S01 EP7 - The Feelings Lounge: using AI for wellbeing and self-coaching, with Kate Hughes (Cambridge University Press & Assessment)
How can micro-experiments help shift mindset and behaviour when used with AI coaching?
Micro-experiments sidestep the common trap of attempting wholesale personality overhauls or operational restructuring. Instead, they involve low-stakes, incremental shifts—such as protecting break time during meetings or approaching a major proposal with conviction rather than hesitation—that compound over weeks , whereas attempting too much at once typically fails.
The power of micro-experiments lies in their psychological accessibility. Kate Hughes discovered that when change feels manageable and non-threatening, it becomes sustainable. Rather than declaring "I will become more confident" or "I will eliminate all stress," a micro-experiment might be as simple as: "This Friday, I will take a ten-minute walk before my leadership meeting." That single, concrete action becomes the testing ground for a new behaviour.
When paired with AI coaching, micro-experiments gain an additional layer of impact. In the Feelings Lounge framework discussed on this episode , every week begins with structured reflection: what energised you, what drained you, what patterns are emerging, and what single micro-experiment will you try this week? AI acts as the thinking partner, identifying patterns across multiple check-ins that humans often miss on their own.
The cumulative effect is profound. Over a series of weeks, these tiny shifts reshape how you approach challenges. One micro-experiment might be: "When I receive critical feedback, I will pause for five seconds before responding instead of defending immediately." Another week: "I will draft that proposal outline today, not wait until it feels perfect." Each small victory builds confidence and creates evidence that change is possible—without the paralysis that comes from aiming for perfection.
"Instead of asking the question, what can AI do for me? It's more a question of what can AI help me notice about myself."
Kate Hughes — Head of Internal Communications and Engagement Business Partnering at Cambridge University Press and Assessment. Hughes leads a team of senior business partners helping leaders communicate clearly and guide organisations through change and complexity. After experiencing burnout, she shifted from conventional wellbeing approaches to using AI as a structured thinking partner, pioneering the Feelings Lounge framework for weekly self-reflection and behaviour experimentation.
This distinction is central to why micro-experiments work. As explored in depth during this podcast conversation , the AI is not imposing change on you—it is helping you recognise patterns you have already been living but not consciously observing. You notice that you hesitate before every high-stakes proposal, or that Mondays drain your energy disproportionately, or that you avoid difficult conversations until they become crises. Once you notice, you can experiment.
What makes micro-experiments particularly effective is their psychological safety. You are not betting your career or identity on a single grand gesture. You are running a small, reversible test. If it fails, the cost is minimal. If it succeeds, you have proof that change is within reach. This creates a feedback loop: each successful micro-experiment validates your agency, which builds the psychological foundation for the next one.
Over time, these accumulated experiments shift your core beliefs about yourself. You move from "I cannot change" to "I can change one thing at a time." That shift in mindset is not a sudden epiphany—it emerges quietly from the evidence you accumulate yourself, week by week, through actions you have chosen and executed.
Why attempts at wholesale change typically backfire
The reason micro-experiments outperform grand resolutions is neurological and psychological. When you attempt to overhaul your entire way of working or thinking, your brain perceives this as a threat. The energy required is overwhelming, the failure rate is statistically high, and after a few setbacks, you abandon the effort entirely.
Micro-experiments bypass this resistance entirely. A change small enough to complete in a single week feels achievable, which means your brain does not trigger the defensive "this is too much" shutdown. You execute, you reflect, you adjust. The full methodology for designing these experiments each week is detailed further in the episode , but the core insight is: incremental action compounds more reliably than heroic ambition.
Kate Hughes observed that when she attempted to "fix her burnout" as a whole, nothing shifted. When she instead chose one micro-experiment—such as protecting time for breaks during meetings—she not only completed it, but noticed how it affected her energy and presence. That single success created momentum for the next experiment.
Micro-experiments replace wholesale overhauls with low-stakes, incremental shifts that avoid triggering psychological resistance to change.
Small, concrete, weekly experiments compound over time, gradually reshaping behaviour and mindset through accumulated evidence of success.
AI coaching identifies behavioural patterns you would otherwise miss, making your experiments more targeted and your reflections deeper.
Psychological safety is essential: when each experiment feels reversible and low-risk, you build agency and confidence that fuels further change.
What is the 'Feelings Lounge' framework and how does it structure weekly self-reflection with AI?
The Feelings Lounge is a weekly Friday morning reflection triggered by AI that guides you through four distinct reflections: what energizes you, what drains you, what loops in the background without your realizing it, and one micro-experiment to try next week. Rather than jumping into problem-solving mode, the framework holds up a mirror to actual patterns—shifting you away from what an anxious brain assumes is happening toward what is genuinely occurring.
From speed to slowness: the shift in how AI supports wellbeing
Kate Hughes initially approached AI as a productivity tool for efficiency, but she discovered something more valuable: using AI as a thinking partner for self-awareness rather than task automation . Conventional wellbeing approaches weren't working for her during a period of overwhelm and burnout, so she began exploring how AI could help her notice patterns about herself rather than just get things done faster.
The Feelings Lounge emerged from this realization. As Kate explains in the episode , the framework deliberately slows you down—a counterintuitive move in a field often obsessed with AI acceleration. The weekly cadence (Friday mornings) creates a consistent anchor point for reflection, while the four-part structure ensures you're capturing the full picture of what's actually influencing you.
The four reflections that reveal hidden patterns
Each Friday, the framework asks four questions. First: what energizes you this week? This positive framing prevents the bias toward only noticing problems. Second: what drains you? This captures the obvious stressors but also the subtle ones you might otherwise ignore.
Third, and perhaps most revealing: what is looping in the background —the thought, worry, or pattern that keeps returning without you actively thinking about it? This is where imposter syndrome, self-doubt, or professional avoidance often hide. Fourth: identify one micro-experiment to try next week based on what you've noticed. This prevents the framework from becoming mere venting and instead makes it actionable.
Over multiple check-ins, AI pattern recognition can detect recurring themes you might miss in isolation. A point discussed at length in this podcast is how Kate discovered her own imposter syndrome pattern only after several weeks of check-ins revealed the same loop surfacing again and again.
"Instead of asking the question, what can AI do for me? It's more a question of what can AI help me notice about myself."
Kate Hughes — Head of Internal Communications and Engagement Business Partnering, Cambridge University Press and Assessment. Kate leads a team of senior business partners who help leaders communicate clearly and guide the organisation through change and complexity. She experienced burnout and overwhelm that prompted her to explore AI not as a productivity accelerator, but as a tool for genuine self-awareness and wellbeing.
One of the most striking details from Kate's experience: she explains how the framework helped her deliver a high-stakes proposal with conviction after weeks of fretting , demonstrating how pattern awareness translates into real professional confidence.
The Feelings Lounge is a structured weekly reflection (Friday mornings) guided by AI, not a productivity tool.
Four questions each week: what energizes, what drains, what loops in the background, and one micro-experiment—this combination reveals hidden patterns.
AI's role is to hold up a mirror to actual patterns over time, not to immediately jump into fixing or advice-giving mode.
Repeated check-ins reveal recurring loops (like imposter syndrome) that single conversations would miss entirely.
How can AI be used as a thinking partner for wellbeing and self-coaching beyond traditional productivity gains?
AI shifts from a speed tool to a genuine thinking partner through structured reflection. Kate Hughes built weekly "feelings lounge" check-ins with Copilot that uncover energizing factors, draining patterns, and overthinking loops—revealing that her struggles came from self-doubt rather than workload, not from productivity gaps at all.
The turning point for Kate Hughes came when she stopped asking "What can AI do for me?" and started asking "What can AI help me notice about myself?" This subtle shift transformed her relationship with technology from an efficiency tool into a genuine reflection partner.
From Speed to Self-Awareness: The Feelings Lounge Method
Kate established a recurring ritual: weekly Friday morning check-ins with Copilot structured around four simple reflection areas. Each session explores what energizes her, what drains her, patterns looping in the background, and one micro-experiment to try that week. Over time, as detailed in this episode of Unprompted , Copilot began detecting recurring themes across these conversations—imposter syndrome, overthinking before high-stakes deliverables, avoidance patterns—that Kate herself had not consciously named.
The breakthrough came when Kate realized her struggles had nothing to do with the volume of work . One proposal she'd fretted over for weeks was delivered with zero pushback once she approached it with genuine conviction. Copilot had helped her notice that her real obstacle was not capacity but self-doubt masquerading as pragmatism. This pattern emerged only because she gave the AI consistent, honest reflections across multiple sessions.
"Instead of asking the question, what can AI do for me? It's more a question of what can AI help me notice about myself."
Kate Hughes — Head of Internal Communications and Engagement Business Partnering, Cambridge University Press and Assessment. Kate leads a team of senior business partners helping leaders communicate clearly and guide organisations through change and complexity. She experienced burnout and overwhelm when conventional wellbeing approaches failed, which prompted her to explore AI as a self-coaching tool.
The beauty of this approach is that pattern recognition happens across time . A single conversation with an AI is generic advice. Thirty conversations, structured and honest, reveal your actual operating system. Copilot evolved naturally from a chat interface into both a wellbeing partner and career coach because Kate gave it permission to remember, notice, and gently reflect back what it was seeing.
As Kate explains in the full Unprompted conversation , this method works precisely because it avoids the performance trap of traditional coaching. You're not trying to impress anyone. You're not optimizing for the "right answer." You're just being honest about what's true this Friday morning, and letting the AI notice what that pattern might mean over weeks and months.
The micro-experiments—small, personal tests each week—turn insight into action without adding pressure. If Copilot notices you're energized by strategic work but drained by reactive meetings, one week's experiment might be to batch reactive tasks and protect two hours for strategy. The experiment fails or succeeds; either way, next Friday you report back, and the pattern refines.
