All Posts

Build Better Bots For Agents and Customers With Knowledge Architecture

 How does knowledge architecture help with automating customer interactions?

Automating routine interactions can provide consistency, speed, and lower costs for customer support. Increasingly it is also allowing organizations to support more sophisticated interactions. But these depend on having not only the right processes in place to accomplish the required task, but also having the right knowledge readily at hand for automated systems to recognize and retrieve.

WATCH: Knowledge Architecture - The Path To Automating Customer Support Interactions

Challenges of developing bots

Automating contact center interactions involves the development of digital assistants and chatbots to support conversational interactions by either text or voice. The goal is to improve customer service while reducing costs or these interactions. However, this is challenging because the current approaches of many bot vendors are not efficient. Training bots is costly and their performance is limited. For example, they often cannot handle a full interaction with the customer; they usually need human intervention.

Remember: Chatbots are a channel – to data, content and knowledge - not a replacement for human reasoning.

One of the reasons for the high costs of automation is people think that setting up a bot requires that an organization re-create all its content and reinvent their processes. Sometimes they'll add organizational structures such as “AI content group” just for automated content, which can really be a showstopper. “Many organizations end up creating new AI functions or groups to specifically develop content for cognitive applications, but really this should be part of existing content operations, not something new. The approaches to content have to be different but the core function is really the same. Content operations need to be evolved to for new AI-powered systems, but this needs to be an extension of enterprise-wide content operations” according to Seth.

Structured knowledge components can be used to improve performance and designing these architectures in a focused way can power intelligent assistants and bots for very precise information retrieval. A decade ago, EIS developed one of the first digital assistants on the market. It was tremendously successful, leveraged component content and now the industry is coming around to this approach for powering so called “cognitive” assistants and applications. Highly functional digital assistants need to be trained with high value and well curated knowledge artifacts and have a flexible, extensible knowledge architecture. That's frequently the piece that's missing, and it is needed in order to develop bots efficiently.

Bots: Friend or foe?

When properly implemented, automated systems such as chatbots can save time and money, and improve customer service. The numbers are impressive.

Smart bot are effective

But when customers get frustrated, the impact on a brand can be strongly negative, both for customers and call center reps who are trying to help customers.

Dumb bots frustrate

Demographic preferences further fuel the trend toward automated customer service. One recent study indicated that 72% of Millennials believe that a phone call is not a good way to obtain customer support. They want online experiences and accomplish their tasks through chat.

Finding the answers

In the past, the user manual was the default source of answers. But the advice to “read the manual,” has long been discarded. Manuals are not well written – or written with the needs of specific users in mind. People don’t have time. A semantic or keyword search can help zero in on a more specific location in a document, but again, that will not necessarily get the person to the answer.

The problem is that many times, knowing the context of the situation is required for any system to find needed information. The relevant information might be in a CRM system or an ERP system, depending on the user’s role or the customer’s needs. Information from a manual cannot just be presented but has to be integrated with information based on the scenario at hand, such as whether the customer is a new one or an existing one, is technical or non-technical, or what specific equipment and configuration is present in their environment. To do this, knowledge bots must be built in a way that embeds human, judgment, knowledge and intelligence into the system to provide situational context.

Without a knowledge structure, bots will be brittle. They will not be able to interpret the user, and they won't perform very well. The content needs to be componentized to allow the bots to find the answers that customers need.

Reusable components are essential

It’s important to consider the expected uses of content when making decisions about how to break it down. It should be broken into content in the form of reusable pieces that reflect typical cases for the organization, Seth advised. For example, if translation and localization are part of what an organization needs, it should not have to be done again completely if only a small piece of the content changes. The content should be designed so only the changed information is replaced. That is where componentizing is usually leveraged.

Another example is an insurance company that has a 300 page policy document. If a customer wants to know what is covered and what is not, those facts should be broken out so they can be accessed individually. Then if question answering systems are built, the same structure and the same reusable components can be ingested into these intelligent applications. This reduces the time to train the AI or cognitive application. It helps leverage the existing organizational structures, boundaries, and processes, and it also lowers internal training costs.

Breaking down the content

Once content is broken down to the right level of detail, it can be used for multiple purposes.

Transforming Content For Multiple Cognitive Channels

Transforming content for multiple cognitive channels

Knowledge architecture is the way to apply the correct structure to those components. Then that content can serve multiple purposes, including customer self-service, chatbots and conversational access to information which will lead to improved access to knowledge for customer facing access and internal knowledge management. But it has to be done right. An inflexible or poorly aligned self-service knowledge architecture (based on user needs) will not reduce calls to customer support. Content analytics can help find the missing pieces so that users will have a high quality experience regardless of their role or context.

