In my previous column, “The Dunning-Kruger Effect with AI in communications is not what you think,” I examined the gap between confidence and proficiency. Across Sequencr AI surveys with more than 1,000 substantive responses, 56 per cent of participants described themselves as Explorers, one level above beginners. Another 20 per cent identified as Adopters. Only 12 per cent considered themselves Skilled Users, and 2 per cent qualified as Power Users.
Those proportions have barely changed over the past two years, even as the technology has improved and confidence in using it has risen. The industry is not standing still, but neither is it progressing as quickly as the headlines might suggest.
The question is how communications teams can move beyond this plateau.
That was the subject of a recent Telum Media webinar on agentic AI, where I joined Jeremy Seow, Chair of the PRCA APAC, and Yeelim Lee, Global Head of Communications at the Tanoto Foundation. Our discussion focused on what more mature adoption requires, from choosing the right use cases and measuring new capabilities, to building better governance, context, and working habits.
The global zeitgeist around Generative AI is still hyper obsessed with the productivity benefits that AI delivers. But, as Jeremy Seow argued during the webinar, productivity is one of the least useful ways to judge its value.
The more important questions are whether AI improves the quality of the work, increases its impact, and makes the work itself more satisfying.
Are outputs more consistent and recommendations better informed? Are teams identifying issues earlier, giving leaders stronger advice, and contributing more meaningfully to decisions? Does the technology reduce frustration and give people more time for the parts of the job that they enjoy?
Jeremy described one test in which two teams spent the same amount of time responding to the same brief. One worked manually; the other used AI. When the results were shown to a client without identifying how each had been produced, the AI-assisted response was judged richer in research and thinking. The benefit was not greater speed; it was greater depth and higher quality.
Jeremy grouped measures of value under three headings for evaluation: efficiency, quality, and effectiveness.
Yeelim offered a simpler test from the perspective of an in-house team: does the technology remove friction and provide access to skills or capabilities the team would not otherwise have? A system that adds complexity without improving the work is unlikely to justify its cost or solve the underlying problem / challenge.
One reason AI adoption has stalled in many organisations is that most have started with the tools rather than the problems they are trying to solve. Teams debate whether they need ChatGPT, Copilot, or Claude, then look for ways to use whichever tool they have chosen.
In our work with communications teams at Sequencr, we have found that the best starting point is the work itself: where teams face the most pressure, which tasks consume disproportionate time, and where existing processes fail to produce the results the organisation needs. Identifying those opportunities requires a deliberate process for ranking use cases and examining recurring friction points, including work that takes too long, depends on too many manual steps, or is harder than it should be.
Some of the most valuable opportunities begin with work the team cannot do today or wishes they could do with AI.
Automating an existing task may save time. Solving a persistent friction point can improve how a team operates, create new capabilities, and change how the organisation delivers value.
The next obstacle is not access. It is behaviour.
Most communications professionals can now access AI through a dedicated company license, a personal account, or AI features embedded in existing software. Yet access has not translated into broad proficiency or application.
As the panelists noted, training can help close the gap, but adoption and application depend on involving people in the process, giving them responsibility for how AI is introduced, and making its use part of the way performance and roles are defined. That can include assigning ownership of use cases, adding AI competencies to job descriptions, and recognising progress through performance goals.
Habits matter just as much. New tools rarely become part of daily work because employees attended a training session. They become useful when people reach for them repeatedly while solving problems. Yeelim made the point that habits form more naturally when adoption begins with the people doing the work and the friction they encounter each day. A mandate imposed from above can have the opposite effect, especially when the technology adds another step rather than removing one.
That makes AI adoption a management challenge as much as a technical one. Teams need clear expectations, visible support from leaders, and room to learn from unsuccessful experiments.
In most organisations, governance and AI policy have acted less as enablers than as brakes on adoption.
Clear rules and guardrails are important. People want to know how to handle confidential information, which tools are approved, when human review is required, and who remains accountable for the final work. But many policies have been written primarily to limit risk. When those policies are vague, overly restrictive, or when they are designed to scare people, they can create a culture of hesitation. Employees avoid using AI altogether, or experiment quietly without sharing what they learn.
As Jeremy argued during the webinar, experimentation and control are not opposites. Effective guardrails create a safe environment in which people can test new uses, learn from failure, and understand when an issue must be escalated.
In our work helping communications teams develop AI policies at Sequencr, we have seen stronger adoption when those policies do more than describe what is prohibited. The more useful policies also explain where experimentation is encouraged, how new use cases should be tested, and who is responsible for reviewing and improving them.
Policy also has a wider purpose. It is a statement of strategic intent and should express how an organisation intends to use AI, how work may change, and which capabilities it wants to develop as a result of the innovation.
For communications leaders, that means translating enterprise policy into the context of the function or agency. A communications AI policy should explain where AI should set a direction for adoption and innovation, not merely define the limits of acceptable use.
Communications teams have long wanted to be more data-driven. The obstacle was not a lack of data, but the form in which it existed. Unlike marketing or sales data, much of the evidence available to communications teams is unstructured. It lives in articles, reports, transcripts, presentations, and stakeholder records rather than in neat rows and columns. AI helps us tap into that intelligence.
This was one of the inspirations behind Stratum, the AI agent platform we have been building at Sequencr for comms teams. It brings together external signals, including media coverage, emerging issues and stakeholder activity, and allows teams to examine how coverage has changed over time, which messages are gaining traction, what preceded an issue, and how different stakeholders have responded.
The same data can also provide a clearer view of what may come next. As Jeremy noted, a year of monitoring reports can support hypotheses about the narratives likely to develop, the signals that tend to precede an issue, and the conditions under which stakeholder concern escalates.
The communications team of 2028 is coming a lot faster than we may expect. During the webinar, I described a future in which individuals begin the day by assigning tasks to a group of agents. One might monitor an issue, another conduct research, and another prepare a report. The work would return later for review, interpretation, and action. The communicator's role would shift from completing every step manually to setting the objective, providing context, and judging the result.
Early versions of this model already exist. In Stratum, an agent can assess the cultural context around a campaign, define an influencer search, identify potential partners, and recommend them against the needs of the brand. Other agents can handle recurring requests for specialised reports that consume large amounts of time inside agencies and in-house teams.
By 2028, managing agents may be as familiar as managing colleagues, suppliers, or agency partners. The strongest teams will know what to delegate, how to direct it, and how to act on that information.
It seems strange to say, given how new the technology still is, but in many ways, we have become so accustomed to AI that it is easy to forget how empowering the technology can be. It gives communications teams the agency to do work that was once too costly, too complex, or simply out of reach.
Realising that potential requires a break with the current inertia. Teams can move beyond isolated prompting and productivity gains and begin redesigning the work itself. Otherwise, a technology that could expand the influence and impact of communications will never deliver on its potential.
Matt Collette is CEO of Sequencr AI, a technology consultancy dedicated to unlocking the full potential of AI for marketing and communications teams. Before founding Sequencr, he was Head of Digital for Edelman Canada and later Global Head of Digital Growth, where he led efforts to embed generative AI across Edelman's global operations. Matt created the firm's AI task force, launched its first campaign powered by generative AI, and developed tools, prototypes, and training initiatives for clients and internal teams.
Read more from our columnists in The Earned View