From Content to Conversation: How AI Can Improve Reader and Member Support

Imagine spending six months producing an industry report. You've brought together researchers, editors, designers and subject matter experts. You've invested a considerable amount of time and money making sure everything is accurate, professionally presented and accessible.

Finally, the report is published. And then someone emails your organisation asking where they can download it.

Another person wants to know whether it's included in their membership. Someone else has trouble accessing the document, and a fourth person wants to know whether there's a newer version available. Suddenly, your team is spending its time answering questions that have very little to do with the actual content.

If you've ever worked with a publishing company, professional association or educational organisation, this probably sounds familiar. We tend to focus enormous attention on creating and distributing information. But there's another part of the experience that doesn't always receive the same attention: what happens when someone needs help?

And I think artificial intelligence has a rather interesting role to play here.

Great content doesn't automatically mean a great experience

Over the past decade, organisations have become considerably better at producing and distributing digital content. Publishers can deliver publications across multiple formats. Professional associations can provide members with access to research libraries, industry standards and educational resources. Training providers can make entire courses available online.

The technology has come a long way. But the experience surrounding that content can still be surprisingly complicated.

A reader might struggle to access a publication they've purchased. An association member might not understand which resources are included in their subscription. A student might have a question about a course, but the relevant administrator isn't available until the following morning.

None of these problems is particularly dramatic. But collectively, they create friction. And when you're dealing with hundreds or thousands of readers, members or students, even relatively simple questions can consume a considerable amount of your team's time.

The traditional response is to improve documentation, create an FAQ section or introduce a support ticketing system. Those are perfectly sensible solutions. But they don't necessarily solve the underlying problem. People still need to find the right information, understand it and work out what to do next. Sometimes, they simply want to ask a question and receive an answer.

There's a difference between finding information and getting help

One of the most promising applications of AI in publishing and knowledge management is making large collections of information easier to search. Technologies such as Retrieval-Augmented Generation, or RAG, allow users to ask questions about documents and retrieve answers grounded in the information those documents contain.

Zentrovia has already explored this in its article on how RAG is changing document discovery for knowledge organisations. And it's an important development. Instead of searching through hundreds of pages, a reader can ask a question and potentially receive a useful answer supported by references to the original material.

But finding information and getting help aren't always the same thing. Consider someone who contacts a professional association and asks: "I renewed my membership yesterday. Why can't I access the latest industry guidelines?"

The answer probably isn't sitting in a PDF. To resolve that request, someone needs to check whether the membership renewal was processed, confirm the person's access rights and perhaps investigate an issue with the member portal. A traditional document search tool can't do much about that.

An AI system connected to the appropriate business systems, however, could potentially help identify the issue or gather the necessary information before passing the request to the right person. That's where things become particularly interesting. We're moving beyond helping people find information towards helping them complete the tasks associated with it.

AI support isn't just for e-commerce companies

When people talk about AI-powered customer service, the examples usually involve online shops. Where's my order? How do I return this product? Is this item available in another size? These are common use cases, and there's a reason businesses are interested in automating them.

But the underlying technology isn't limited to retail.

Think about a professional association. Members have questions about subscriptions, renewals, events, certifications and access to publications. Or consider an academic publisher. Readers might need assistance with institutional access, licensing, subscriptions, publication availability and account-related issues. An educational organisation faces a similar set of challenges. Students need help accessing courses, understanding enrolment requirements, navigating learning platforms and retrieving certificates.

The specific questions are different, but the underlying problem is remarkably similar. An organisation has information spread across various systems, and someone needs help accessing or using it.

In many cases, the first part of that interaction is fairly predictable. A well-designed AI assistant could answer routine questions using approved information, collect relevant details and direct more complicated enquiries to the appropriate team.

Of course, not every question should be automated. Academic enquiries involving expert interpretation, complex licensing disputes or sensitive student information may require human judgement. But that doesn't mean every routine question needs to land in someone's inbox.

