Agentic AI is the new darling of the tech world — software that can make autonomous decisions, learn from context, and act without waiting for us to click the next button. Its promise is alluring: smarter tools, frictionless journeys, and systems that anticipate our needs before we even articulate them.
But anticipation cuts both ways. The same technology that can help us might also quietly take agency away from us. And the definition of “help” isn’t neutral — it’s set by the people, companies, and algorithms designing the system.
In the rush to design “human-centred” digital systems, we often forget that humans are not separate from the environments we inhabit. A truly grounded humanism recognises that our technologies, economies, and cultures are embedded within ecological systems — the rivers, soils, forests, and climates that sustain us. When we centre only the human user without considering the non-human world, we risk building tools that meet short-term desires while undermining the long-term viability of life itself. In this sense, a more expansive humanism isn’t about placing people above nature, but about understanding our co-dependence with it. This means designing AI systems — agentic or otherwise — that measure success not just in personalisation metrics or efficiency gains, but in their capacity to support the resilience of the living systems we all depend on.
From reactive to proactive — and why it matters
For decades, our digital lives have been reactive. We act; the system responds. Agentic AI flips this model, shifting power toward systems that initiate action. In theory, this could mean better health alerts, faster resolutions to problems, and more relevant experiences.
In practice, it raises harder questions:
- Who decides what actions are “in our best interest”?
- How do we ensure proactive AI isn’t just another layer of manipulation for profit?
- Can we prevent the automation of bias, exclusion, and control?
Human-centred — but whose human?
The industry likes to talk about “human-centred design” as if humans are a single, uniform category. In reality, the centre is always chosen. It reflects cultural values, economic incentives, and political assumptions — often those of the most powerful.
If agentic AI is to serve us, it must be co-designed with the people most affected, across different contexts and perspectives. That’s not a nice-to-have; it’s the only way to prevent the tech from narrowing our possibilities instead of expanding them.
Designing for trust, not just efficiency
True human-centred growth in the age of agentic AI means shifting our success metrics away from pure efficiency and engagement. We should be asking:
- Does this system build trust over time?
- Does it make power more accountable?
- Does it strengthen — not replace — human decision-making?
The future of AI will be shaped less by its technical capabilities and more by the boundaries we set around its use. Agentic AI can be a tool for autonomy or a mechanism of control. Which one it becomes depends on how fiercely we defend the human in the loop.
Digital.
Agentic AI: From Anticipation to Autonomy — Who’s Steering the Ship?
In the swelling tide of AI hype, one term is starting to surface alongside the now-familiar “generative AI”: agentic AI.
Where generative AI builds content, code, and concepts on demand, agentic AI is built to act on our behalf — anticipating our needs, making decisions, and in some cases executing them without waiting for human approval.
For technologists, this feels like the next natural step in automation.
For those of us in the Communication Generation community, it raises sharper questions: when systems act for us, whose values are they carrying forward? What parts of human agency are we trading away in the name of convenience?
Agentic vs Generative AI — The Core Distinction
- Generative AI: Creates new artefacts — text, images, videos, music, code — based on patterns learned from vast datasets. Think ChatGPT, Midjourney, or GitHub Copilot.
- Agentic AI: Goes beyond creation to autonomous decision-making and action. These systems operate in dynamic environments, deciding what to do next in pursuit of a defined goal. Think personal assistants that book travel, negotiate contracts, or adjust supply chains — without a human pushing the buttons each step.
Why this matters:
As Douglas Rushkoff (Team Human) reminds us, tools are never neutral — they carry the values of the people and organisations that design them. Agentic AI collapses the space between intent and action. Once your intent is interpreted, the machine runs with it. If the interpretation is flawed, biased, or serving a misaligned interest, so too will be the action.
Why Human-Centred Isn’t Enough
At Something Digital, the “Personalised customer journeys” panel on agentic AI promises to explore more intuitive, anticipatory, and deeply personalised experiences. These sound like positives — until we ask: personalised for whose benefit?
As Paris Marx (Tech Won’t Save Us) often points out, the history of digital “personalisation” is largely a history of extracting more value for platforms and advertisers, not for people.
If agentic AI systems are designed with the same underlying incentives as our current algorithmic feeds, we’re not talking about liberation. We’re talking about more subtle, more pervasive behavioural nudging — at scale.
The New Power Equation
Agentic AI changes the communication equation in at least three ways:
- From reactive to proactive
Systems no longer wait for explicit input. They act on inferred signals, shrinking the space for conscious decision-making. - From static to adaptive relationships
Interactions aren’t just one-off — the AI builds a persistent model of “you” and updates it continuously. - From choice to orchestration
Instead of offering you options, agentic AI may simply choose. This creates efficiency, but also risk — especially if the “choice” serves another party’s goals.
