infrastructure that people with disabilities rely on to navigate the web. When done well, the instructions for various assistive technologies serve as instructions for AI agents. Which means that as businesses invest in making their websites easier for agents to use, those investments may also affect accessibility. The consequences could be positive, negative, or largely independent of accessibility improvements. We don’t yet know which direction this will take.

There is a good reason, then, that the Accessible Platform Architectures (APA) Working Group and the Accessible Rich Internet Applications (ARIA) Working Group appear in the coordination section of the proposed W3C Agentic Web Working Group charter.

What interests me about all this is how the commercial push to make websites more usable by AI agents might influence the technologies that people depend on for accessibility, and whether the standards community can recognize those consequences before the architectural decisions become difficult to change.

AI Agents, Accessibility, and an Uncertain Future for the Web - A Digital Identity Digest
A Digital Identity Digest
AI Agents, Accessibility, and an Uncertain Future for the Web
Loading
/

You can Subscribe and Listen to the Podcast on Apple Podcasts, or wherever you listen to Podcasts.

And be sure to leave me a Rating and Review!

Why AI Agents Are Interested in the Accessibility Tree

To understand the connection, it helps to know a little about how browsers make web content available to assistive technologies. I had to do quite a bit of research on this as I had no idea how far this rabbit hole could go, but I’m glad I did. It is both useful and relevant.

When a browser processes a web page, it creates a Document Object Model (DOM) tree representing the page’s elements, attributes, and text. From that structure, the browser also generates an accessibility tree, which represents information relevant to accessibility. (Thank you, Mozilla, for some great documentation on the subject.) Platform-specific accessibility APIs make that information available to assistive technologies, including screen readers.

As Mozilla’s documentation explains, accessibility tree objects expose properties such as a name, description, role, and state. A button, for example, might have a name describing its purpose, a role identifying it as a button, and a state indicating whether it is enabled or disabled. The accessibility tree can also represent available actions, such as following a link or entering text.

Much of this information comes from appropriately structured HTML. The Web Accessibility Initiative (WAI)-ARIA provides additional semantics where necessary, particularly for complex interface components. But adding ARIA attributes isn’t a substitute for using semantic HTML correctly or designing accessible interactions.

This structured information is also useful to AI agents. An agent trying to complete a form needs to identify the relevant fields, determine what information belongs in them, find the submission control, and recognize whether the operation succeeded. Accessibility-related information can help it understand those controls without relying entirely on visual interpretation.

Agents have several ways to interact with websites, however. They may inspect the DOM, consume accessibility snapshots exposed through browser automation tools, or analyze screenshots. They can also combine those approaches.

A recent AISmith comparison of DOM inspection, accessibility snapshots, and visual interpretation illustrates the trade-offs. Accessibility snapshots can provide compact, structured descriptions of interactive controls, while DOM inspection and screenshots can expose information missing from that representation.

One example from the article involves an online bookstore. An agent needs to identify a book that meets particular rating criteria, but the product ratings are displayed visually and don’t appear in the accessibility snapshot. The agent can identify products and purchase controls, yet it lacks the information needed to complete the task.

For the agent, the solution may be to inspect the DOM or analyze a screenshot. For a person who relies on a screen reader, the missing information may pose an accessibility barrier.

An AI agent’s ability to work around inaccessible information doesn’t mean the underlying website has become more accessible. And successful agent automation tells us relatively little about whether people with different disabilities can perceive, understand, navigate, and operate the same interface.

A New Commercial Incentive

Web accessibility has been a persistent challenge, despite decades of standards development, advocacy, and legal requirements.

The 2026 WebAIM Million report, which analyzed the home pages of one million popular websites, found an average of 56.1 automatically detectable accessibility errors per page. These included problems with text contrast, alternative text, form labels, and other accessibility requirements.

Automated testing cannot identify every accessibility barrier, and those results don’t establish whether a complete website is accessible. But they illustrate how widespread detectable problems remain.

Now consider the growing commercial interest in AI agents.

Retailers may want shopping assistants to discover products and complete transactions. Travel companies may want agents to book reservations and manage itineraries. Other businesses may want agents to interact with their services reliably rather than struggle with complex websites designed primarily for human navigation. What this means, basically, is that making websites easier for software to understand is acquiring an additional business rationale.

