Using AI Wisely in Knowledge Operations

AI has everyone very excited but also a little scared. AI promises endless possibilities but also potential problems.

This document summarizes ScreenSteps’ view of AI in Knowledge Operations and how we are implementing AI into the ScreenSteps product.

Guiding Principles

As we adopt AI, we want to make sure that we keep the following principles in mind:

End-users should be able to trust the information they receive
Permissions around access to information must be respected
AI should make work more effortless

A shorter version of this would be:

Can I trust it?
Is it secure?
Is it easy to use and actually a net benefit to my workflow?

Why do these principles matter?

AI has many strengths but some pretty significant weaknesses. Because AI’s capabilities seem so amazing, we can get ourselves and our employees into trouble if we don't track these guiding principles.

Why did we choose these guiding principles?

AI has a tendency to invent information or “hallucinate” when it doesn’t have all the information it needs. This can lead to information that can’t be trusted.
Many AI systems work on a single body of knowledge. They are not designed to separate out bits of knowledge based on permission or security levels. This can lead to the wrong people having access to information they shouldn’t see.
Ultimately, AI should make our work more efficient and enjoyable, but it shouldn’t be seen as a tool that will just “do everything for you.”

Where Can AI Offer Value While Still Respecting the Guiding Principles?

The areas where we see AI offering the most immediate value are:

Summarizing or extracting information
Helping content authors apply best practices when creating content
Speeding up the content authoring process
Helping authors jump-start the formation of training programs

We will address each of these in the sections below.

Introducing Clarify AI — ScreenSteps AI-Powered Knowledge Ops Assistant

The ScreenSteps AI tools are called Clarify AI since their purpose is to help you transfer knowledge more effectively.

Now Available in Beta:

Summarize Search Results and Extract Information

AI is very good at summarizing and extracting information as long as the AI engine has access to accurate, up-to-date content.

With AI Summary, we can give people access to accurate information with less effort.

For example, we may have an article on our account cancelation policy. As part of that article, there may be information on exceptions to the policy.

If the user asks, “What are the exceptions to our cancellation policy?”, the AI Summary can take the provided articles, analyze them, and provide a summarized result.

The user gets the information they need without having to dig through the full article.

To ensure that information is accurate, the AI engine is instructed to rely only on content from the knowledge base, not outside sources.

We also respect user permissions during this process. Summaries are only provided for content that the current user has permission to see.

Tip: Using AI summaries works very well when you are trying to get an answer to a factual question. If the question is more procedural, like “How do I…?”, then viewing the full article with checklists, screenshots, and decision trees will be more useful.

Now available in beta:

Capture Procedures

Capturing procedures is not only a time-consuming process, but it can be difficult to capture those procedures clearly.

Using AI, we can dramatically speed up the process.

We are currently focusing on speeding up the process of capturing complex procedures. Many of these procedures are non-linear. There are many branches, variables, etc.

Using Clarify’s Audio AI Article Generator, users can record themselves talking through a procedure, explaining all the variables, dependencies, etc. Clarify AI will then generate an actionable outline that includes branching logic.

The resulting outline can then be converted into a decision tree in ScreenSteps using the AI Outline tool.

The results are procedures that take much less time to write and which are much easier to follow.

Coming in late Q3:

Capture Foundational Knowledge

Foundational knowledge is different than procedural knowledge (or what we like to call “actionable knowledge”). Foundational knowledge explains the what, why, and when, not the how.

This information is often written in a narrative format, which is hard to consume quickly.

Using Clarify AI, users can record background information on a topic and have Clarify AI generate a foundational article. These articles break down foundational knowledge into an easy-to-scan format, making it easier for subject matter experts to capture information and for end-users to consume it.

Currently in Internal Beta:

Suggest Foundational Topics

When building effective training programs, it is important to separate foundational and actionable knowledge. Actionable knowledge can be captured in checklists, how-to articles, and decision trees. But if we don’t teach the right foundational knowledge, end-users will struggle to use the actionable knowledge.

Clarify AI can suggest foundational topics and even provide course outlines based on the actionable knowledge you are trying to document.

These tools speed up the development of Find & Follow training programs and help authors identify blind spots they may have when developing training content.

Coming in Q4:

Clarify Existing Procedures

We use the same AI tools to suggest improvements to existing content. This helps authors of all skill levels produce guides that are clear, complete, and easy to follow.

More to Come

We are very excited by all the possibilities of AI. As we identify additional ways to use AI to improve knowledge transfer, we will continue to add them to the ScreenSteps framework.

Request to Join the Clarify AI Beta