Converting documentation into conversational flow

Content generated in PDF or word processing documents can be put into short segments so that a customer can follow step-by-step instructions, pausing or rewinding to review the steps, as Henry describes here.

Technical manual publishing map

The same content can be used for self-service or for assisting a customer service representative in talking though a process with a customer either on the phone or by sharing a screen.

Conversational bot immediately available

Accelerators for bot development

Products are now available for accelerating bot development so that the process does not have to be done manually. The Rulai enterprise conversational computing platform is one such product.

In the clip below is a demo that shows how easily the platform can generate a bot conversation. In this case, it can parse PDF content or imports componentized content (using DITA or another content architecture) generate reference flows and path flows.

HubSpot Video


”What's different now is the ability to very precisely answer questions, based on the knowledge architecture, but also the mechanisms for processing content through imports and generating flows automatically rather than building them one-by-one manually. They still need some tweaking but this is a really great step forward for bots” according to Seth.

Where to start

Information architecture overall can be quite complex, and that can be daunting, but start by taking a subset of enterprise content, and build out a knowledge architecture that is very contextualized and specific to the problems that you're trying to solve. Look for routine, high frequency topics and build those use cases before starting to look at the complex ones. Make the information readily available, which will help whether the bot is simple transaction support or a bot that is helping a human do a job. Building a helper bot is a good approach so that you can test it before releasing it into the wild. The customer service representative can use it as an aid as it is tested and fine-tuned without leaving the customer stranded if the bot does not end up being responsive to the question. This becomes an opportunity to build greater functionality through ongoing learning.

The content of this blog is a summary of the discussion between Henry Truong, VP, Enterprise Solutions at, and Seth Earley, CEO of Earley Information Science during a webinar, Knowledge Architecture - The Path to Automating Customer Support Interactions.

View Webcast



Seth Earley
Seth Earley
Seth Earley is the Founder & CEO of Earley Information Science and the author of the award winning book The AI-Powered Enterprise: Harness the Power of Ontologies to Make Your Business Smarter, Faster, and More Profitable. An expert with 20+ years experience in Knowledge Strategy, Data and Information Architecture, Search-based Applications and Information Findability solutions. He has worked with a diverse roster of Fortune 1000 companies helping them to achieve higher levels of operating performance.

Recent Posts

[Earley AI Podcast] Episode 26: Daniel Faggella

Human Cognitive Science Guest: Daniel Faggella

[RECORDED] Master Data Management & Personalization: Building the Data Infrastructure to Support Orchestration

The Increasing Criticality of MDM for Personalization for Customers and Employees Master data management seems to be one of those perennial, evergreen programs that organizations continue to struggle with. Every couple of years people say, “we're going to get a handle on our master data” and then spend hundreds of thousands to millions and tens of millions of dollars working toward a solution. The challenge is that many of these solutions are not really getting to the root cause of the problem.  They start with technology and begin by looking at specific data elements rather than looking at the business concepts that are important to the organization. MDM programs are also difficult to anchor on a specific business value proposition such as improving the top line. Many initiatives are so deep in the weeds and so far upstream that executives lose interest and they lose faith in the business value that the project promises. Meanwhile frustrated data analysts, data architects and technology organizations feel cut off at the knees because they can't get the funding, support and attention that they need to be successful. We've seen this time after time and until senior executives recognize the value and envision where the organization can go with control over its data across domains, this will continue to happen over and over again. Executives all nod their heads and say “Yes! Data is important, really important!” But when they see the price tag they say, “Whoa hold on there, it's not that important”. Well, actually, it is that important. We can't forget that under all of the systems, processes and shiny new technologies such as artificial intelligence and machine learning lies data. And that data is more important than the algorithm. If you have bad data your AI is not going to be able to fix it. Yes there are data remediation applications and there are mechanisms to harmonize or normalize certain data elements. But looking at this holistically requires human judgment: understanding business processes, understanding data flows, understanding dependencies and understanding of the entire customer experience ecosystem and the role of upstream tools, technologies and processes that enable that customer experience. Until we take that holistic approach and connect it to business value these things are not going to get the time, attention and resources that they need. In our next webinar on March 15th, we're going to take another look at helping organizations connect master data to the Holy Grail of personalized experience. This is an opportunity to bring your executives to a webinar that will show them how these dots are connected and how to achieve significant and measurable business value. We will show the connection between the data, the process that the data supports, business outcomes and the and the organizational strategy. We will show how each of the domains that need to be managed and organized to enable large scale orchestration of the customer and the employee experience. Please join us on March 15th and share with your colleagues - especially with your leadership. This is critically important to the future of the organization and getting on the right track has to begin today.

[Earley AI Podcast] Episode 25: Michelle Zhou

Data Tells the Story Guest: Michelle Zhou