That changing perspective is visible in a June 2026 Search Engine Journal article about accessibility trees and AI agents. Its significance isn’t so much that it establishes a new accessibility requirement as that it highlights that accessibility-related markup is being discussed with an audience interested in optimizing websites for AI automation.

But the accessibility community has reasons to be cautious. Accessibility specialist Adrian Roselli has criticized advice encouraging developers to add ARIA attributes for AI agents, highlighting the risks of treating ARIA as generic machine-readable metadata rather than semantics designed to support accessibility.

The difference matters because developers optimizing for agent task completion may make choices that aren’t appropriate for assistive technologies.

This tension was also explored at TPAC’s AI and Society mini-workshop. In a presentation titled “Web Accessibility and Agentic AI: Machine-Readable Doesn’t Always Mean Accessible,” accessibility researcher Jonathan Zong addressed the relationship between emerging AI interfaces and the accommodations people with disabilities have long needed.

In an earlier Tech Policy Press article coauthored with Frank Elavsky, Zong examines how developers are creating machine-readable representations for AI systems without necessarily providing equivalent accessibility improvements for people.

A particularly useful example involves providing documentation as plain text for AI systems. That may help a language model process an entire document, but a person using a screen reader may need headings, landmarks, descriptive links, and other structural information to navigate that material efficiently. Machine readability and human accessibility overlap, but they aren’t interchangeable.

Zong and Elavsky also raise a broader concern: accessibility improvements that people with disabilities have requested for decades may suddenly attract investment because they are commercially valuable to AI systems. Yet the people who originally needed those improvements may have little influence over what gets built.

That perspective is worth keeping in mind. Accessibility is already of value to society. It is not merely a potential benefit of making websites easier for machines to use.

Three Possible Directions

I see three broad possibilities for how this new commercial pressure might affect accessibility. They aren’t mutually exclusive, and different websites may follow different paths.

The first possibility is that agent-oriented development produces improvements that benefit people as well.

If developers discover that AI agents perform better with well-structured pages, meaningful labels, and predictable interactions, they may have additional incentive to improve those aspects of their websites.

A product rating previously conveyed only through a visual representation might acquire a meaningful text equivalent. A poorly labeled form might finally receive the necessary attention. Improvements originally motivated by agent compatibility could also make those interfaces more usable by people relying on assistive technologies.

That would be a useful outcome, although better markup alone wouldn’t address every accessibility requirement.

The second possibility is that optimizing for agents introduces new problems.

Developers might begin adding agent-oriented instructions to accessibility-related markup or creating structures intended primarily to influence automated behavior. It wouldn’t surprise me at all if this is already happening.

The accessibility tree has a specific purpose. Information that helps an agent decide what to do isn’t necessarily appropriate information for a screen reader to announce to someone navigating a page. If agent-oriented descriptions begin displacing meaningful accessibility information, or if developers add hidden controls that interfere with accessible navigation, the result could be a worse experience for people.

This concern has already appeared in standards discussions. At a July 2026 APA Working Group meeting, participants discussed whether WebMCP-related information should appear in the accessibility tree and raised concerns about mixing agent-specific information with accessibility semantics.

Those discussions don’t establish that these problems will occur, but they identify plausible risks worth considering during the design process.

The third possibility is that agent-oriented development largely bypasses accessibility improvements altogether.

Consider WebMCP, which proposes ways for websites to expose structured capabilities directly to browser-based agents. Instead of analyzing a page, finding a form, and determining how to complete it, an agent could invoke a capability described by the website.

There are good architectural reasons to explore this approach. Agents shouldn’t necessarily have to simulate human interactions to accomplish tasks, particularly when a structured interface can be more reliable and efficient.

But suppose a business invests in making those agent-facing capabilities work exceptionally well while leaving its human-facing website difficult to use. The agent can complete the transaction. A person who wants to interact directly with the website may still encounter the same barriers.

There is also a question about capabilities that are available to agents but aren’t available, or aren’t discoverable, through the ordinary interface. Some functions may legitimately be intended for automation, but we should understand the implications when an agent can accomplish something that a person cannot independently access.

The result could be an increasingly capable agent-facing web alongside a human-facing web whose accessibility problems receive little additional attention.

All three directions are plausible. We don’t have enough evidence (yet) to know which will become most common, and an individual website could demonstrate elements of all three.

