We are going to walk through an example of intelligent automation, the process, and the decisions that need to be made along the way.  This isn’t about adopting a specific tool or providing a how-to tutorial on how to address a specific problem, but rather to give you an understanding of how to go about an intelligent automation effort.

Before we jump in, if you haven’t already read the whitepaper on Intelligent Automation that sets the foundation for what we’re trying to do, you can get a copy on the NovoCircle website under the resources section at the top.  There is another paper in there titled “The End of the Information Economy as we know it” that explains some of the technology trends taking place and how they will impact both companies and individuals in “information worker” job roles.  I’d highly encourage you to take a look at those for some context.

Common Mistakes of Intelligent Automation

There are 3 common mistakes people make when trying to automate a business process or a set of job tasks.  I’ve done all of these myself, and as a team, we talk about these regularly to keep each other in check.

1. Jumping straight to picking tools
2. Trying to find one tool that does the whole job
3. Spending time trying to make new cool tools work instead of focusing on the outcome

I’m not going to say that diving down the rabbit hole of AI tools isn’t fun, or that I haven’t learned a lot and gained some new ideas about what is possible, but when we talk about Intelligent Automation, it is important to remain focused on the goal.  We do that by applying what we refer to as the “NovoCircle Method,” which is really nothing more than four questions that we use to ground our thinking, so we don’t get distracted by all the bright and shiny things in our environment. If we find ourselves getting caught up in a trap or veering off course, we go back to those questions and, if needed, answer them again.

The “NovoCircle Method”

1. What is the problem we are trying to solve?
2. What are the pieces of the puzzle?
3. Where are we trying to get to (or what does success look like)?
4. How are we going to get there?

Human In the Loop

In the Intelligent Automation whitepaper, we talked about the one “non-negotiable rule” for intelligent automation: Always keep a human in the loop and in control of the process you are automating.  Intelligent Automation isn’t about replacing humans; it’s about figuring out where humans add true value in a process and differentiating that from busywork that technology can do for us.  There are certain things that the human brain can do that AI (as it exists today) simply cannot.

• Creativity – AI is a math machine; the things it creates are by definition derivative works based on things that have been done before.
• Clarification of intent and ambiguity
• Guardrails on automated tasks
• Critical thinking and review of outputs to assess how they will be perceived by humans

As we are working through the process of Intelligent Automation, this is something that should be considered early on – not at the end

Scenario: Creating Videos to accompany written content

I create a lot of written content that gets published on our website and on LinkedIn.  But when I go to LinkedIn, I don’t really want to read through a bunch of text.  I’d rather watch an interesting video that explains the topic or perspective, and then if I want more information, I can go read the source document.

Maybe I’m lazy or just fatigued, but if I were to venture a guess, I’d say others feel the same way.  So here’s the scenario for our Intelligent Automation project.  I want to be able to create videos based on my written documents or articles.  They need to be professional.  They need to reflect my ideas. (not AI-generated slop) They need to use my voice and reflect our company’s brand. And, most importantly, they need to be engaging and interesting.

Let’s use the NovoCircle method that we talked about to break this down

What is the problem we are trying to solve?

I want to produce good-quality videos to accompany my written articles and papers.  I need them created quickly (because I don’t have much time to spend on this, and I also want to be able to publish new video content a few times/week).

What are the pieces of the puzzle?

We are going to answer this question by looking at the process (the sequence of tasks) involved with creating a video.

The Inputs

• Source Material – the written article or paper on which the video is based. We put these through a rigorous review process to make sure the content is clear. Fact checked and aligned to our company messaging (not AI-generated fluff) before we publish it.
• Reference Content – in this case, I’m going to use our company’s website to provide context about what we do, our philosophy, and other articles and papers that we’ve published.
• Brand Assets – We have a logo, font, and color scheme that I want applied to the video
• Re-usable Elements – Eventually, we’ll get to a reusable intro/outro segment for videos, but for now, we just use a title slide and end slide to serve as bookends, like we’d do if we were giving a live presentation

The Process

1. Create the storyline – if I were to do this manually, I’d use sticky notes on a wall to create a storyboard of what I want the video to cover.  In my opinion, this is the most important part of the process
2. Write a script – sometimes I just start talking and dictate a story as a starting point, but even if I do that, refining it so it flows is essential. Think about being on a video call when the slides don’t work.  If the story is clear and stands on its own, the message still gets delivered
3. Illustrate the story – here is where the sticky notes or storyboard are helpful. We break the script into scenes, and each scene gets an illustration.  It may be a slide, a screenshare, a stock video or animation, or just me talking.   The key here is that the illustrations support the story, not the other way around.  In most cases, I’d use PowerPoint to create a slide deck to accompany my story.
4. Decide what format to use. Do I want this to be audio over some slides?  Do I want an animation?  Do I want a video of talking heads- either mine or multiple people?  Do I want a split-screen or picture-in-picture format?  This is important because it will guide us on what needs to be recorded.
5. Record the audio/video. In most cases, I just do audio because I want people to focus on what I’m saying or showing, not on my face.  There are some exceptions.  I usually try 2-3 takes of a recording because I rarely get a clean version the first time.
6. Rough cut editing. Here is where the multiple takes are cut and spliced to create a clean audio track, and the visual components are layered for each scene.  Sometimes I try to do this myself, but I usually ask a video editor on my marketing team to do it for me.  Any reusable elements are added at this step.
7. Rough cut review. This is where I take a critical look at the rough cut and ask whether the story is clear and engaging, whether it gets the message across, and whether the visuals support or detract from the overall message.  If there is something I don’t like, this is the time to re-record, swap out a visual, or make a change.
8. Final cut editing. Here is where the finishes are applied to the video.  Transitions are smoothed, captions are added (if desired), Chapters are assigned, and any metadata that is going to be needed for publishing.  A key thing to keep in mind here is what the thumbnail will look like (what shows up when someone shares a link to the video, the frame the video will be on when the viewer hits play, and the end frame when the video ends).  I like to have any website links or calls to action on that last frame.
9. Final rendering. The final video format will be guided by where the video will be published and viewed.  If this is going to be projected on a big screen, we may want a high-resolution rendering.  If it is going on YouTube or LinkedIn, we may want a different format.  If it will be shared via email, we may need to consider how to reduce the file size.
10. Here is where we post the video wherever it needs to go.  We may need to add in some descriptive text or even a full transcript in some cases.