What building Mando taught me about AI support

This is something I've spent a considerable amount of time thinking about while building Mando AI, an AI-powered customer support and business automation platform.

When we started developing Mando, one of our main objectives was to help businesses make better use of their existing knowledge. The idea was straightforward: connect AI to company information so customers could receive useful answers without waiting for someone to respond manually.

But as we started working with real businesses, we discovered that answering questions was only one part of the challenge. Sometimes a customer needs information from another system. Sometimes they want to perform an action. And sometimes they just need to speak with a human. That's why Mando gradually evolved into a broader platform combining AI-powered knowledge retrieval, integrations, conversational AI, a shared inbox and human handover.

One thing became increasingly clear to me during that process. The best AI support experience isn't necessarily the one that automates the most conversations. It's the one that helps people get what they need with the least amount of friction.

An AI assistant that answers immediately but gives incorrect information isn't particularly useful. Neither is a chatbot that keeps asking questions when the customer clearly needs human assistance. And there's nothing more frustrating than explaining your problem to an AI assistant, only to repeat the entire conversation when a support agent takes over. These are the kinds of details that matter much more in practice than they might appear during a product demonstration.

While Mando primarily focuses on business communications and customer engagement, I believe many of the same principles apply to publishers, associations and educational organisations. The goal isn't to replace the people responsible for supporting readers or members. It's to give those people more time to deal with situations where their expertise genuinely matters.

One organisation, multiple communication channels

There's another issue worth considering. People don't necessarily communicate with organisations through a single channel anymore.

Someone might discover a publication through social media, visit the publisher's website and eventually send an email when they encounter a problem. Another person might prefer to contact an organisation through WhatsApp. A student might send a message through a website chat widget because they need help outside normal office hours.

For smaller organisations, keeping track of these conversations can become surprisingly difficult. You might have one employee monitoring email, another responding to website enquiries and someone occasionally checking social media messages. Before long, customer conversations are scattered across multiple systems. And that's where context starts getting lost.

Imagine a reader contacts a publisher through its website about an access issue. The following day, they send an email because the problem hasn't been resolved. Ideally, the support team should know about the earlier conversation. The reader shouldn't have to explain everything again.

Connecting communication channels through a shared system can help address this, particularly when AI assists with categorising enquiries, preparing responses and summarising previous interactions.

However, it's important to distinguish between communication and access to sensitive information. An AI assistant shouldn't automatically receive unrestricted access to membership records, payment details or personal information simply because it can communicate across several channels. The appropriate permissions and safeguards still need to be in place.

Good customer support depends on trust. And that doesn't change just because AI is involved.

Your existing content can become part of the support experience

One of the advantages knowledge-intensive organisations have is that they often possess a considerable amount of valuable information already. Publishers have catalogues, subscription information, licensing policies and help documentation. Associations have membership guides, event information, professional standards and frequently asked questions. Educational organisations have course descriptions, enrolment policies, student handbooks and technical instructions.

A lot of the information required to answer routine questions is probably already available somewhere. The challenge is making it accessible and keeping it up to date.

This is where content management and AI-powered customer support begin to overlap. Rather than maintaining one collection of documents for internal use and another for customer service, organisations can explore ways to make approved information available across multiple applications.

For example, imagine a professional association updates its membership benefits. Ideally, that change would be reflected not only on its website but also in the information used by its AI support assistant. Otherwise, the organisation risks creating an awkward situation where its website says one thing and its chatbot says another. And nobody wants to spend their afternoon explaining why the AI is giving members outdated information.

Of course, making this work requires some discipline. Documents need to be maintained. Outdated information needs to be removed or clearly marked. Access permissions must be respected.

But the principle is straightforward. The information used to support people should be as reliable as the information published for them.

Start with the questions your team already receives

There's a temptation with AI projects to begin by asking what the technology can do. I think a better question is what your organisation actually needs.