Reframe AI from a productivity tool to a reflection partner by asking what it helps you notice about yourself, not what it does for you.
Structure recurring check-ins (weekly works) around consistent reflection prompts: what energizes, what drains, background patterns, and one micro-experiment.
Pattern recognition emerges over time—a single conversation is generic; thirty honest conversations reveal your actual operating system and blind spots.
Use micro-experiments to turn insights into action without performance pressure, treating each week as data for the next reflection.
S01 EP1 - Experimenting with AI agents in internal comms, with Rebecca Lindsay (hyperexponential)
What tool has made building AI agents significantly more accessible for internal communicators?
Glean has removed the need for internal communicators to rely on software engineers to build AI agents, connecting across company tools and platforms to draw on a wealth of organizational knowledge. By allowing agents to be built both manually through steps and via natural language commands, Glean has leveled the playing field—enabling communications professionals to create sophisticated tools without any coding background.
From Gatekeeping to Self-Service AI
Traditionally, building AI agents required technical expertise that internal communications teams simply didn't have. The barrier to entry was high: you needed developers, software engineers, or specialized AI knowledge to even attempt agent creation. This meant many valuable internal communications use cases never made it off the drawing board.
Glean changed that equation by removing technical prerequisites while maintaining powerful functionality. As Rebecca Lindsay explains in the episode , the ability to build agents through natural language is "massive" and "one of the most significant advantages for using AI in internal comms." The tool doesn't discriminate between a technical team member and a communications professional—both can create equally functional agents.
This democratization is critical because it aligns with the broader vision of AI in internal communications: giving employees tools to communicate better themselves , rather than centralizing all communication through traditional gatekeepers. When internal comms teams can rapidly prototype and deploy agents, they can respond to organizational needs at actual speed.
Two Pathways, Same Outcome
Glean's power lies in its flexibility. You can build agents either by specifying steps manually or by describing what you want in plain English. Neither approach requires coding knowledge. This dual pathway means teams can choose the method that fits their workflow—whether that's a step-by-step guided creation process or a more conversational approach using natural language.
This flexibility matters because internal communicators work in diverse contexts. Some teams may have very clear, structured processes they want to automate; others may need to iterate and experiment quickly. Glean accommodates both, removing the common objection that "we'd have to bring in a developer for this." A point detailed in this podcast is how quickly and independently teams can now move—something that was impossible before such tools existed.
"We want to be giving you the tools to do great comms yourself, and that is precisely what AI can enable us to do at scale."
Rebecca Lindsay — Internal Comms Manager at Hyper Exponential, a tech company where AI adoption is embedded across all roles. As part of her company's AI enablement squad, Rebecca has built multiple agents including tone-of-voice checkers, annual leave assistants, and emotionally intelligent communications tools—all without software engineering involvement—demonstrating how non-technical communicators can deliver sophisticated AI solutions at organizational scale.
To understand more about how Rebecca has implemented these agents in practice and the specific challenges she's solved, listen to the full episode , where she walks through her hands-on experience building agents from scratch.
Glean eliminates the technical barrier—internal communicators can build agents without software engineering skills.
Agents can be created through manual steps or natural language, offering flexibility for different workflow preferences.
Removing dependency on developers means internal comms teams can prototype, test, and deploy AI solutions independently and at speed.
This accessibility directly enables the shift from internal comms as a service provider to internal comms as an enabler of self-service communication tools.
How is AI fundamentally changing the role of internal communicators beyond just creating content?
AI is shifting internal communicators from being 'post boxes' that distribute messages to enablers who give others the tools to communicate well themselves . Rather than creating all content centrally, communicators can now guide managers and teams through AI-powered agents that help them craft their own messages, turning the role into one of strategic support at scale.
The transformation addresses a real challenge facing internal comms teams today. As explained in Unprompted: Real AI in IC , people across organizations are time-poor, yet they still need to communicate effectively with their teams. Instead of bottlenecking all requests through a central comms function, AI agents can distribute the capability to communicate well directly to the people who need it most—managers, team leads, and individual contributors.
This shift means internal communicators stop being the gatekeepers of every message. Instead, they become architects of systems that empower others. As Rebecca Lindsay describes it: "We want to be giving you the tools to do great comms yourself, and that is precisely what AI can enable us to do at scale."
From bottleneck to enabler: redefining the internal comms role
Traditional internal communications operates as a centralized function—messages flow through the comms team, creating inevitable delays and a dependency on their capacity. This model breaks down when organizations need rapid, localized communication from managers and team leads. AI agents solve this by placing communication tools directly in the hands of those who need them , guided by frameworks and guardrails set by the internal comms team.
The role becomes less about authoring every message and more about ensuring that those messages—written by dozens or hundreds of people across the organization—align with brand voice, tone, and strategic intent. Rebecca Lindsay's work at Hyper Exponential demonstrates this in practice through AI agents that support everything from annual leave messaging to emotionally intelligent communications checking, freeing internal comms to focus on strategy rather than production.
This also enables better conversations between managers and their teams. When managers have tools to craft thoughtful, considered messages—rather than sending off-the-cuff replies—the quality of internal dialogue improves. The internal comms team becomes the strategic guide rather than the sole creator, multiplying their impact across the entire organization without requiring proportional growth in headcount.
"We want to be giving you the tools to do great comms yourself, and that is precisely what AI can enable us to do at scale."
Rebecca Lindsay — Internal Comms Manager at Hyper Exponential, where she leads AI enablement and experimentation with AI agents designed to democratize communication across the organization. Her work includes building tone of voice agents, annual leave assistants, and emotionally intelligent communications checkers that empower employees at all levels to communicate more effectively.
The annual leave assistant asks users when they're going on leave and when returning, offering to draft both internal and external out-of-office messages.…
What is the role of layered data in building more effective internal communications strategies with AI?
Individual data points are useful, but combining multiple sources—internal surveys, employee satisfaction data, event feedback, and external research indexes—creates exponentially richer insights for AI-powered communications decisions. This layered approach gives your AI agents and human team the full context they need to craft more effective, targeted messaging.
The power of layered data lies in depth. Rather than relying on a single metric or feedback channel, Rebecca Lindsay compiles a personal internal comms brain that pulls from multiple simultaneous data streams . This includes hard metrics like internal comms survey results and employee satisfaction scores, alongside qualitative inputs like event feedback and benchmarked research from reputable sources.
As detailed in this episode of Unprompted: Real AI in IC , external research frameworks like the Institute of Internal Comms index become particularly valuable when layered with your own organizational data. These indexes reveal macro trends—such as the finding that people are increasingly time poor —which you can then cross-reference with your own employee feedback to understand how that pressure manifests in your specific culture.
Why a single data point falls short
One survey result, one satisfaction score, or one event feedback comment tells you what happened, but not why it matters or how to act on it. Layering changes that calculus entirely. When survey data aligns with event feedback and external benchmarks all point to the same tension, your AI agents have unambiguous direction on where to focus comms efforts.
Rebecca emphasizes that richer context enables richer output from both AI tools and human judgment. Instead of defaulting to generic messaging, you can craft internal communications that directly address the specific pressures and priorities your people face. As discussed in the full podcast episode , this principle applies whether you're designing an AI agent or writing a strategic brief for your leadership team.
"We want to be giving you the tools to do great comms yourself, and that is precisely what AI can enable us to do at scale."
Rebecca Lindsay — Internal Comms Manager, Hyper Exponential. At Hyper Exponential, a tech-focused organization with an AI-maximalist leadership culture, Lindsay has pioneered the use of AI agents and tools to solve real internal communications challenges. Her work spans tone of voice agents, emotionally intelligent communications checkers, and employee enablement systems—all grounded in rigorous data layering and behavioral design principles.
The practical implication is clear: before you build an AI agent or refresh your comms calendar, audit your data sources. Look for gaps where you're flying blind, then systematically layer what you learn across all future strategy. This layered approach is what transforms AI from a content-generation shortcut into a strategic decision-making engine for internal communications .
Multiple data sources combined—surveys, satisfaction scores, event feedback, and external benchmarks—deliver far deeper insights than any single metric alone.
External research indexes like those from the Institute of Internal Comms reveal macro trends that, when layered with your own data, unlock actionable context for comms planning.
Richer data context enables both AI agents and human communicators to craft more targeted, effective messaging aligned to real organizational pressures.
A personal internal comms brain that systematically layers data sources becomes a strategic asset for all future communications decisions.
How does an emotionally intelligent communications checker agent work and what are its limitations?
Rebecca Lindsay built an agent that scores written communications against emotional intelligence principles using a defined rubric, but it only analyzes text and has no access to human context—missing critical nuance that matters to real conversations.
A rubric-driven approach to emotional tone
The agent works by evaluating written communications—emails, messages, announcements—against key emotional intelligence principles . Rebecca built it to give instant feedback on tone and approach, mimicking what an emotionally aware reviewer would flag.
The system uses a clear scoring rubric to evaluate each message. This keeps the feedback consistent and reproducible, though it also locks the agent into a fixed framework.
The blind spot: context and nuance
The agent's biggest limitation is that it cannot see the human context . An emotionally intelligent response depends heavily on who wrote the message, why, for whom, and what happened before—none of which the agent can access.
As a result, it flags patterns based purely on words and tone, missing the real story. A sharp response might be exactly right if the recipient has been asking the same question five times; it might be tone-deaf if delivered to someone in crisis. The agent can't tell the difference.
"We want to be giving you the tools to do great comms yourself, and that is precisely what AI can enable us to do at scale."
Rebecca Lindsay — Internal Comms Manager, Hyper Exponential. Rebecca leads AI experimentation across internal communications at a company where the CEO explicitly champions AI adoption across all roles. She has built multiple agents—tone of voice checkers, annual leave assistants, and emotional intelligence reviewers—as part of the company's AI enablement squad, turning organizational friction points into scaled solutions.