Communication Generation’s Stake in This
Our brand exists at the intersection of ethics, technology, and cultural transformation.
Agentic AI isn’t just a technical trend — it’s a cultural architecture in the making. The design decisions being made today will ripple through the ways we work, socialise, and even think.
Questions we’ll be carrying into this conversation:
- How do we ensure transparency when actions are taken without explicit prompts?
- What mechanisms allow users to override, pause, or interrogate agentic AI’s decisions?
- Can “human-centred design” truly hold its shape when the system’s primary loyalty might be to profit or efficiency?
The People Behind the Session
This panel will bring together voices from multiple sectors, moderated by Philippa Fleming, Senior Director of Bid Management at NTT DATA Australia.
Fleming’s background in complex ICT services, stakeholder negotiation, and the development of AI tools for enterprise gives her a practical lens on how agentic AI could operate in real business contexts.
Her industry perspective is valuable — but equally important is the framing we, as attendees and citizens, bring to the table.
It’s on us to make sure the questions don’t stop at “how can we use this?” but extend to “should we use this — and in what way?”
Why This Conversation Is Urgent
Agentic AI is not five years away. It’s already here, quietly embedded in customer service, finance, logistics, and personal productivity apps.
Every rollout without public scrutiny sets new norms in the shadows. If we don’t interrogate them now, we’ll find ourselves living inside systems we had no meaningful role in shaping.
As Rushkoff warns: “When we accept the machine’s frame, we’re not the players anymore — we’re the playing field.”
Further Reading & Listening
- Paris Marx – Tech Won’t Save Us: Critical perspectives on technology’s social and political impacts.
- Douglas Rushkoff – Team Human: A rallying cry for reclaiming agency in the digital age.
- Ethan Zuckerman – Rewire: On building a more human internet.
- Better Internet Initiative: Practical efforts to create healthier online ecosystems.
About the Session Lead
This conversation on agentic AI is part of Something Digital Brisbane, led by a panel including Philippa Fleming, Senior Director of Bid Management at NTT DATA Australia, alongside industry experts from Creative HQ, Flight Centre, and others. Fleming brings two decades of experience in complex ICT services, stakeholder negotiations, and team leadership. She has also worked directly on developing generative AI tools for high-performance bid and pursuit teams — giving her a practical perspective on how AI is already transforming professional workflows.
While the panel features diverse voices, Fleming’s background in both technology adoption and human team dynamics offers a grounded, operational view of how “agentic” systems move from concept to reality.
Why This Matters to Communication Generation
At Communication Generation, we’re interested in more than the buzz around AI — we’re looking at the direction of travel. Sessions like this offer a window into how industry leaders are framing the shift from reactive to proactive technology, and what “human-centred” actually means when high-stakes automation is in play.
We see agentic AI as both an opportunity and a pressure point. It’s an opportunity if it genuinely gives people more control, time, and insight in their work and lives. But it’s a pressure point if “personalisation” becomes a euphemism for steering human behaviour at scale without transparency or consent.
Our interest is in how these technologies are designed, governed, and tested against real-world impacts — especially for the people who have the least say in shaping them. This session isn’t just about customer experience; it’s about who defines value, who holds the power, and how much human agency we’re willing to trade for convenience.
Agentic AI vs Generative AI
Why this matters now
Agentic AI and generative AI are on a collision course. One is built to act for us, the other to create for us — and when they merge, the stakes shift from “what can machines make?” to “what decisions will they make without us in the loop?” For those of us thinking about the future of communication, this isn’t a tech glossary exercise. It’s about the cultural architectures we’re building: whose values get encoded, whose voices are amplified, and whether human-centred design will still mean human-centred power.
| Dimension | Agentic AI | Generative AI |
|---|---|---|
| Core Function | Acts on behalf of a user or system, making autonomous decisions to achieve defined goals. | Produces new content (text, images, audio, video, code) based on learned patterns. |
| Primary Skill | Planning, reasoning, and executing actions in context. | Creative synthesis and expression of information. |
| Goal Orientation | Proactive — anticipates needs and initiates action without waiting for prompts. | Reactive — responds to prompts or input with generated output. |
| Human Role | Defines objectives, constraints, and values; may step back during execution. | Guides generation through prompts, edits, and feedback loops. |
| Potential Benefit | Reduces friction, saves time, handles complex or multi-step tasks. | Expands creativity, produces ideas or artefacts at scale. |
| Potential Risk | May overstep, removing human agency or making opaque decisions. | Can generate persuasive but false, biased, or low-quality outputs. |
| Ethical Focus | Accountability, transparency, and alignment with human values in autonomous action. | Integrity, originality, and safe use of synthetic media. |
| Cultural Question | Do we want systems to act for us — and on whose terms? | Who owns and shapes the culture these systems create? |
