It had to be sometime in the late 1990s that I heard a Pitney Bowes communicator speak at a conference about “Message Mission Control”. It was a great concept then and remains a great concept today. With generative AI tools finding their way into the workplace, it’s time to revisit it.

Too many companies use what I call the “Godspeed Method” to introduce new messaging technologies to employees: “Here you go, everybody. We’re launching Teams/Slack/Yammer/Whatever. Godspeed.” Using the Godspeed Method, IT flips a switch, announces the tool is available, and leaves employees to fend for themselves. It is guaranteed to fail; employees will either ignore the new tool or use it incorrectly.

Someone needs to be in charge of the organization’s messaging culture. That department or individual—Message Mission Control—is responsible for marketing the tool, explaining why it is being deployed, how employees should (and shouldn’t) use it, what it replaces, and why it’s better than existing tools.

In almost every case, no department is better suited to the role of Message Mission Control than Employee Communications.

The sudden rise of generative AI tools available in the workplace means Message Mission Control must expand its scope beyond just messaging. AI adoption is even more problematic than rolling out collaborative tools like Slack and Teams. Even experts are proclaiming tools like ChatGPT a “personal assistant” and urge employees just to use it. Since it’s freely available in the form of Google Bard,  Microsoft Bing Chat, and Anthropic’s Claude (among others), IT doesn’t even have the opportunity to deploy it. Employees are just using it, with or without guidance from the company. Even if their company has banned its use and blocked access, employees can still access it on their phones and tablets. ChatGPT even has iOS and Android apps.

IT departments are exploring more structured uses of AI, but that isn’t stopping employees from prompting these tools using any input that occurs to them to produce any number of outputs.

Employee use of AI without guidance or best practices can lead to problems—even catastrophes. If your organization has not yet communicated guardrails for the use of generative AI, don’t put it off any longer. Among the issues to address is the fact that anything entered in a prompt, along with its result, becomes part of the Large Language Model (LLM) and could appear in the results a competitor or a bad actor gets from THEIR prompts.

However, best-practice guidance can address a bigger issue: the need to revisit workflows when using AI. There are two ways to consider workflows:

1. How to inject AI into your current workflows
2. The new workflows you will need to adapt to account for AI

Injecting AI into Your Current Workflows

Most employees in most organizations have workflows that guide their efforts. Processes begin with step one before moving to step two, then step three, and so on. Whether paying an invoice, developing a proposal, launching a new product, or issuing a press release, the roadmap has usually been well established.

AI’s capabilities require every function to consider its workflows to determine how AI can simplify, streamline, and/or improve it. Consider the workflow for a media pitch. A media relations practitioner might start with a generic pitch, then spend considerable time identifying outlets, editors, and reporters who might be interested in the story. Next, they would rewrite the pitch based on what they know about each journalist in an effort to increase the likelihood that they will respond. Depending on the scope of the assignment, this could be a day or two worth of work.

Using an AI service now available, PRophet, that work can be performed in a matter of an hour or two. PRophet identifies the journalists and reviews their previous work in a matter of minutes, then crafts personalized pitches based on criteria the media professional lists (including, for example, tone: casual, professional, etc.).

Developers are introducing new AI tools every day that can alter workflows. A blogger might want to repurpose a lengthy blog post into a listicle for publishing on additional platforms. There’s an AI for that. It’s called Listicle Club. Just enter the post’s URL and Listicle Club generates a visually appealing listicle.

Of course, communication isn’t the only discipline that can benefit from generative AI. For example, AI-powered generative platforms can create flowcharts based on textual descriptions provided by users. AI-powered tools can understand the given instructions and automatically generate a workflow diagram, saving time and effort for operations staff. These tools can also automate repetitive tasks, like creating customer support tickets.

The fact that AI can generate those support tickets means customer support representatives can shift their focus to high-level support. That’s yet another example of workflow modification that involves redefining roles and responsibilities. AI can even play a role in decision-making, providing summaries and generating insights from data. Companies need to modify workflows that involve decision-making to include input from AI tools.

It should be Message Mission Control’s role to work with every department and team to identify workflows that AI can improve by reducing cycle time or improving outputs (or both).

Adopting Workflows to Account for AI

In both of these use cases, the mere fact that an employee is using AI means they will have to adopt new workflow activities. The media person will need to review each pitch PRophet created before letting the service send it. The blogger will need to edit the listicle to ensure it accurately reflects the contents of the post and the blogger’s writing style.

In no case, ever, should anyone simply cut and paste content into a prompt or use an AI tool’s output without review.

Shortly after ChatGPT was released, I was tasked with converting the raw notes from a SWOT analysis into a document company leaders could use as part of their strategic planning process. Tackling this assignment manually would have taken me a couple of days. Instead, I used ChatGPT and delivered a finished document in about two and a half hours. While that’s a significant time savings—time I was able to spend on more strategic and creative activities—it still seems long. I did not simply paste the notes into ChatGPT with instructions on how to edit them, then deliver the results.

Instead, I scrubbed the notes to ensure nothing I would share with ChatGPT was proprietary. I removed all references to the company. That ensured that material wouldn’t be gobbled up by the LLM and possibly delivered as part of a response to someone else’s prompt.

When I got the results, I performed a side-by-side comparison with the original notes to be certain ChatGPT hadn’t removed anything important. I ensured all the bullets were listed in appropriate categories and the category labels made sense. I also rewrote some of the bullets for consistency and style.

In other instances, employees will need to check outputs for baked-in biases, make sure no copyright has been violated, and that nobody’s privacy has been violated, for example. Each of these activities needs to become routine for every employee using these tools.

It is not enough to communicate these steps to employees, nor is a training program adequate (though companies will need to update training programs to include AI solutions). Message Mission Control drives messaging culture by recognizing employees demonstrating the right behaviors and reinforcing them through ongoing communication. Supporting a cultural shift to accommodate AI also means addressing fear of AI, which some employees are likely to feel, especially given Hollywood’s take on it. (AI is the villain in several newer releases, including the new “Mission Impossible” movie.)

It is unlikely that, in these early days, IT departments are giving much thought to workflows. For communicators, this is a rare opportunity, for once, to be ahead of the curve.