The unexpected value: reactions reveal what matters
Despite its limitations, the agent has already sparked something valuable: it triggers real conversations when people receive unexpected scores . When someone gets a low emotional intelligence rating they weren't expecting, they push back—and those conversations surface disagreements about what "good" communication actually looks like.
That friction is information. It reveals where people's intuitions differ from the rubric, and where the definition of emotionally intelligent communication is actually contested within the organization. Rebecca describes this in the episode as one of the most valuable outcomes—not the scores themselves, but the conversations those scores unlock.
What prompt engineering techniques help ensure better AI agent outcomes?
Start by asking the AI tool to clarify your requirements before running the full prompt —Rebecca Lindsay emphasizes you don't know what you don't know. Instruct the AI to ask follow-up questions and remember them for future interactions, and be explicit about which knowledge sources the agent should draw from and what outcome you're trying to achieve.
Three foundational principles for stronger prompts
The first step is pre-clarification. Rather than launching straight into a complex request, ask the AI to help you define what you actually need. This upfront conversation prevents misalignment later and forces both you and the agent to name assumptions that might otherwise remain hidden. As Rebecca explains in the episode , many people rush into their request without first unpacking what they actually want to achieve.
Second, embed permission for the agent to ask clarifying questions within your initial prompt. This transforms the interaction from one-directional to conversational. By instructing the AI to probe deeper, you allow it to surface gaps in your brief and propose refinements. Rebecca describes this as giving the agent a voice to shape the outcome, not just execute a predetermined instruction.
Third, be explicit about scope and knowledge boundaries. Define which sources, tools, or data the agent should reference and which it should avoid. Clarity on what the agent is supposed to know—and what it should ignore—prevents hallucinations and ensures outputs stay grounded in your organizational context. This is particularly important at Hyper Exponential, where the tone of voice agent and annual leave assistant had to be trained to draw only from specific internal documentation and policies.
"We want to be giving you the tools to do great comms yourself, and that is precisely what AI can enable us to do at scale."
Rebecca Lindsay — Internal Comms Manager, Hyper Exponential. Rebecca has been part of the company's AI enablement squad and is known for building multiple AI agents to solve internal communications challenges, including tone of voice checkers, annual leave assistants, and emotionally intelligent communications tools. Her work is grounded in a company culture where the CEO describes himself as an AI maximalist and encourages all teams to experiment with AI tools in their roles.
To dive deeper into how these prompt principles translate into real agent deployment and the behavioral design choices that follow, listen to Rebecca's full conversation on the episode , where she also discusses how to identify which organizational problems are right for agent solutions in the first place.
Pre-clarify your requirements with the AI before running the full prompt; acknowledge gaps in your own thinking upfront.
Build clarifying questions into your initial prompt instructions so the agent refines its understanding of your goal.
Explicitly define which knowledge sources and tools the agent should draw from, and which it should exclude.
This framework prevents misalignment and ensures AI agents produce outputs that fit your organizational context and objectives.
What are the key steps for implementing and rolling out an AI agent in an organization?
AI agent rollout requires three sequential steps: identify the need through data or observed friction , create the agent itself, then design the behavioral rollout so it reaches users at their moment of greatest need. Success depends on placing the agent at the exact point where employees encounter the problem it solves.
Starting with Evidence, Not Assumptions
The first step is not to guess which problems your organization should solve. Data from surveys or direct feedback reveals the actual pain points that employees face daily. This grounds your AI investment in real friction, not intuition.
As Rebecca Lindsay explains in the episode , Hyper Exponential identified the annual leave assistant as a priority after survey data showed this was genuinely pulling people's attention away from other work.
Timing Is Everything: The Moment-of-Need Principle
Once the agent exists, the rollout strategy determines whether employees actually use it . Launching an agent and hoping people find it is passive; designing the moment of adoption is active adoption strategy.
Rebecca's approach integrated the annual leave assistant directly into the HR platform so employees received the agent link at the exact moment they submitted their annual leave request. They did not have to remember the tool existed or search for it—it appeared where the problem manifested. This behavioral design pattern dramatically increases adoption compared to sending a blanket announcement and waiting for uptake.
This principle applies across all AI agents: the episode explores how strategic placement turns a tool nobody remembers into a solution people use instinctively.
"We want to be giving you the tools to do great comms yourself, and that is precisely what AI can enable us to do at scale."
Rebecca Lindsay — Internal Comms Manager, Hyper Exponential. As part of the company's AI enablement squad, Lindsay has built multiple agents to solve internal communications challenges, including tone-of-voice checkers and emotionally intelligent communication tools, all designed to scale human capability rather than replace it.
The full conversation also covers how Hyper Exponential's CEO takes an AI-maximalist stance , creating organizational permission for experimentation and iteration—a context that allowed these rollout strategies to develop in the first place.
Identify problems through data and surveys, not guesswork, to ensure the agent solves a real need.
Create the agent with clear functionality aligned to that identified friction point.
Design the behavioral rollout to reach employees at their moment of need, not through announcement alone.
Embed the agent within existing workflows (e.g., HR platform integration) to maximize discovery and adoption without extra effort from users.
How can internal communicators identify which problems are suitable for AI agent solutions?
Combine hard evidence from employee surveys with intuitive listening to spot where AI agents can genuinely help. Rebecca Lindsay at Hyper Exponential uses internal communications survey data to identify friction points—like her annual leave assistant, created after survey results revealed employees struggled when returning from time off—and pairs this with her instinct for catching workplace grumbles where better conversations could transform the employee experience.
Data-Driven Discovery Meets Employee Listening
The foundation of identifying suitable AI agent problems starts with real survey results and feedback loops . When Rebecca Lindsay saw that return-to-work transitions were causing stress among her team at Hyper Exponential, she didn't guess—she let the data guide the solution. An annual leave assistant was built directly in response to this evidence, solving a concrete pain point rather than chasing AI adoption for its own sake.
However, data alone isn't enough. As an internal communicator, Rebecca relies on her lived experience of organizational life—the conversations in corridors, the repeated complaints, the moments when people express frustration about processes that feel unnecessarily complicated. This intuitive listening amplifies what surveys reveal , surfacing problems that employees might not articulate in formal feedback channels. The combination of quantitative evidence and qualitative observation creates a map of where AI agents can make a real difference.
Discussed in depth in Unprompted: Real AI in IC , Rebecca's approach reflects a broader principle: prioritize employee outcomes over technology trends . The Institute of Internal Comms index 2026 highlights that people are time poor at the moment, making this need for friction-reduction even more acute. An AI agent should ease the burden, not create another step in an already taxed workflow.
"We want to be giving you the tools to do great comms yourself, and that is precisely what AI can enable us to do at scale."
Rebecca Lindsay — Internal Comms Manager, Hyper Exponential. Rebecca works at Hyper Exponential, a tech-focused company where the CEO describes himself as an AI maximalist and encourages all employees to use AI in their roles. She has been part of the company's AI enablement squad and is known for experimenting with AI agents and tools to solve internal communications challenges, including tone of voice agents, annual leave assistants, and emotionally intelligent communications checkers that help employees communicate more effectively.
The full episode reveals exactly how Rebecca built multiple agents from this same principle—each one addressing a real need discovered through the combination of data and organizational intuition. Listen to the episode to hear the specific workflows and tools she deployed, from ChatGPT to Copilot Studio to Glean, and how she measured their impact on employee experience.
Survey data reveals concrete pain points; use annual engagement results to identify where employees struggle most.
Internal communicators' intuitive listening skills are as valuable as quantitative feedback—pay attention to recurring grumbles and friction.
Pair evidence with empathy: an AI agent solution must genuinely reduce friction, not add complexity to already time-poor teams.
Build agents to scale human capability, not replace it—the goal is to enable better conversations, not bypass them.
S01 EP6 - Reimagining internal comms with AI, with Adele McIntosh
Why does productivity with AI initially decrease before increasing, and how should communicators plan for this learning curve?
Productivity with AI initially decreases because teams must invest significant time in learning, experimentation, and training rather than immediately reaping rewards. As Adele McIntosh explains, there is no shortcut to unlocking AI's deeper capabilities —organizations that stick to surface-level uses like email writing miss the transformative potential entirely.
The Learning Investment That Changes the Game
The Economist piece that Adele references illustrates a counterintuitive truth: AI adoption requires a deliberate productivity dip upfront . Teams spend time experimenting with new tools, reading documentation, running tests, and discovering use cases beyond obvious automation. This is not wasted time—it's the investment period that determines long-term returns.
Adele has witnessed this pattern directly at Arm, where ChatGPT Enterprise was rolled out in early 2025 to approximately 10,000 employees. She emphasizes that organizations must dedicate genuine time to learning , moving beyond quick wins like automating routine emails to understanding how AI can reshape core workflows, decision-making, and creative work.
The stakes are real: according to Microsoft's 2026 Work Trend Index, 58% of AI users now produce work they could not have a year ago . That shift only happens when communicators and teams accept the learning curve as non-negotiable, as Adele discusses in this episode .
"You have to put time aside and put time in to learn and experiment and read and do the training and experiment again and that does take time there's no kind of shortcut to it really."
Adele McIntosh — VP of Internal Communications and Community at Arm. At Arm for over eight years, McIntosh leads internal communications and other functions within a major technology company of approximately 10,000 employees operating in the semiconductor space. Arm designs and builds chips for applications ranging from smartwatches and phones to data centers and robots.
Planning for the Productivity Curve as a Communicator
For internal communications teams, the implication is clear: plan for an initial productivity dip when introducing AI tools . This is not a sign of failure—it's a sign of real adoption happening. Teams that rush through onboarding or expect instant payoff will underestimate the time needed to build competence.
The path forward, detailed further in this episode of Unprompted: Real AI in IC , involves messaging this investment cycle to leadership and teams alike. Communicators should set realistic expectations: productivity falls in month one or two while learning happens, then climbs as people move beyond surface-level experimentation to high-value applications in content creation, data synthesis, and strategic thinking.