What matters is that the goals driving development are not necessarily identical. Businesses may measure agent compatibility through task completion rates, reliability, or reduced integration costs. Accessibility involves enabling people with different needs to use and understand the service.

Those goals may reinforce each other. They may also diverge. I really want to insert a <shrug> emoji here.

Distinguishing Agents from Assistive Technology Users

Another issue complicates the picture. Websites have legitimate reasons to distinguish automated agents from people interacting directly with a service. Agents can perform operations at scale, operate without continuous human supervision, and pose security or operational risks that differ from those of ordinary browsing. But reliably identifying agent-mediated interactions remains difficult, particularly when agents operate through ordinary browser interfaces.

We also need to be careful about how that distinction is established. A website might attempt to identify automation through interaction patterns, such as the speed or sequence of actions. But people interact with websites in many different ways, including through screen readers, voice control, switch devices, and other assistive technologies.

Behavior that differs from conventional mouse-and-keyboard interaction shouldn’t automatically be interpreted as evidence of an AI agent.

There’s an important technical qualification here: using a screen reader or accessing an accessibility tree does not necessarily expose that activity to the website. Browser automation, platform accessibility APIs, programmatic DOM interactions, and direct network requests have different observable characteristics. The concern is about detection mechanisms that use unreliable behavioral signals, rather than accessibility-tree access being inherently detectable.

There is also a distinction that makes any simple division between agents and assistive technologies inadequate: some people may use AI agents as assistive technology. I find it highly likely that people already do. An agent can be both an automated system and an accommodation that enables someone to interact with a website.

That suggests separating several questions. Is an automated system performing the action? Who authorized it? What is the system permitted to do? Does the website actually need to know anything about the person’s accessibility needs?

The answers do not need to be bundled together, and there are several reasons to think hard about what you’re doing before you take that action.

A website might reasonably require additional authorization before an agent makes a payment or changes account settings. That doesn’t mean it needs to know whether the person directing the agent has a disability.

The proposed Agentic Web Working Group charter acknowledges related concerns, including the possibility that agent identification could expose accessibility needs or enable discrimination. It also excludes bot detection from the group’s scope.

That’s a useful boundary, but it doesn’t eliminate the wider architectural challenge. Mechanisms for identifying agents, authorizing actions, and managing automated access will need to coexist with the many ways people already use the web.

Why Accessibility Expertise Matters to the Agentic Web

These questions explain why APA and ARIA need to be involved in agentic web standards development. The two groups bring different expertise. APA examines accessibility implications across web technologies, while ARIA develops specifications for accessible semantics and their use in web content.

APA has already been investigating AI-related accessibility questions, including through its draft Accessibility of Machine Learning and Generative AI.

The W3C’s October 2026 State of the Web in the AI Era report also identifies the relationship among WebMCP, HTML, and ARIA as an area requiring attention.

One particularly useful detail from APA’s July discussion concerned whether capabilities exposed to agents would also be available and understandable to people using assistive technologies, including those who might not want to enable an AI agent for privacy or security reasons. That is exactly the kind of question that can be overlooked when a specification is evaluated primarily on whether the proposed technology functions as intended.

There’s a standards process consideration here as well.

W3C specifications receive horizontal reviews for accessibility, security, privacy, and internationalization. Those reviews are important, but relying exclusively on them can mean that fundamental architectural choices have already been made before specialists examine their consequences. The emerging agentic web offers an opportunity to involve accessibility experts much earlier. As Zong argues in his TPAC workshop contribution, that should include the meaningful participation of people with disabilities in designing the technologies that affect them. Their involvement should shape requirements and architectural decisions, rather than simply validate implementations after the fact.

For the proposed Agentic Web Working Group, that means coordination with APA and ARIA should involve more than exchanging specifications for review. It means understanding where agent capabilities intersect with accessibility semantics, when agent-facing functionality might create barriers for people, and which questions should be resolved collaboratively before technical approaches become established.

These are not all problems the Agentic Web Working Group can or should solve itself. But it will need to recognize which of its decisions affect work elsewhere, and make sure those dependencies are understood by the groups responsible for them.

What Happens Next?

I approach this subject as someone whose expertise is in digital identity and standards development, with a growing interest in web architecture, rather than accessibility. The TPAC discussions made me aware of questions I hadn’t previously considered in enough depth, and the work of accessibility researchers and practitioners has helped me understand why they matter.

