We're Changing How We Use AI. Are We Changing How We Manage It?AI is becoming more involved in how we work. A recent announcement from OpenAI raises questions about what happens when the technology we've come to depend on changes. If you've spent hours building a custom GPT, you might want to pay attention to December 11, 2026. That's the date OpenAI plans to retire custom GPTs and move users toward plugins. For anyone who has invested time creating a specialized AI assistant, this is a significant development. Think about what goes into building one of these tools. You write instructions, upload documents, establish standards and test the results. You correct what the system gets wrong and refine its responses until it becomes useful for a particular kind of work. For communications professionals, that might mean an assistant built around an organization's brand voice, executive messaging, editorial standards or communications procedures. Over time, it becomes something you rely on because you've invested considerable effort in making it work the way you need it to. Then the company providing the technology announces that the product is going away. In a September article for Ragan, Stephanie Nivinskus explained what the retirement means for communications teams and outlined steps for preserving their work. OpenAI is providing a migration path to plugins, although some capabilities will require additional work and the replacement may behave differently. The immediate task is fairly clear. Organizations need to identify the custom GPTs they use, preserve their instructions and reference materials, migrate what they can and test the replacements. I've been thinking about what this announcement tells us about the way AI is changing. We're beginning to build important parts of our professional work around technology that can change with relatively little notice. At the same time, the technology itself is taking on more responsibility. Those 2 developments deserve attention, particularly from people responsible for communications, marketing and organizational reputation.
AI Is Becoming Part of How Work Gets DoneThink back to how most people started using generative AI. We opened ChatGPT, typed a question and waited for an answer. We asked it to summarize a document, suggest headlines, improve an email or generate ideas for a presentation. The interaction was straightforward. We gave the system an assignment and received something to review. That remains a common way to use AI, and there's nothing inherently wrong with it. But the range of work we can give these systems has expanded considerably. We can create specialized assistants with instructions and reference materials, connect AI to other applications and ask it to conduct research across multiple sources. We can also assign work that involves several steps and requires the system to move between different kinds of information. Products such as ChatGPT Work are designed to carry out longer assignments that previously required us to move between applications and manage each stage ourselves. For a communications professional, the difference is substantial. Consider preparing for an executive interview. An early use of ChatGPT might have involved asking it to suggest 10 questions a journalist could raise. A more developed workflow might involve gathering the executive's previous interviews, reviewing recent coverage of the organization, identifying recurring issues, comparing messaging with public statements and preparing a briefing document. The professional still needs to decide which issues matter, what advice to give and whether the briefing accurately reflects the situation. But more of the preparation can now happen within an AI-assisted process. As these capabilities develop, we're likely to spend more time assigning, reviewing and directing work that AI systems perform. That changes the skills we need and raises questions about how we maintain oversight when technology handles an increasing share of the process. We're Putting More of Our Professional Knowledge Into AIOne reason custom GPTs became useful was that people could give them context. A general-purpose model might produce a perfectly acceptable press release, but acceptable is rarely enough when you're writing on behalf of a company, an executive or an institution. Communications work depends on details. An organization may have specific language it uses to describe its business. A CEO may have strong preferences about how messages are structured. Certain subjects require legal review, while others carry historical sensitivities that wouldn't be obvious to someone unfamiliar with the organization. Experienced communicators learn these things over time. Some of that knowledge exists in formal documents, but much of it comes from experience, conversations, previous decisions and an understanding of the people involved. Custom AI assistants have given professionals a way to document some of that knowledge and make it available for recurring assignments. I think that's one of the more consequential developments in how we're using AI. We're translating parts of our professional expertise into instructions that technology can follow. There's a practical benefit to doing this. A team can establish standards once and apply them across multiple assignments. New employees can access approved reference material, and routine work can become more consistent. But there's also a management responsibility that deserves attention. Who owns those instructions? Who checks whether they're still accurate? What happens when an executive changes direction, an organization updates its strategy or a communications policy becomes outdated? And what happens when the platform containing that knowledge changes? The retirement of custom GPTs makes those questions immediate. OpenAI says its migration process can transfer instructions and knowledge files into plugins. However, custom actions don't transfer automatically, selected models don't carry over and migrated tools may produce different results. That means organizations will need to test whether their replacements still perform the work correctly. A familiar AI assistant can become part of a team's routine long before anyone formally decides who is responsible for maintaining it. We need to be more deliberate about that responsibility. AI Is Moving Closer to Our Information and SystemsAnother change is happening as AI becomes connected to the applications people use