One additional insight worth hearing from the full conversation: Adele explores how organizations can position AI adoption narratives to address employee concerns about job displacement and build trust in the transition—a consideration that runs parallel to managing the productivity dip itself and shapes how teams receive training and experimentation time.
Productivity with AI initially decreases because teams must invest time in learning and experimentation beyond surface-level uses.
There is no shortcut to unlocking AI's deeper capabilities—organizations that rush through adoption risk missing transformative potential.
58% of AI users now produce work they could not have produced a year ago, but only after accepting the learning curve.
Internal communicators should plan for and explicitly message the productivity dip to leadership as a normal, necessary phase of adoption.
What new skill sets are required for internal communications teams managing AI tools and GPT agents?
Internal communications teams now need skills to understand AI platforms, maintain AI agents, and monitor their performance —fundamentally different competencies from traditional IC work. This isn't simply about using AI as a tool; it's about managing AI as an active team member that requires oversight, training, and continuous optimization.
From Tool User to AI Manager
The shift is profound. When Adele McIntosh deployed ChatGPT Enterprise at Arm in early 2025, her team faced a new reality: AI agents behave like employees and need active management . Keeping those agents updated, ensuring they stay aligned with organizational goals, and catching performance issues before they escalate became part of the job description.
This represents a break from how internal communications teams have traditionally operated. As outlined in the Unprompted episode , the skills gap isn't about learning software—it's about understanding the principles behind AI behavior and how to govern it effectively.
"You have to put time aside and put time in to learn and experiment and read and do the training and experiment again and that does take time there's no kind of shortcut to it really."
Adele McIntosh — VP of Internal Communications and Community at Arm, where she has led internal communications and related functions for just over eight years. At Arm, a 10,000-person semiconductor company designing chips for applications ranging from smartwatches to data centers, McIntosh has overseen the integration of AI tools including ChatGPT Enterprise since early 2025.
What This Skillset Actually Looks Like
The new competencies fall into three categories. First, understanding how AI tools work —not as a user, but as an operator. Teams need enough technical literacy to recognize when an AI agent is producing reliable outputs and when it's hallucinating or drifting from intended use. Second, maintaining the agents themselves: updating prompts, refreshing training data, and keeping the system aligned with evolving business needs. Third, measuring and reporting performance like you would for any team member.
The adoption curve is steep, and as discussed in this podcast , there is no shortcut. Organizations like Arm are investing in ongoing learning and experimentation to build these capabilities from the ground up. For a deeper look at how Arm is reframing the entire AI adoption narrative internally, the full episode explores how to communicate AI's role to your workforce .
What is the future role of intranets and trusted information sources as AI agents become the primary interface for employee communications?
Intranets will not disappear in the short term because organizations need a single trusted source of truth from which AI pulls information . An AI layer will sit over this infrastructure as the primary interface, delivering personalized content to employees, but communicators must ensure people see organizationally important messages—not just algorithmically preferred ones—and preserve in-person experiences where human connection happens.
The intranet as infrastructure, not interface
The future architecture is a layered one. The intranet remains the authoritative repository—the backend engine where all organizational knowledge lives and is actively curated. Rather than disappearing, it evolves into the backbone that AI agents query to surface relevant information to individual employees.
As detailed in Unprompted: Real AI in IC , Adele McIntosh emphasizes that without this trusted source of truth, AI cannot reliably deliver accurate, vetted organizational content. The intranet's role shifts from being the place employees visit to being the place AI pulls from on their behalf.
Proactive delivery meets human curation
The AI layer becomes the new front door—the conversational interface employees interact with daily. It learns preferences and delivers personalized information proactively, saving employees from having to search. Yet communicators must actively prevent algorithmic bias from drowning out strategically important messages that employees may not have explicitly requested.
This distinction is critical: ChatGPT or a Teams-integrated AI agent might naturally surface popular, engaging content. But internal communicators must ensure compliance updates, leadership announcements, and organizational priorities break through the algorithmic noise and reach their intended audiences regardless of engagement metrics.
"You have to put time aside and put time in to learn and experiment and read and do the training and experiment again and that does take time there's no kind of shortcut to it really."
Adele McIntosh — VP of Internal Communications and Community at Arm. With over eight years leading internal communications at the semiconductor company (approximately 10,000 employees), McIntosh is navigating the shift to AI-driven employee engagement while maintaining organizational voice and cultural coherence across global operations.
The irreplaceable human element
Even as AI automates content delivery and personalization, in-person experiences remain non-negotiable . No algorithm can replicate the informal conversations, spontaneous connections, and human trust that happen when colleagues gather physically or through live conversation.
The best internal communication strategies of the next five years will blend three layers: a curated intranet (infrastructure), AI agents (interface), and deliberate human gatherings (culture). Communicators who design for all three—not just optimizing for algorithmic reach—will sustain organizational alignment as the technology transforms how information flows.
How should organizations frame internal narratives around AI adoption when addressing concerns about job displacement and ethical issues?
Organizations should anchor their AI narrative around philosophy, ambition, and human judgment rather than technology alone —then cascade the CEO's vision downward through leaders and managers with structured support. This approach combines AMAs with leadership, transparent viewpoints on ethical and environmental concerns, and continuous FAQs that evolve as organizational understanding develops.
The foundation is rejecting the "big launch" mentality. Instead of announcing AI adoption as a single organizational event, the strategy involves multiple touchpoints designed to address skepticism directly. As Adele McIntosh explains in the podcast , this means preparing leaders at every level—not just the C-suite—to have authentic conversations about both opportunity and concern.
Building trust through acknowledgment, not reassurance
The most effective framing doesn't dismiss fears about job displacement; it acknowledges them explicitly and positions AI as a tool that amplifies human capability rather than replaces it. At Arm, this involves crafting messaging that centers on human judgment and decision-making as irreplaceable, even as the technology changes what work looks like.
This requires preparation for managers. They become the primary communicators to their teams, so they need clear talking points, FAQs for difficult questions, and permission to say "I don't know yet." According to the approach detailed in this episode , organizations should provide managers with templates and frameworks so their answers feel consistent without being robotic.
Making the narrative evolve deliberately
One striking element of this strategy is recognizing that the AI narrative cannot stay static . The technology moves faster than internal understanding, so organizations should plan for continuous FAQs and regular all-hands AMAs where leaders directly address new concerns as they emerge.
Adele's team at Arm also addresses ethical concerns—environmental impact, algorithmic bias, data privacy—not in a separate ethics channel but woven into the main narrative. This prevents a "business vs. ethics" false split and demonstrates that the organization is thinking holistically about AI adoption. For more on how internal communicators can embed ethical thinking into AI strategy, the full discussion covers additional frameworks and real examples .
"You have to put time aside and put time in to learn and experiment and read and do the training and experiment again and that does take time there's no kind of shortcut to it really."
Adele McIntosh — VP of Internal Communications and Community at Arm, where she has led internal communications and related functions for just over eight years. Arm is a semiconductor leader with approximately 10,000 employees, designing chips for applications ranging from smartwatches and phones to data centers and robots. McIntosh brings hands-on experience managing AI adoption communications for a major technology organization navigating the precise balance between opportunity and employee concern.
What approach should organizations take when redesigning processes for AI integration rather than simply adding AI to existing workflows?
Start with a blank sheet of paper —understand what AI can do now and what it might do in the future, then ask how to reach your goal entirely differently. Avoid bolting AI onto existing processes; instead, redesign from the ground up, which sometimes reveals that automation or a different solution altogether is the real answer, not AI.
When organizations begin their AI journey, the natural instinct is often to look at what they already do and find where AI might fit in. But as discussed in the Unprompted podcast , this approach misses the real opportunity. The shift requires letting go of established norms that dictate how work has always been structured.
Rethink the path, not just the tools
The blank sheet of paper exercise is foundational. You start by asking: what does AI enable us to do that we couldn't do before? What might it do in the next year or two? Then—and this is critical—you ask a different question: how would we do this entirely differently if we designed from scratch? This is not incremental improvement; it's fundamental redesign.
One of the paradoxes Adele McIntosh uncovered during her work at Arm, a semiconductor company with roughly 10,000 employees, is that sometimes the answer isn't AI at all. When you remove the preconception that "we need to use AI here," you might discover that automation or a completely different solution solves the problem faster, cheaper, or better. Using AI for the sake of using it is a costly mistake.
The Microsoft 2026 Work Trend Index found that 58% of AI users report producing work they could not have produced a year ago . But that leap in capability doesn't happen by accident—it happens when you fundamentally rethink the process. There's a deeper insight embedded in the podcast episode about how organizations trained and iterated on these new approaches, worth exploring the full conversation .
"You have to put time aside and put time in to learn and experiment and read and do the training and experiment again and that does take time there's no kind of shortcut to it really."
Adele McIntosh — VP of Internal Communications and Community, Arm. With over eight years at Arm, McIntosh leads internal communications and related functions for the global semiconductor design company, bringing deep experience in scaling communication strategies across organizations navigating rapid technological change.
Begin process redesign with a blank sheet of paper: understand AI's current and future capabilities before restructuring around them.
Resist the impulse to add AI to existing workflows; instead, ask how you would solve the problem differently if building from scratch.
Sometimes the answer is automation or an entirely different tool—using AI without that interrogation is waste.
Genuine AI-driven outcomes require real time investment in learning, experimentation, and iteration; there are no shortcuts.
How can internal communicators use AI beyond basic email writing to elevate strategic work?
Move AI from basic automation to intelligent partnership by creating custom AI personas that act as specialized thinking partners —an internal comms expert, a marketing strategist, an engineer—brainstorming together on real strategic challenges. Pair this with an AI-powered chief-of-staff agent that monitors your email, Slack, and Teams channels to flag urgent and overdue items, letting you focus your mental energy on high-impact decisions rather than triage.