There are already significant accessibility problems across the web. There is also growing interest in making websites more understandable and usable by AI agents. The overlap between those efforts presents opportunities, but it introduces risks and creates the possibility of substantial investment that does little to improve human accessibility. We don’t yet know how those competing pressures will play out.

I’ll be particularly interested in whether agent compatibility encourages better semantic HTML and accessible interactions, whether agent-specific capabilities develop independently of human-facing interfaces, and how efforts to identify automated activity affect people who rely on assistive technologies.

More fundamentally, I’ll be watching how accessibility expertise influences the architectural decisions being made now.

The web has spent decades developing technologies and standards intended to make it usable by people with different abilities and different ways of interacting. As we begin adapting the web for AI agents, we need to understand how those changes will affect the people who already depend on it.

Those consequences are not predetermined, which makes this an especially important time to be asking the questions.

📩 If you’d like to be notified of new posts rather than hoping you catch it on social media, I have an option for you! Subscribe to get a notification when new posts go live. No spam, just announcements of new posts. [Subscribe here]

Transcript

Welcome to the Digital Identity Digest, the audio companion to the blog at Spherical Cow Consulting. I’m Heather Flanagan, and every week I break down interesting topics in the field of digital identity, from credentials and standards to browser weirdness and policy twists.

If you work with digital identity but don’t have time to follow every specification or hype cycle, you’re in the right place.

Let’s get into it.

One of the more thought-provoking conversations I had at the W3C’s annual TPAC meeting in Dublin last month concerned something I hadn’t spent nearly enough time thinking about: the relationship between AI agents and web accessibility.

I’ve written quite a bit about AI agents lately, particularly questions around identity, delegated authority, and what happens when software starts acting on behalf of people.

However, the accessibility discussions introduced another dimension to the problem. AI agents are increasingly using some of the same infrastructure that people with disabilities depend on to navigate the web.

As businesses invest in making websites easier for AI agents to use, those investments may have consequences for accessibility that nobody really intended.

Some of those consequences could be positive. Others could introduce new barriers. Alternatively, we might end up with a web that’s considerably easier for AI agents to use while the accessibility problems people have been dealing with for decades remain largely unchanged.

The reality is that we don’t know yet which direction this will take.

This is also a good illustration of why the proposed W3C Agentic Web Working Group needs to coordinate closely with the groups responsible for accessibility standards.

I want to explore what I’ve been learning and some of the questions I think we need to be asking.


How Browsers Make Websites Accessible

First, let’s look at a little background on how browsers make websites accessible.

When a browser loads a web page, it creates a structure called the Document Object Model, or DOM.

From that structure, the browser also generates what’s known as an Accessibility Tree. This provides information that assistive technologies, including screen readers, can use to understand and navigate a page.

The Accessibility Tree describes things such as:

  • The purpose of a button.
  • The name of a form field.
  • Whether a control is enabled or disabled.
  • How different parts of a page relate to one another.

A properly constructed website provides much of this information through ordinary HTML. Additional accessibility information can be supplied through ARIA, which stands for Accessible Rich Internet Applications.

This infrastructure is very useful for people who interact with websites in ways other than the familiar combination of a screen, keyboard, and mouse.

As it turns out, it’s also very useful for AI agents.

Imagine an agent trying to fill out a registration form. It needs to identify the fields, understand what information belongs in each one, find the buttons to submit the form, and determine whether the operation succeeded.

The Accessibility Tree can provide a structured description of all those controls. That’s potentially much easier than trying to work out what everything means by looking at a screenshot.

Of course, agents have other options. They can also:

  • Inspect the DOM directly.
  • Analyze screenshots.
  • Combine different techniques to understand a page.

I recently read a very interesting comparison of these approaches that included an example involving an online bookstore.

The agent needed to find a book meeting a particular rating criterion: five stars.

It could identify the products and purchase buttons from an accessibility snapshot, but the ratings themselves were represented visually and weren’t included in that information.

The agent could work around the problem by looking at a screenshot or inspecting the underlying page structure.

But consider what that means for someone relying on a screen reader. For that person, the missing rating information might be a genuine accessibility barrier.

The AI agent’s ability to work around the problem doesn’t mean the website has become more accessible.