Before introducing an AI assistant, I'd suggest looking at the questions your team receives most frequently. Not the most complicated questions. The ordinary ones. The messages that arrive every week and receive almost identical replies.

For a publisher, that might involve subscription access, publication availability or download instructions. For a professional association, it could involve membership benefits, renewal procedures or event registrations. For an educational organisation, it might be questions about enrolment, course access or certificates.

Once you understand those patterns, you can begin identifying which enquiries could reasonably be handled through automation, and which ones should always involve a person.

I wouldn't recommend automating everything from day one. Start with a limited collection of accurate information and a handful of common enquiries. Test the responses carefully. Pay attention to how the system behaves when it doesn't know the answer. Then gradually expand its responsibilities as you gain confidence in its performance.

It sounds less exciting than launching a fully autonomous AI agent capable of handling everything. But it's usually a more sensible way to introduce technology into an existing organisation.

Measure whether people are actually getting help

Another lesson I've learned from building AI products is that it's remarkably easy to get distracted by impressive numbers. An AI assistant handled 10,000 conversations. Excellent. But how many of those conversations were actually resolved?

Did people receive the information they needed? Did they have to contact support again? And when a human needed to step in, did the system make the process easier or more complicated? These questions matter.

For publishers and associations, there are also broader considerations. Are readers finding it easier to access their publications? Are members receiving assistance more quickly? Are employees spending less time answering repetitive questions? And are the conversations revealing useful information about what people find confusing?

That last point is particularly interesting. If dozens of members repeatedly ask how to access a particular resource, perhaps the issue isn't with customer support. Perhaps the website needs improving. If students regularly misunderstand the same enrolment requirement, perhaps the instructions aren't clear enough.

Support conversations can tell organisations a great deal about how people experience their products and services. AI can help identify these recurring patterns. But someone still needs to pay attention to what they're telling you.

The human side of AI support

I don't believe the future of customer support involves removing people from the process entirely. In fact, I think that would be a mistake.

A professional association isn't simply a collection of documents and subscription records. It's a community of people who expect a certain level of service and professional understanding. The same is true of educational organisations, where students may need guidance and reassurance beyond what an automated response can provide. And publishers often maintain long-standing relationships with authors, researchers, institutions and subscribers. Those relationships matter.

AI can help organisations become more responsive, but responsiveness and good service aren't necessarily the same thing. Sometimes, a quick automated answer is exactly what someone needs. Other times, they need a thoughtful conversation with a person who understands their situation. The challenge is recognising the difference, and designing systems that make it easy to move between the two.

This is why I think human handover should be treated as a fundamental part of AI support rather than an afterthought. A good AI assistant should know its limitations. And when it reaches them, getting help from a person shouldn't feel like starting all over again.

Making valuable content easier to use

Publishers, associations and educational organisations have spent years improving how they create, manage and distribute digital content. AI is now creating opportunities to make that content more accessible through natural-language search, intelligent discovery and personalised experiences.

But there's another opportunity sitting alongside those developments: making the experience of interacting with the organisation itself considerably easier. Helping a reader resolve an access issue. Helping a member understand their benefits. Helping a student navigate an online learning platform. Or simply making sure someone receives a useful answer without waiting until the following morning.

These aren't particularly revolutionary applications of artificial intelligence. But they solve real problems. And across organisations handling large volumes of content and thousands of interactions, small improvements can make a meaningful difference.

I suspect the organisations that benefit most from AI won't necessarily be those introducing the largest number of tools. They'll be those that understand where people are struggling and use technology to make those experiences better.

After all, the value of publishing isn't simply in producing information. It's in making that information useful to the people who need it. And sometimes, that begins with something as simple as answering their questions.

About the author

Malaz Madani is the founder of Mando AI, an AI-powered customer support and business automation platform combining knowledge retrieval, integrations, conversational AI, a shared inbox and human handover. This is a guest post; the views expressed are the author's own.