Using AI as a collaborative thought partner
The shift from tool to partner starts with how you configure your AI. Instead of using ChatGPT or ChatGPT Enterprise only for one-off tasks, create distinct personas that represent different expertise domains. As Adele McIntosh explains in the Unprompted episode , she sets up personas like an internal comms expert, a marketing expert, and an engineering expert, then runs them through collaborative brainstorms on real strategic questions.
This approach surfaces cross-functional perspectives without requiring lengthy meetings . You get rapid iteration on messaging, audience strategy, or organizational challenges by prompting each persona with the same prompt and comparing their distinct angles. It's simultaneous thinking from multiple disciplines, compressed into minutes.
Automating awareness without losing human judgment
Beyond brainstorming, McIntosh created a chief-of-staff agent that monitors her communications in real-time across email, Slack, and Teams . This agent flags urgent items, reminds her of overdue follow-ups, and surfaces context she might have missed in the constant flow of messages and requests.
The goal is not time savings—it's decision clarity. Strategic work in internal communications demands that you know what's actually urgent versus what feels urgent, what stakeholders are waiting on, and where organizational friction exists. An AI agent handling this triage means you spend your cognitive energy on judgment calls and strategy, not on manually scanning messages.
"You have to put time aside and put time in to learn and experiment and read and do the training and experiment again and that does take time there's no kind of shortcut to it really."
Adele McIntosh — VP of Internal Communications and Community at Arm. With over eight years at Arm, a semiconductor company of approximately 10,000 employees, McIntosh leads internal communications and manages the strategic alignment of messaging across a large, distributed organization where engineering, marketing, and business functions operate under competing priorities and timelines.
This quote underscores a critical truth often missed in AI adoption: the tool is only as effective as your willingness to invest in learning it. McIntosh details her own experimentation process with ChatGPT Enterprise at Arm, rolled out in early 2025, emphasizing that shortcuts do not exist for deep AI fluency. You must practice, fail, refine, and practice again.
Reframing AI impact: quality over speed
According to Microsoft's 2026 Work Trend Index, 58% of AI users report producing work they could not have created a year ago —a statistic that matters far more than time-saving metrics. This is the actual value at stake: not "I finish my emails 30 minutes earlier," but "I can now model three different crisis communication scenarios before the leadership call, or stress-test a message across personas I never would have had time to consult."
Strategic work in internal communications requires this shift in thinking. The personas and monitoring agents are not shortcuts; they are force multipliers for judgment, creativity, and stakeholder understanding.
Set up multiple AI personas representing different expertise areas (internal comms, marketing, engineering) to run simultaneous collaborative brainstorms on strategic questions.
Deploy an AI chief-of-staff agent to monitor email, Slack, and Teams in real time, flagging urgent and overdue items so your focus shifts to judgment and strategy.
Invest genuine time in learning and experimenting with AI—there are no shortcuts to developing fluency that translates to higher-quality output.
Measure AI impact by the quality and novelty of work produced, not by time saved; 58% of AI users report creating work they could not have a year ago.
S01 EP2 - Balancing AI and human judgement in IC, with Amanda Atkinson
What framework does Amanda recommend for internal communicators learning about AI?
The Anthropic AI Fluency course—free and structured around the four Ds (direction, diligence, delegation, and discernment)—is the framework Amanda recommends for internal communicators learning AI. Discernment emerges as the most critical pillar: knowing when to use AI, when to step back, and how to maintain your own judgment. Beyond technique, Amanda stresses that internal communicators must remember their own humanity and expert value remain irreplaceable assets in an AI-driven world.
The four Ds provide a practical scaffold that moves beyond simple tool literacy. Direction means knowing your intent before touching an AI tool —clarity about what you're trying to solve or explore. Diligence is about rigorous testing and validation; you don't accept AI output at face value. Delegation is understanding which tasks genuinely benefit from AI assistance versus which need your human expertise and judgment. And discernment ties it all together: the ability to step back and ask whether AI is truly the right lever for this moment.
As Amanda explains in the episode , this framework emerged from her own experience navigating AI as both an individual practitioner and a leader guiding teams through adoption. She's clear that the course is free and accessible , removing any barrier for communicators who want structured guidance without the noise of vendor marketing or hype.
Why discernment matters more than speed
Internal communicators often feel pressure to move fast—to embrace AI, automate faster, ship content quicker. But discernment asks the harder question: Is this the moment for speed, or is this the moment for human voice, nuance, and judgment? Discernment is about knowing the difference , and Amanda underscores that this is the foundation on which all other AI use rests.
This connects directly to a broader truth she returns to throughout this conversation : your expertise as an internal communicator—your understanding of organizational culture, your ability to sense what a message truly needs, your credibility with teams—doesn't disappear because AI exists. These human strengths complement AI; they don't compete with it. The framework works only if you hold onto that clarity.
"Where I start with AI, I find myself just getting a mess. Where I start with my own work and then use AI to enhance and challenge, that's where it works best for me."
Amanda Atkinson — Communication and Culture Leader at the Science Based Targets Initiative and Founder of Create Connect. Over seven years at Salesforce leading internal comms for EMEA, Amanda developed a deep tech-driven mindset and now positions herself as an AI intrepid explorer, committed to exploring AI's possibilities while honestly acknowledging the learning curve ahead.
This insight—starting with your own clarity before adding AI into the equation—mirrors the "direction" pillar of the framework. Too many communicators throw their half-formed ideas at ChatGPT or Claude, hoping AI will rescue them. Instead, Amanda's path reverses that: own your thinking first, then let AI enhance or challenge it. A detailed breakdown of how to apply this practically in your daily workflow appears in Amanda's method for capturing organizational insights , which complements this framework beautifully.
Holding onto your humanity in an AI world
Perhaps the most grounding part of Amanda's answer is her reminder: AI is exciting and must be embraced, but internal communicators must not forget their own humanity and the expert value they bring. That value is not going anywhere. The framework acknowledges AI as a genuine tool—powerful, useful, worth learning—while rejecting the narrative that adopting it means abandoning who you are or what you know.
For a full discussion of how this plays out in team dynamics and strategic work, explore Amanda's thoughts on AI productivity gains versus strategic work , where she digs into why simply saving time doesn't automatically translate to better outcomes.
The Anthropic AI Fluency course provides a free, structured framework built on four pillars: direction, diligence, delegation, and discernment.
Discernment—knowing when to use AI, when to hold back, and when to trust your own judgment—is the most important element of the framework.
Start with your own clarity and expertise first; use AI to enhance or challenge your thinking, not to rescue half-formed ideas.
Your humanity and expert value as an internal communicator remain irreplaceable, even in an AI-driven environment.
What is the relationship between AI productivity gains and strategic work in internal communications?
No—saving time through AI does not automatically translate to strategic work, warns Amanda Atkinson. If you are already stuck in tactical work with limited time to be strategic, AI will simply help you do tactical work faster, not break the cycle . The real opportunity is to invest those saved hours into exploring AI itself and identifying specific use cases that align with your actual needs.
The assumption that efficiency gains equal strategic freedom is seductive but misleading. Amanda emphasizes that time saved without intentional redirection will vanish into the same tactical patterns that already consume your day. You must consciously decide where reclaimed time goes, or it defaults to filling the gaps left by faster work.
This distinction matters enormously in internal communications, where practitioners often report having only about 30 minutes or less per day to communicate with their teams. Adding AI-assisted tactical tasks to this already fractured schedule does not solve the underlying structural problem—it just makes the problem move faster.
The trap of efficiency without intention
Amanda's caution cuts against a common narrative in AI adoption: the idea that tools automatically liberate knowledge workers. Instead, she argues that the burden of intentional work allocation shifts entirely to you . If you lack clarity on what strategic work looks like or how to prioritize it, AI acceleration becomes a distraction rather than a catalyst.
As explained in this episode of Unprompted , the most effective approach is to start by understanding your own work first, then use AI to enhance it. Amanda describes her own process: "Where I start with AI, I find myself just getting a mess. Where I start with my own work and then use AI to enhance and challenge, that's where it works best for me." This principle applies equally to reclaimed time—you must know your priorities before you can redirect savings toward them.
The consequence is practical: if you are truly resource-constrained and struggling to find breathing room for strategy, the first investment should be time spent exploring AI use cases specific to your role and organization . This exploration itself becomes the bridge between tactical acceleration and strategic clarity. Without it, productivity gains simply compound the problem.
Where to invest the time AI saves
Amanda recommends a deliberate reallocation strategy. Rather than assuming freed time will naturally flow toward strategy, you should invest those hours into learning what AI can genuinely do for your comms challenges. Discussed at length in the full conversation , this approach treats AI exploration as strategic work in itself—because identifying the right tools and workflows is the foundation for all downstream decisions.
This perspective reframes the entire relationship between productivity and strategy. Time saved becomes an investment fund, not a vacation . The question shifts from "What will I do with extra hours?" to "What do I need to understand about AI to unlock real value for my team?"
"Where I start with AI, I find myself just getting a mess. Where I start with my own work and then use AI to enhance and challenge, that's where it works best for me."
Amanda Atkinson — Communication and Culture Leader at Science Based Targets Initiative and Founder of Create Connect. Amanda has built her career in internal comms across multiple industries, spending seven years leading internal comms for EMEA at Salesforce, where she developed a deeply tech-driven mindset. She describes herself as an AI intrepid explorer, committed to understanding AI applications while remaining honest about the learning curve ahead.
One practical detail that surfaces in the episode but extends beyond this question alone: Amanda's use of a personal Slack channel to capture insights throughout her day—voice notes, forwarded emails, ideas, and observations—which she later synthesizes. This micro-habit hints at a larger workflow philosophy that shows why understanding your own work patterns comes before optimizing them with AI.