That’s actually a rather important distinction to keep in mind.

An agent successfully completing a task isn’t evidence that a person with a disability can successfully complete the same task.


The Growing Commercial Interest in AI Friendly Websites

Web accessibility has been a challenge for a very long time.

The 2026 WebAIM Million report, which examined the homepages of one million popular websites, found an average of more than 56 automatically detectable accessibility errors per page.

Of course, automated testing doesn’t catch everything, and those results don’t tell us whether an entire website is accessible. However, they do suggest that a great deal of work remains to be done.

Meanwhile, businesses are becoming increasingly interested in making their websites usable by AI agents.

For example:

  • Retailers want shopping assistants to find products and complete transactions.
  • Travel companies want agents to book reservations.
  • Financial service providers are exploring ways for agents to interact with their systems.

Suddenly, making a website easier for software to understand has another commercial justification.

That shift is already visible in discussions aimed at website developers and search optimization professionals. Accessibility-related markup is increasingly being discussed as a way to make websites work better for AI agents.

Which raises an uncomfortable question.

People with disabilities and accessibility advocates have spent decades arguing for better structured content, meaningful labels, and interfaces that work with assistive technologies.

Will businesses finally invest in those improvements because AI agents need them? And if they do, will the results actually benefit people?

At the AI and Society Mini Workshop during TPAC, accessibility researcher Jonathan Zhong discussed work he had submitted for consideration.

In related work with Frank Eldavski, he has examined how developers are creating machine-readable information for AI systems that resembles accommodations people with disabilities have long requested, without necessarily meeting those people’s needs.

For example, an AI system might be perfectly happy receiving an entire document as plain text.

However, someone using a screen reader may depend on headings, landmarks, and meaningful links to navigate that document effectively.

Both benefit from structured information, but their requirements aren’t interchangeable.

And frankly, accessibility shouldn’t suddenly become important only because it has acquired commercial value for AI.


Three Possible Futures for AI Agents and Web Accessibility

I see three possible directions for how this could develop.

1. Improvements for AI Agents Benefit Everyone

The first possibility is that investment in agent-friendly websites produces improvements that benefit everyone.

If AI agents work better with properly structured pages, meaningful labels, and predictable interactions, businesses may finally have more reasons to fix problems they’ve previously neglected.

Perhaps that visually displayed book rating gets a proper text equivalent. Perhaps poorly labeled forms finally receive attention.

Improvements made for agents could also make websites easier for people using assistive technologies.

That would be a very welcome outcome.

2. Agent Focused Changes Introduce New Accessibility Problems

The second possibility is a little less encouraging.

Developers might start changing accessibility-related information primarily to influence AI agents, without considering how those changes affect people.

Imagine adding instructions intended for an agent into descriptions that a screen reader might announce to a user.

Those descriptions have a particular purpose. Filling them with information intended for a different kind of software could make them confusing or even misleading.

This concern has already surfaced in discussions within the W3C’s accessibility community, particularly around whether information intended for WebMCP and AI agents belongs in the Accessibility Tree at all.

There are very good reasons to keep those purposes distinct.

3. Websites Bypass the Human Interface Altogether

The third possibility is that developers largely bypass the human interface altogether.

WebMCP, for example, is exploring ways for websites to expose structured capabilities directly to agents.

Instead of navigating a page and filling out a form, an agent could invoke an operation that the website explicitly makes available.

There are good reasons to do this. Agents shouldn’t necessarily have to imitate every interaction a person makes with a website.

But what happens if a business invests heavily in those agent-facing capabilities while leaving its human-facing website just as inaccessible as it was before?

An agent can complete the transaction, but the person who wanted to use the website directly still cannot.

We could end up with a considerably more capable web for AI agents without making much progress on accessibility for people.


Why These Three Possibilities Matter Together

The interesting thing is that these three possibilities aren’t mutually exclusive.

Different websites may take different approaches, and the same website could demonstrate all three.

What’s important is that the incentives driving agent development and the requirements for human accessibility aren’t necessarily the same.

There’s another complication, too.

Websites have legitimate reasons to distinguish automated agents from people interacting directly with a service. Agents can operate at scale, perform actions without continuous human supervision, and create security or operational risks.

However, reliably identifying agent-mediated activity remains difficult, especially when agents interact through ordinary browsers.