The human in the Loop

Here is where I like to stop and ask the question about the human in the loop.  If I were to use AI or other tools to help automate this process, where would I want to keep humans in the driver’s seat?  Stated differently, where do humans create unique value in this process?  Here’s what I came up with (and it’s okay if you disagree)

• Creative Intent – If this video is going to reflect my ideas, I need to be able to guide the story’s direction and express my intent for the messages I want to convey and how they are conveyed.
• Editing the Script –AI generates usable content 85% of the time, but I need to be able to fix the gaps
• Editing Visual Elements – There is some nuance around tone, formality, information, and engagement that is best to have a human touch. AI-generated visuals sometimes don’t come across well to viewers.
• Recording the Audio/Video – I’ve tried some tools that sample my voice and use it for AI voiceover. I haven’t found one that sounds natural, and even if I did, the audio component is my chance to connect with the audience on a personal level.
• Rough Cut Review – This is perhaps the most important part of the entire process.  Even if I hired other people to write the script, create visuals, and a voice actor to do the recording, it is important that I provide critical feedback as the production comes together.
• Final Cut Editing – This part of video editing is an art. I’m not that kind of artist, so I don’t do this myself, but this is a step in the process that really needs the experience and eye of someone who knows what they are doing.

As we look at these activities, there is still a lot of work that doesn’t necessarily need to be done by a human.  This is where intelligent automation comes in – leveraging technology to augment the humans and enable them to be more effective and efficient.

If we were to look at this process through the lens of how many hours of effort are needed to produce say a 5 minute video, it would probably be somewhere in the range of 30-50 hrs in total over a couple week duration.

Where are we trying to get to (or what does success look like)?

In an ideal situation, this process would be heavily augmented by technology, but retaining humans in the loop for the critical activities we identified.  The human effort would be reduced to 5-10 hrs, and the total time from a source article to final rendering would be 1-2 days.  There would be 2 roles involved: the presenter and a video editor.  Interactions between them would be clearly defined and limited to avoid repetitive feedback loops.  The process would be asynchronous, so there could effectively be multiple projects going on at the same time.

How are we going to get there?

There are 2 ways to approach intelligent automation of a process like this: task augmentation or process re-engineering.

Task Augmentation

Using task augmentation, we’d look at the individual steps in the process and ask the question: “How could I use AI and other tools to make this step faster, cheaper, more effective, or higher quality?  We might use a combination of different tools to support different activities.  We wouldn’t have to address all tasks at the same time.  We’d probably look for low-hanging fruit – either things that are easy to automate or things that have a big impact on effort or cycle time.  The risk is low, it’s relatively easy to get started, and you start to see value right away.

There are a few downsides to this approach.  You are constrained by existing structural assumptions about the process, such as the sequence of events, or you may be trying to automate something that was only necessary as a mitigation to some other issue.  You will likely also need some form of orchestration mechanism (human or automated) to move things along in the workflow.  And the potential benefits will not be as high as through process re-engineering.

Process Re-Engineering

Using process re-engineering, we’d look at the process as a whole and ask, “Is there a better way of achieving the end result by approaching the process in a different way?”  We might be looking for a single tool or a couple of tools to support the end-to-end capabilities.  The benefit is that you realize the full value of intelligent automation at once, rather than through incremental improvements.  The potential to achieve higher levels of efficiency/effectiveness are also there.

The downsides of the re-engineering approach are that it takes more time and effort up front to figure out what the new process/tool combination would look like.  Unless there is a fit-for-purpose tool that meets our exact needs, we are likely to spend considerable time evaluating alternatives and experimenting to see if we can find a tool that does what we want.

So, what did we decide to do?

This isn’t about what we decided to do… its about you.  At NovoCircle, our mission is “to help our friends and colleagues be successful”.   What are the things that you do in your job (or in your life) where you’re spending time that you’d rather be spending doing something more meaningful?  How can Intelligent Automation help you achieve more with less effort?  How can we help?

Visit novocircle.com to access resources and articles to help you ground your understanding of what’s possible, then when you’re ready, fill out the contact form and request a consultation with one of our Intelligent Automation experts.