AI saves time on tactical tasks, but does not automatically redirect you toward strategy unless you have a deliberate plan for those reclaimed hours.
If you are already trapped in tactical work, faster tactical work via AI will deepen the cycle, not break it—you must interrupt the pattern through intention, not speed.
The highest-value use of AI-saved time is exploring specific AI use cases and workflows that match your real organizational needs and communication challenges.
Starting with your own work and priorities, then enhancing them with AI, produces better results than starting with AI capabilities and retrofitting your workflow around them.
How can internal communicators practically capture and synthesize organizational insights using AI?
Build a personal Slack channel to capture raw insights throughout your day —ideas, voice notes, videos, forwarded emails, and files—then synthesize this collection regularly using AI tools like Gemini to extract principles and actionable checklists. This method removes friction from knowledge capture and transforms scattered feedback into structured communication frameworks without requiring additional meeting time.
A single channel as your thinking partner
Amanda Atkinson's approach starts simple: designate a dedicated Slack channel as your personal inbox for organizational signals . Throughout the day, she captures everything—quick observations from one-to-one meetings with leadership, tone shifts in team feedback, ideas that emerge during conversations, even forwarded emails that contain valuable context. No filtering, no editing at capture time. The goal is velocity, not polish.
This is not a journal or a brainstorm document. It's raw material deliberately gathered for later synthesis . Think of it as collecting fragments of organizational truth as they happen, before the context fades or the nuance gets lost in translation.
From capture to framework in one synthesis pass
Once or twice per week, Amanda reviews this accumulated material and feeds it into an AI synthesizer—in her case, Gemini. The AI doesn't generate her thinking; it helps her organize and extract the patterns she's already sensed. This synthesis step turns scattered observations into communication principles and practical checklists that directly inform her regular biweekly communications to the organization.
As Amanda explains in the episode , this method works best when applied to work she's already done herself. For example, feedback and tone from CEO one-to-ones carries context and weight that she notes during the meeting. When that material is synthesized, it slots naturally into her communication strategy without requiring her to revisit or re-contextualize those conversations.
"Where I start with AI, I find myself just getting a mess. Where I start with my own work and then use AI to enhance and challenge, that's where it works best for me."
Amanda Atkinson — Communication and Culture Leader at Science Based Targets Initiative and founder of Create Connect. After seven years leading internal comms for EMEA at Salesforce, she has built a career using technology and automation to amplify organizational communication. She describes herself as an AI intrepid explorer, committed to honest exploration of both the opportunities and the learning curve ahead.
A deeper look at how Amanda navigates the tension between AI automation and human judgment is explored further in the full episode , where she also discusses why starting with a strong point of view matters more than letting AI lead the analysis.
Why this works when time is scarce
Internal communicators report having only about 30 minutes or less per day to connect with their teams . The personal Slack channel method preserves focus during meetings and conversations—you're not stopping to document, you're simply flagging insights as they surface. The synthesis work happens asynchronously, in batches, using AI as a partner to spot patterns that might otherwise require hours of manual review.
This shifts AI from "content generator" to "sense-maker"—a role that respects both the communicator's judgment and the organization's lived experience. Amanda's full conversation on Unprompted also covers how this principle applies to strategic use cases like analyzing all-staff feedback, showing the breadth of her approach across different communication challenges.
Create a dedicated personal Slack channel to capture ideas, voice notes, emails, and feedback in real time throughout your day—no filtering required at capture time.
Synthesize this raw material weekly using AI tools like Gemini to extract patterns, principles, and actionable checklists that inform your communication strategy.
Start with your own thinking and let AI enhance and organize it, rather than using AI to generate ideas from scratch—this yields clearer, more authentic frameworks.
This approach respects time scarcity: meetings stay focused on connection, and synthesis happens asynchronously, turning scattered insights into structured communication without extra meeting overhead.
Does AI need to lead data analysis work, or does the user need clarity on desired outcomes first?
Yes—you absolutely need to know what you want to achieve before using AI for data analysis. A clear objective allows AI to help you find correlations and compare data points across different lenses, a task that would have required extensive manual analytics work and significant time investment in the past.
Starting with AI and letting it lead the analysis typically produces unfocused results. The most valuable approach combines your own work and insights with AI enhancement, rather than asking AI to find patterns in raw data from scratch. Amanda Atkinson, Communication and Culture Leader at Science Based Targets Initiative, describes this as the critical difference between exploratory floundering and purposeful analysis.
The distinction matters because while there is merit in exploring what AI tools can do, exploration without direction wastes both time and computational resources. As discussed in the episode , the real power of AI in data analysis emerges when you pair it with a human point of view—your hypothesis, your business question, your desired outcome—and then let the tool accelerate the correlation-finding and cross-lens comparison work that would have previously required dedicated analytics support.
"Where I start with AI, I find myself just getting a mess. Where I start with my own work and then use AI to enhance and challenge, that's where it works best for me."
Amanda Atkinson — Communication and Culture Leader and Internal Comms Person at Science Based Targets Initiative. Amanda has built her career in internal comms across several industries, always seeking ways to leverage technology and automation. She spent seven years at Salesforce leading internal comms for EMEA, where she developed a deeply tech-driven approach to her work. She describes herself as an AI intrepid explorer, enthusiastically investigating AI applications while maintaining a grounded acknowledgment of the learning curve ahead.
For a practical example of how this plays out, Amanda used AI to analyze feedback from the Science Based Targets Initiative's first all-staff off-site in Amsterdam, a meaningful detail expanded further in a dedicated segment of the podcast .
Why clarity of intent matters more than tool capability
The temptation to let AI-driven exploration lead your analysis is understandable; the tools are powerful and appear to offer unlimited analytical possibility. However, boundless exploration without a hypothesis typically produces noise rather than insight. AI is most effective as an accelerator of your thinking, not as a replacement for it.
When you arrive at AI with a clear question—such as "What themes emerge most frequently in our team feedback?" or "Which departments report the highest engagement?"—the tool can compare data points across multiple dimensions far faster than manual analysis would allow. This transformative speed gain only materializes when you know what question you are asking.
What is a concrete strategic use case for AI in internal communications?
Use AI to bridge the gap between how employees feel about an event and whether your strategic messages actually landed. Amanda Atkinson combined post-event feedback data with original event objectives, then asked AI to identify gaps between emotional response and message absorption—revealing that while cultural connection was strong, key messaging around strategy and ways of working had not been fully absorbed.
Turning Event Data into Strategic Insight
When the Science Based Targets Initiative held its first all-staff off-site in years in Amsterdam, Amanda saw an opportunity to test a real strategic application of AI. Feedback was overwhelmingly positive—employees felt energized and connected. But "positive" doesn't always mean "strategically effective." She took the raw feedback data and combined it with the original session objectives and desired outcomes for each part of the event.
The next step was direct: Amanda fed both the feedback and the objectives to AI and asked it to identify gaps between what people felt emotionally and whether the strategic messages had truly been absorbed. This isn't AI replacing judgment—it's AI enhancing the analysis that Amanda herself would have to manually piece together , but faster and more systematically.
What the Data Revealed
The results told a clear story: cultural connection was strong , but key messaging around strategy and ways of working had not fully landed with the audience. That gap would have been harder to articulate without the structured AI analysis. Amanda now had a data-backed narrative for stakeholders—not just "people loved it," but "people felt connected, and here's where we need to reinforce strategy messaging next time."
As Amanda explains in the episode , this kind of analysis starts with her own work: "Where I start with AI, I find myself just getting a mess. Where I start with my own work and then use AI to enhance and challenge, that's where it works best for me." She brought the human strategic thinking first, then let AI accelerate the pattern-matching.
"Where I start with AI, I find myself just getting a mess. Where I start with my own work and then use AI to enhance and challenge, that's where it works best for me."
Amanda Atkinson — Communication and Culture Leader and Internal Comms Person at Science Based Targets Initiative. Amanda has spent seven years at Salesforce leading internal comms for EMEA, where she developed a tech-driven approach to communications. She is also founder of Create Connect and describes herself as an AI intrepid explorer, balancing enthusiasm for emerging AI applications with honest acknowledgment of the learning curve ahead.
If you want to hear how Amanda handled the specific challenge of analyzing feedback at scale without losing the human context , the full episode dives deeper into her toolkit and the unexpected obstacles she encountered.
How should internal communicators approach using AI for writing and content creation?
Don't start from scratch with AI—it produces poor results. Instead, begin with your own core narrative and use AI to adapt and scale it into supporting pieces like event invitations, Slack posts, and promotional copy. The real value emerges when you start with human work and use AI to enhance it, not the other way around.
Why starting with AI leads to a mess
Using AI as your first step in the writing process tends to generate generic, uninspired output that requires heavy reworking. Where you start with AI, you find yourself just getting a mess. The problem isn't that AI lacks capability—it's that without human direction rooted in your organisation's voice, values, and context, AI produces content that feels disconnected and requires substantial revision. This approach wastes time rather than saves it.
The winning approach: human foundation, AI acceleration
The most effective workflow begins with you writing the core message yourself. This might be a key memo, a strategic narrative, or your authentic take on a topic. Then use AI to enhance and challenge that work , multiplying it across different formats and channels. As discussed at length in this episode of Unprompted: Real AI in IC , internal communicators report that this sequence—human insight first, AI adaptation second—unlocks genuine productivity gains.
The practical application is immediate: write your all-hands talking points, then ask AI to turn that into a Slack announcement, an event invitation, or a calendar reminder in your preferred tone. AI becomes a scaling engine for work already grounded in human judgment and organisational knowledge. You maintain control, authenticity, and context while eliminating repetitive rewriting.
"Where I start with AI, I find myself just getting a mess. Where I start with my own work and then use AI to enhance and challenge, that's where it works best for me."