every day. Rather than uploading a document every time we need assistance, connected systems can access information from approved sources, subject to the permissions and controls in place. A marketing professional might use AI to examine campaign reports across several files. A communications director might ask it to review internal documents and prepare a summary for a leadership meeting. An executive assistant might use connected tools to help coordinate information across a calendar and other applications. These capabilities create new questions about access, confidentiality and accuracy. Communications teams routinely work with sensitive information. We see financial results before they're announced, help prepare responses to employee matters and legal disputes and advise executives on subjects that may never become public. The ability to connect AI to organizational information requires careful decisions about what each system should be allowed to access, who can authorize that access and how information should be handled. It also requires us to think about the quality of the information being used. An AI assistant may be able to retrieve a document without knowing that the document contains outdated guidance. It may find an approved message from 2 years ago without understanding why the organization no longer uses that language. It may combine information from several sources that were never intended to be used together. These are familiar information-management problems, but AI can make them harder to notice because it produces such polished responses. A well-written answer can create the impression that the underlying information has been checked. For communications professionals, that distinction matters. Our responsibility includes knowing where information came from, whether it's current and whether it's appropriate for the audience and situation. Connecting AI to more information makes those responsibilities more important, especially when the resulting work may influence decisions or become part of an organization's public communications. We're Beginning to Delegate Work That Continues Without UsThere's another development I find particularly interesting. AI systems are increasingly able to work on assignments that extend beyond a single conversation. Depending on the product and permissions, users can schedule tasks, request recurring work or assign projects that involve several stages of research and execution. The experience begins to resemble delegating work to an assistant. You describe the assignment, establish expectations and review the result. Of course, an AI assistant doesn't have the same understanding, accountability or professional obligations as a human colleague. That's precisely why the management side of AI deserves attention. When we delegate work to another person, we usually make decisions about their experience, authority and level of supervision. We know that a new employee preparing a sensitive executive statement needs more guidance than a senior communications adviser who has worked with the CEO for years. We also know that different assignments carry different consequences. Drafting a routine meeting summary is one thing. Preparing a public response during a crisis is another. As AI systems become capable of carrying out longer assignments, we need similarly deliberate standards for their use. Which tasks can proceed without continuous review? Which require approval before an action is taken? When should the system stop and ask for direction? Who is responsible for checking the work? These decisions should reflect the risk of the assignment and the consequences of an error. For communications leaders, this is particularly important because our work often involves judgment that depends on circumstances the technology may not fully understand. A system might produce a technically accurate statement that would be poorly received by employees. It might recommend a response that follows an established procedure without recognizing that the circumstances call for an exception. It might identify a pattern in media coverage without understanding the relationships or history behind it. Those limitations tell us where professional judgment needs to remain active, even as AI becomes capable of handling more of the work. What Happens to Our Own Expertise?There's another question I think we should be asking as AI takes on more responsibility. What happens to our professional capabilities when we stop practicing certain skills? Communications professionals develop judgment through experience. We learn to identify weak arguments by working through them. We become better writers through drafting, editing and receiving feedback. We learn to anticipate difficult questions by preparing for conversations and observing what happens in the room. Some of that learning involves work that can feel repetitive or time-consuming. AI can reduce the time required for those activities, which is often useful when teams face growing demands with limited resources. But we should be thoughtful about which activities we delegate and what we continue to practice ourselves. Consider writing. If AI produces every first draft, how does a less experienced communicator develop the ability to structure an argument independently? If a system routinely analyzes media coverage, how do team members learn to recognize patterns and question the conclusions? If AI prepares every executive briefing, when do emerging professionals develop the research and synthesis skills that make someone a trusted adviser? These are questions about professional development. Experienced professionals may be able to review AI-generated work using judgment developed over many years. Someone entering the profession may not yet have that experience. We can't assume that the ability to recognize a good result develops automatically from reviewing AI output. Organizations will need to think about how people learn when technology performs more of the tasks that traditionally helped them develop expertise. That includes creating opportunities to practice independent thinking, explain decisions and receive meaningful feedback. It also means being careful about how we measure productivity. Completing an assignment faster tells us something about the process. It doesn't necessarily tell us whether the person doing the work is developing the capabilities they'll need for more complex responsibilities. If we want future communications leaders who can