Attempts to identify automation through patterns of behavior could also create problems for people using assistive technologies.

People navigate websites in many different ways. Some use:

  • Voice control.
  • Screen readers.
  • Switch devices.

These technologies may interact with websites quite differently from someone using a mouse.

That doesn’t mean their behavior should be considered suspicious, bot traffic, or anything of the sort.

To be clear, simply using a screen reader doesn’t necessarily tell a website that a screen reader is being used. Similarly, an agent accessing accessibility information doesn’t necessarily make that activity visible to the site.

The concern is about detection mechanisms that make unreliable assumptions about what normal human behavior looks like.

There’s another important consideration. Some people use AI agents precisely because they need accessibility accommodations.

An AI agent and a person using assistive technologies aren’t always separate categories.

That means we need to distinguish between:

  • Identifying an automated system.
  • Understanding who authorized its actions.
  • Learning something about the person who uses it.

The last of these has its own privacy implications.

A website may have legitimate reasons to know whether an agent is authorized to make a payment. It generally doesn’t need to know whether the person directing that agent has a disability.

These are the kinds of questions that deserve careful treatment as agent identification mechanisms develop.


Why the W3C Agentic Web Working Group Needs Accessibility Expertise

This brings me back to the proposed W3C Agentic Web Working Group.

Its draft charter includes coordination with the Accessible Platform Architectures Working Group, or APA, and the Accessible Rich Internet Applications Working Group, or ARIA.

These groups bring important expertise to the discussion:

  • APA examines accessibility implications across web technologies.
  • ARIA develops specifications that help define accessible semantics.

The accessibility community is already discussing these challenges, including the implications of WebMCP and the possibility of agent-facing capabilities that aren’t available or understandable to people using assistive technologies.

There’s an important lesson here about how standards are developed at the W3C.

Specifications undergo reviews for accessibility, privacy, security, and other considerations. Those reviews are critical.

However, waiting until a specification is largely designed before asking accessibility experts to evaluate it can mean that important architectural decisions have already been made.

Jonathan Zhong’s work also emphasizes the importance of including people with disabilities in decisions about the technologies that affect them.

For the Agentic Web Working Group, meaningful coordination needs to begin while the requirements and designs are still taking shape.

That doesn’t mean the group must solve every accessibility problem on the web. It does mean recognizing when its decisions may affect people who rely on existing web technologies, and working with the communities that understand those consequences.


Accessibility Must Be Part of the Conversation From the Start

Throughout all of this, I’m approaching the subject as someone whose expertise lies in digital identity, standards development, and increasingly, web architecture, rather than accessibility.

The conversation at TPAC helped me recognize questions I hadn’t considered deeply enough. The work of accessibility researchers and practitioners has helped me understand why those questions matter.

We already know that the web has substantial accessibility problems. We also know that businesses are becoming more interested in making their services usable by AI agents.

What we don’t know is how these developments will interact.

Will AI compatibility encourage improvements that benefit everyone?

Will it introduce new barriers?

Or will we build parallel systems that work beautifully for agents while people continue struggling with the same inaccessible websites?

At this point, these outcomes are not predetermined.

That’s why the involvement of accessibility experts, including people who rely on assistive technologies, matters now, while the architecture is still taking shape.

If you think about it, the web has spent decades developing technologies intended to make it usable by people with different abilities and different ways of interacting.

Now, as we adapt that web for AI agents, we need to understand what those changes will mean for the people who already depend on it.


Final Thoughts

There were quite a lot of links in this post. If you’d like links to the research, W3C groups, and specifications I’ve mentioned, please take a look at the written version on my website.

With that, thank you for listening. I’ll be back next week.

And that’s it for this week’s episode. If it helps make things a little clearer, or at least a little more interesting, share it with a friend or colleague and connect with me on LinkedIn at @hlflanagan.

If you enjoyed the show, be sure to subscribe and leave me a rating and review on Apple Podcasts or wherever you listen to podcasts.

You can also find the full written post at sphericalcowconsulting.com.

Stay curious, stay engaged, and let’s get these conversations going.

Heather Flanagan

Principal, Spherical Cow Consulting Founder, The Writer's Comfort Zone Translator of Geek to Human

Leave a Reply

Your email address will not be published. Required fields are marked *.

This site uses Akismet to reduce spam. Learn how your comment data is processed.