Amanda Atkinson — Communication and Culture Leader at Science Based Targets Initiative and Founder at Create Connect. Amanda has built her career in internal comms across several industries and has always sought ways to leverage technology and automation. She spent seven years at Salesforce leading internal comms for EMEA, where she developed a tech-driven mindset. She describes herself as an AI intrepid explorer, excited about exploring AI applications while maintaining honest acknowledgment of the learning curve ahead.
In conversations about AI adoption in internal comms, context and confidence matter enormously. Amanda Atkinson explores in detail how managers who openly support and model AI use increase employee adoption dramatically—employees with managerial AI support are 7.4 times more likely to say it helps them do their work. This same principle of human leadership applies to your writing: your human judgment must lead, and AI must follow.
Never use AI to generate content from zero—it produces generic, contextless output that requires heavy reworking and wastes time.
Start every piece with your own core narrative, voice, and organisational insight, then use AI to adapt and multiply it across formats.
Scale internal comms efficiently by using AI to convert one well-crafted message into event invites, Slack posts, emails, and promotional copy.
Maintain authenticity and control by keeping human judgment at the start of your workflow, not at the end.
S01 EP5 - Using AI as a change-comms wingman, with Ana-Maria Thijssen
What is 'bot sitting' and how widespread is it among workers using AI tools in 2024–2025?
Bot sitting is the time workers spend checking AI outputs, feeding missing context, and debugging errors instead of benefiting from efficiency gains. According to a study of 6,000 workers across the US, the UK, and Australia, 69% of workers now admit to bot sitting—wasting an average of 6.4 hours per week on these tasks.
The term captures a critical gap between AI's promised productivity boost and workplace reality. Workers are not using AI to eliminate work; they're using it to generate content that requires substantial rework and oversight.
This hidden cost of AI adoption is reshaping how organizations measure return on investment. What looks like AI acceleration on paper becomes a different story when workers spend their "saved" time validating, correcting, and contextualizing the tool's output. As discussed in the episode with Ana-Maria Thijssen , this reality hits especially hard for change communications teams managing large-scale transformations where accuracy and tone matter immensely.
The hidden workload behind AI efficiency
Bot sitting reflects a fundamental truth about AI tools: they are fast at generating text, but they are not smart about organizational context. A communications professional using Microsoft Copilot or similar tools may generate a draft in minutes, but that draft still needs fact-checking, tone refinement, and alignment with real organizational circumstances that the AI cannot know.
The 6.4-hour weekly average is not evenly distributed. Knowledge workers, particularly those in roles like change communications, often invest far more time in validation and correction than teams doing routine tasks. Ana-Maria Thijssen, working as the sole communications professional in her organization's global SAP transformation programme, highlighted how AI became her virtual colleague precisely because the tool's output always requires her expert judgment before it can be shared with 6,000 programme users across dozens of locations.
"AI became my virtual colleague. It's not a person that does my job. It's more like an entity that helps me bounce ideas and not feel so alone."
Ana-Maria Thijssen — Change Communications Lead (OCM/CMO team). Thijssen joined her current role approximately a year and a half after her organization's SAP IT transformation programme had already started, meaning she had to onboard rapidly with no formal induction. She works as the sole communications professional embedded within an Organizational Change Management team operating across a global, multi-wave programme spanning dozens of locations worldwide. Her colleagues, involved since January 2015, are a key source of institutional knowledge she leverages alongside AI tools to manage communications for the transformation's 6,000 users.
Bot sitting also reveals why organizations cannot simply hire fewer communicators or assume AI will slash internal communications budgets. The work is not eliminated—it is transformed into oversight, validation, and quality assurance. As explored in Unprompted: Real AI in IC , this shift demands different skills and creates new demands on the professionals who already manage multiple competing priorities.
Bot sitting describes time spent checking, debugging, and contextualizing AI output rather than eliminating work altogether—69% of surveyed workers admit to this practice.
Workers waste an average of 6.4 hours per week on bot-sitting tasks, turning AI's speed advantage into a hidden cost for organizations measuring productivity gains.
Change communications roles are particularly vulnerable to bot sitting because organizational context, accuracy, and tone cannot be delegated to AI without expert validation.
AI tools complement human expertise in high-stakes communication roles, but they do not reduce the need for experienced professionals to oversee output quality.
Will AI replace internal communications professionals, according to Ana-Maria Thijssen?
Thijssen is unequivocal that AI will not take communications jobs, arguing that AI lacks the human experience, organisational context, and people-centred…
What is the emotional value of AI for solo internal communications professionals, according to Ana-Maria Thijssen?
Thijssen identifies emotional support as one of the most significant, if unexpected, benefits of using AI: when working alone, self-doubt and professional…
How is Ana-Maria Thijssen's team preparing to scale change communications across 30 locations in the next wave of their SAP programme?
With the next wave covering 30 locations across Europe and America — compared to just three locations in earlier waves — Thijssen's team is focusing on…
What are the risks of relying on AI-generated references in change communications work, and how can practitioners mitigate them?
AI frequently produces dead links, fabricated references, and citations that lead only to a single blogger's unverified opinion. Ana-Maria Thijssen's mitigation is blunt: check every reference manually — confirm the website, magazine, or academic paper actually exists — and always ask AI specifically for academic research to raise the baseline quality of sources it returns.
Hallucinated citations aren't a edge case — they're the default risk
When Thijssen uses AI to support her change communications work, sourcing is where things break down fastest. The tool confidently produces URLs that return 404 errors, authors who don't exist, and journal articles that were never published. These aren't rare glitches — they're a predictable pattern that any practitioner leaning on AI for evidence-based communications needs to account for.
The danger in a change management context is particular. Communications that cite Deloitte, Gartner, or Harvard Business Review carry real authority with senior stakeholders. A fabricated citation to any of these sources — one that a colleague later googles and finds nowhere — damages credibility far more than having no citation at all. As Thijssen explains in Unprompted: Real AI in IC , the stakes of getting this wrong are not abstract.
Manual verification and the "ask for academic sources" prompt — Thijssen's two-step fix
Thijssen's approach has two distinct moves. First, she does not trust any reference AI returns without checking it herself — clicking through to the actual site, paper, or magazine to confirm it exists and says what the AI claims it says. There is no shortcut here: the check is manual, every time.
Second, she has refined how she asks for sources in the first place. Requesting "academic research" rather than generic references pushes the model toward peer-reviewed material rather than blog posts. It doesn't eliminate hallucination, but it raises the floor — a point she develops in detail in this episode of Unprompted .
This two-step discipline reflects a broader posture she describes throughout the conversation: AI is a starting point, not a finished product. Every output requires a human layer of scrutiny before it reaches any stakeholder, internal or external.
"AI became my virtual colleague. It's not a person that does my job. It's more like an entity that helps me bounce ideas and not feel so alone."
Ana-Maria Thijssen — Change Communications Lead, OCM/CMO team, large global corporation (undisclosed).
Thijssen is the sole communications professional embedded within an Organizational Change Management team running a global, multi-wave SAP IT transformation programme. She joined the programme roughly a year and a half after it had already launched — with no formal induction period — and has had to build contextual knowledge rapidly ever since. Her colleagues have been involved since January 2015 and remain a critical source of institutional memory. The next wave of the programme alone will span 30 locations worldwide. Working in isolation as a change comms specialist, Thijssen has developed AI-assisted workflows — including her reference-verification discipline — out of practical necessity, not experimentation for its own sake. Her full approach is covered in the Unprompted podcast .
That framing — AI as colleague, not replacement — is precisely why verification matters. A colleague can be wrong. You check their work. Thijssen applies the same logic to every source her AI wingman hands her, a discipline detailed further in S01 EP5 of Unprompted: Real AI in IC .
What approach does Ana-Maria Thijssen recommend for prompting AI to find reliable external research and case studies?
Thijssen insists on always asking AI to provide references alongside every idea it generates — not in a single list at the end, but paired with each…
How did Ana-Maria Thijssen use Microsoft Copilot to develop audience personas for a global change communications programme?
Thijssen's team fed documents, org charts, communication models, and examples into Microsoft Copilot and asked it to identify the most distinct personas…
How did Ana-Maria Thijssen use AI to get up to speed quickly when joining a mid-flight SAP transformation programme?
Thijssen arrived on a global SAP transformation programme roughly a year and a half after it had launched — with no formal onboarding and a team already fluent in dense IT and programme-specific jargon. She used AI to rapidly decode that language, eventually creating her own personal file she called a "Jargon Buster" — translating complex SAP and OCM terminology into plain, digestible content so she could contribute meaningfully from her very first days.
Arriving 18 months late — with no time to catch up the slow way
The programme Thijssen joined had been running since January 2015. Her colleagues carried deep institutional knowledge built over years. She had none of it — and there was no structured induction waiting for her.
Rather than spending weeks shadowing colleagues or working through documentation, she turned to AI as an accelerator. The tool let her query unfamiliar terms in real time, surface definitions, and cross-reference SAP-specific concepts against broader change management practice — at her own pace, without burdening a team that was already mid-delivery.
The result was the Jargon Buster: a personal reference file built with AI's help, capturing the programme's specific vocabulary in plain language. It was not a generic glossary. It was tailored to her programme's context, and she built it herself through iterative AI-assisted translation of real content she encountered on the job. You can hear her explain the approach in full on Listenly's episode page for Unprompted .
AI as the colleague who already knew everything — without making her feel behind
One of the subtler points Thijssen makes is about psychological safety. As the sole communications professional embedded in an Organizational Change Management team, she had no peer to turn to with a basic question about SAP terminology without risking looking uninformed in front of programme stakeholders.
AI removed that friction entirely. She could ask it anything — however elementary — and receive a clear, patient explanation with no social cost. It translated technical IT language into content she could actually use in communications, allowing her to produce relevant, credible materials even before she had fully absorbed the programme's history.
The programme itself was large: ultimately set to impact 6,000 users across the company, with the next wave alone covering 30 locations worldwide. Getting fluent fast was not optional — it was a condition for being useful.