exercise sound judgment, we need to consider how they're going to acquire it. AI Management Is Becoming a Leadership ResponsibilityThe custom GPT announcement is a useful reminder that AI products will continue to change. Features will be introduced, revised and retired. Companies will make decisions about their products that affect the people and organizations using them. Some changes will make our work easier. Others will require us to rebuild processes we've spent time developing. Organizations should expect this and plan accordingly. For communications and marketing leaders, I think several practices deserve attention. Document the knowledge behind your AI systems. Keep copies of approved instructions, reference materials, editorial standards and important examples in locations your organization controls. Make sure the documents are current and that someone is responsible for maintaining them. The knowledge your team has developed should remain accessible even when a particular AI product changes. Know which AI tools your team depends on. A department may have several specialized assistants created by different employees. Some may support routine assignments, while others may contain detailed instructions for work that only a few people understand. Leaders should know what exists, who created it and what would happen if it became unavailable. This is particularly important when individual employees develop AI workflows that eventually become part of the team's regular operations. Establish clear expectations for review and approval. The amount of human review should depend on the assignment. A social media brainstorm and a response to a regulatory investigation require different levels of scrutiny. Teams need to understand those distinctions before they begin delegating more complex work. They should also know when an AI-generated recommendation needs to be questioned, revised or rejected. Continue developing professional skills. AI training should include the ability to evaluate information, question recommendations, recognize weak reasoning and explain decisions. Learning how to operate a particular product is useful, but those broader capabilities will remain relevant as products change. Leaders should make sure their teams have opportunities to develop judgment through experience rather than relying exclusively on AI-generated work. Plan for technology changes. When an AI system becomes important to a recurring process, consider how the work would continue if the product were unavailable or its behavior changed. That might mean maintaining accessible source documents, testing alternative processes or making sure more than one person understands how the workflow operates. The more important the assignment, the more deliberate that planning should be. None of these practices requires a large AI department. They require leaders to treat AI-supported work with the same care they give other important business processes. The Next Phase Will Require Different HabitsI expect the way we use AI will look quite different a year from now. Some of the tools we're using today will change. New capabilities will become available, and tasks that currently require several prompts may become easier to assign and complete. I also expect people will become more comfortable allowing AI to perform work that once required their direct involvement. That could free communications professionals to spend more time on activities that benefit from their experience, including advising leaders, understanding audiences, managing relationships and making difficult decisions. Whether that happens will depend partly on how organizations use the time and capacity these systems create. There's no guarantee that adopting more capable technology will automatically produce better communications or stronger professional judgment. Those outcomes still require thoughtful decisions about how work is organized, how people learn and where responsibility sits. Which brings me back to the retirement of custom GPTs. For teams that have invested in these assistants, December 11 is a practical deadline. There's work to do before then. But the announcement also gives us a reason to examine how much of our professional knowledge and daily work now depends on systems we don't control. If an AI assistant your team relies on disappeared tomorrow, what would you lose? Would you lose a convenient way to complete certain assignments? Or would you discover that important instructions, knowledge and processes had become difficult to recover? The answer would tell you quite a bit about how your organization is managing AI. And it might be a useful place to begin your next conversation about it. |
Comms FutureLab is a media platform for communications and marketing leaders who are endlessly curious about the future of our profession. Through thoughtful analysis of AI, media, leadership, trust and organizational change, we help today's leaders prepare for tomorrow's challenges and opportunities.
When Every Meeting Is Recorded, What Happens to Candor? AI meeting assistants are becoming part of everyday work. New research raises questions about what happens when those tools start evaluating the people in the room. I recently attended a meeting with almost as many AI agents as human participants. š Between meeting assistants, recording devices and smart glasses, Iām starting to wonder how much of our working lives are being documented, analyzed and stored somewhere. A few weeks ago, I...
Your Embargo Needs an Internal Comms Plan Iām working on a project with an organization that has a longstanding practice: employees hear company news 24 hours before itās announced publicly. I understand the reasoning. Employees shouldnāt routinely learn important news about their company from a news alert, LinkedIn post or reporter. But 24 hours can be a long time when youāre also working with media under embargo. An embargo gives reporters access to information before an announcement with...
AI Has a Communications Problem Sam Altman has been making the rounds, and in a recent interview with Bloomberg, he acknowledged that the AI industry has done a poor job communicating the benefits of the technology. Itās not often you hear ācommunications challengeā in a Bloomberg cover story, so of course I was instantly invested. Altman talked about the fear surrounding AI and the need to give people a greater sense that it will increase their agency rather than take it away. I agree with...