"AI became my virtual colleague. It's not a person that does my job. It's more like an entity that helps me bounce ideas and not feel so alone."
Ana-Maria Thijssen — Change Communications Lead, OCM/CMO team, large global corporation (SAP IT transformation programme)
Thijssen joined her current role approximately 18 months after the programme had already started, stepping in with no formal induction into a team whose senior members had been involved since January 2015. She operates as the only dedicated communications professional within an Organizational Change Management function, supporting a multi-wave, multi-location global rollout. Working without a communications peer to consult, she built her practice around AI tools — including Microsoft Copilot — to compensate for missing institutional knowledge and to maintain the quality and speed her programme demanded.
How can AI act as a 'wingman' for internal communications professionals working alone on large change programmes?
Ana-Maria Thijssen describes AI as her 'virtual colleague' rather than a tool that does her job: it allows her to bounce ideas, check for best practice, and…
S01 EP4 - Getting people on board with AI at work, with Charlotte Halligan
What is the risk of organizations that publicly championed AI then scaling back their usage due to cost or poor ROI?
Charlotte notes that some organizations are currently scaling back AI usage because the economic cost is becoming prohibitive and the ROI is not there,…
How should organizations link their AI rollout narrative to company values and purpose rather than promoting AI for its own sake?
AI communication must be anchored to an organization's values, purpose, and mission — not floated as a standalone initiative. Charlotte Halligan argues that pushing people to use AI simply to hit an adoption target, without identifying where it genuinely adds value to the actual work being done, fails to address the underlying value problem. The golden thread linking AI to organizational purpose is the same principle that should underpin any internal communications campaign.
Halligan is particularly direct about what happens when the "why AI?" conversation gets disconnected from the "why we exist" conversation. Adoption numbers may climb, but hollow usage — people using tools mechanically to satisfy a metric — produces none of the outcomes organizations are hoping for. The question she pushes communicators to ask is not "how do we get people to use this?" but "where does this genuinely help our people do better work for our customers?"
In her view, the ideal AI rollout narrative goes one level deeper: it connects responsible and economically viable generative AI usage directly to how the organization helps its customers succeed. That framing transforms AI from an IT deployment into something that belongs to everyone's daily purpose. This episode of Unprompted: Real AI in IC explores exactly how internal communicators can build that connection credibly, drawing on Halligan's hands-on experience navigating a complex, multi-role, multi-jurisdiction global rollout.
"It's not necessarily that AI is going to take your job, but somebody who knows AI is going to take your job. So you kind of have to get on the bus or get off the bus."
— Charlotte Halligan, Comms Leader, Unprompted: Real AI in IC, S01 EP4 About Charlotte Halligan
CH Charlotte Halligan Comms Leader / Internal Communications Leader Charlotte Halligan brings more than 20 years of experience in internal communications to this conversation — a career that has spanned numerous technology implementations and organizational changes. That depth of experience means she has seen, firsthand, which change communication strategies hold up under pressure and which ones quietly collapse. She describes herself as someone who was initially slow to adopt generative AI tools personally: she did not rush to embrace ChatGPT or similar platforms at the outset. But once she began using them seriously, she recognized their real potential quickly and shifted her perspective. That arc — from cautious outsider to informed advocate — gives her particular credibility when advising others who are reluctant or anxious. Halligan then took on a role in which rolling out generative AI tools at enterprise scale was a central responsibility, running her own adoption journey in parallel with an organization-wide deployment. The complexity she navigated was significant: a global organization spanning multiple types of roles, legal frameworks, and regulatory environments. Her guidance on linking AI to values is not theoretical — it is the practical lesson she drew from doing this work live, under real organizational pressures.
What mistakes did Charlotte Halligan make during an enterprise AI rollout that others should avoid?
Charlotte identified three key mistakes: first, allowing clashing announcements — for example, headcount reduction notices going out at the same time as…
Why is plain-English policy writing critical to psychological safety during an AI rollout, and what did reducing jargon actually achieve?
Charlotte's organization initially had an acceptable use policy that ran to seven pages filled with terms such as LLM, large language model, and prompt…
How should organizations communicate with frontline or production-line employees who feel existentially threatened by AI and automation?
Charlotte found that production-line staff were far more terrified of AI than, for example, finance teams, because they saw automation as an existential…
How should internal communicators approach the hearts and minds side of an AI rollout when traditional know-feel-do models fall short?
Charlotte found that the traditional know-feel-do sequence does not work well for AI adoption, and that starting with the 'do' piece was far more effective.…
What is the biggest challenge of rolling out generative AI to employees compared to other technology implementations?
The core challenge is that generative AI is entirely unbounded: unlike rolling out a tool such as Salesforce — where success is clearly defined and adoption is eventually guaranteed — neither society nor organizations have yet defined what success or the future state actually looks like. Without a concrete destination to communicate, internal teams have no future-state story to rally employees around, and every individual arrives at the rollout with a wildly different starting point shaped by external narratives the organization simply cannot control.
With a conventional technology deployment, communicators can say with confidence: here is what the world looks like once we've crossed the finish line. That narrative anchor is what drives engagement and manages anxiety. Generative AI removes it entirely. Employees bring their own pre-formed assumptions — from news coverage, social media, and public debate — into the workplace on day one, and those assumptions vary enormously. Some arrive convinced AI will take their jobs; others dismiss it as hype; others are already power users of tools like ChatGPT in their personal lives. The organization inherits all of that variation without having shaped any of it.
The absence of a locked-down future-state vision also has direct knock-on effects on messaging. As Charlotte Halligan experienced first-hand when leading an enterprise-level generative AI rollout across a complex global organization, key messaging frameworks can take months to stabilize — in her case, roughly six months into the rollout before core messaging was fully locked down. This is not a communications failure; it reflects a genuine organizational reality that has no equivalent in a standard technology programme. The full episode on Listenly explores how she navigated these compounding challenges across multiple roles, legislations, and geographies.
"It's not necessarily that AI is going to take your job, but somebody who knows AI is going to take your job. So you kind of have to get on the bus or get off the bus."
— Charlotte Halligan, Comms Leader / Internal Communications Leader
About Charlotte Halligan
CH Charlotte Halligan Comms Leader / Internal Communications Leader Charlotte Halligan brings more than 20 years of experience in internal communications, which means she has lived through a significant number of technology implementations and organizational change programmes across her career — long enough to know what a "standard" tech rollout looks like and exactly where generative AI departs from it. By her own account, she was initially slow to adopt generative AI personally, approaching it with caution rather than enthusiasm. That changed quickly once she began using it directly, and she soon recognized its substantive potential for the profession. What gives her perspective particular weight on this question is that her personal adoption journey and her professional deployment challenge unfolded at the same time: she took on a role in which rolling out generative AI tools across an entire enterprise was a central responsibility, navigating a large, complex global organization spanning multiple workforce types, regional legislations, and regulatory frameworks. That dual vantage point — adopter and change lead simultaneously — is what makes her account of the challenge genuinely authoritative rather than theoretical.
How can AI be used to capture and replicate a senior leader's tone of voice for internal communications?
To replicate a senior leader's tone of voice with AI, feed the model approved writing samples and unscripted videos the leader created without any external scripting input. Those materials establish what the leader authentically sounds like. Once that baseline is set, draft your communication as usual, then run it through the AI as a tone check — and always review every output yourself before anything goes out.
This is the method Charlotte Halligan described on Unprompted: Real AI in IC . The key word throughout is "authentically." The source material fed into the AI must reflect the leader's own unfiltered voice — not content that communications teams have already shaped. That distinction matters because the whole point is to train the AI on what the leader genuinely sounds like, not on what the comms function has historically made them sound like.
The second — and equally important — principle is never to treat the AI's output as final. Halligan is explicit: AI will subtly insert generic best-practice ideas into its suggestions. Those insertions may read as perfectly competent, but they are not things an experienced communicator would choose, and they dilute rather than reinforce a leader's distinctive voice. The communicator must stay in charge of every editorial decision.
"It's not necessarily that AI is going to take your job, but somebody who knows AI is going to take your job. So you kind of have to get on the bus or get off the bus."
— Charlotte Halligan, Unprompted: Real AI in IC, S01 EP4
The method, step by step
1 Source authentic material only Collect writing and video the leader produced without any scripting from comms. This is the only source that reflects their real voice. 2 Programme the AI with those samples Input the approved material to build a tonal baseline the model can reference when checking subsequent drafts. 3 Draft your content as normal Write the communication using your own professional judgement. The AI is a check, not a ghost-writer. 4 Run the draft through the AI tone check Ask the AI to review the draft against the leader's established tonal profile and flag any mismatches. 5 Always review the output yourself AI will insert generic best-practice language that no experienced communicator would choose. Read every suggestion critically before accepting anything. About Charlotte Halligan
CH Charlotte Halligan Comms Leader / Internal Communications Leader Charlotte Halligan brings more than 20 years of experience in internal communications to a conversation that many IC practitioners are still approaching with caution. Over the course of her career, she has lived through numerous technology shifts and enterprise-wide implementations — giving her a grounded, unsentimental view of what actually changes when new tools arrive and what stays stubbornly the same.
She describes herself as a slow personal adopter of generative AI: she was not an early enthusiast. That makes her perspective distinctly valuable. Her turning point came when she experienced first-hand what AI could genuinely do, at which point she moved quickly. She subsequently took on a role where rolling out generative AI tools across a large, complex global organisation became a central responsibility — navigating the reality of multiple workforce types, jurisdictions, legislative requirements, and varying levels of digital readiness simultaneously with her own adoption journey.
That combination — deep IC experience, initial scepticism, and hands-on enterprise deployment — gives Halligan's practical advice on tone-of-voice replication a weight that purely technical commentary lacks. She knows both what communicators fear about AI and exactly where it earns its place in a professional's toolkit.