It doesn’t speak. It doesn’t take coffee breaks. But it gets work done quickly.
In meetings, it quietly joins, listens, and then delivers a detailed analysis, a risk summary, and a neatly organised action plan. It’s not a colleague in the traditional sense, but it’s still reshaping the team dynamic.
This is Agentic AI – software that can reason, plan and act with minimal prompting. It doesn’t wait for instructions. It identifies goals, takes action, and adapts in real time.
With this power comes complexity.
Leaders must understand not only the functionality but also its implications because this next phase of AI isn’t just technological; it’s cultural. Navigating it well will separate the teams and organisations that thrive from those that struggle.
What does this mean for how we lead and collaborate?
The answer lies in understanding both the promise and the pitfalls and then redesigning work so that humans and machines elevate one another.
What is agentic AI?
Large language models (such as ChatGPT, Claude, and Perplexity) are generative AI models. They can answer questions and problem-solve.
Agentic systems can pursue objectives, plan, reason and execute end-to-end processes.
Microsoft’s 2025 Work Trend Index describes three phases:
- Assistants who remove drudgery
- Agents that run discrete workflows under human direction; and
- Digital colleagues who manage end-to-end processes while people handle exceptions.
Why the excitement?
Over the past couple of years, we have learned how AI can boost the productivity of individual knowledge workers, ranging from consultants to lawyers to
However, most knowledge work isn’t purely an individual activity; it often occurs in groups and teams. Teams are more than a collection of individuals – they provide critical benefits that individuals alone typically can’t, including better performance, sharing of expertise, and social connections.
So, what happens when AI acts as a teammate?
Thanks to research, we now know. Researchers conducted a randomised controlled trial of 776 professionals at Procter and Gamble. Their research found that when working without AI, teams outperformed individuals by a significant amount (0.24 standard deviations). Individuals working with AI performed just as well as teams without AI (0.37 standard deviation improvement over the baseline). The conclusion was that using AI can replicate the performance benefits of having a teammate.
However, when teams used AI, they saved 12 – 16% of time and were “…significantly more likely to produce these top-tier solutions”.
It’s worth reading Professor Ethan Mollick’s article for more details.
Interestingly, Mollick and colleagues’ research also found that people using AI reported more positive emotions than those who weren’t using AI. This finding aligns with Gallup’s advice to “play the long game” when it comes to AI. Organisations will achieve greater gains when they utilise AI in conjunction with their employees.
What’s the downside?
As I’ve written about before, you want to approach AI wisely.
It’s easy to get caught up in the hype. We’ve all seen tech bubbles before, and heard stories about AI bots hallucinating; it’s also easy to try and avoid change that makes us feel uncomfortable.
Neither approach is helpful.
Employees hesitate when systems feel opaque, rigid and overly complex, and they won’t trust using AI if they don’t trust their leaders.
As KPMG reported, two-thirds of people now use AI regularly, yet only 36% trust it, signalling both appetite and caution. For example, research reveals that top performers sometimes avoid powerful AI, worried it could deskill them or erode their edge.
There are also well-placed fears about job displacement, and headlines fuel anxiety. There are debates about whether older, more experienced workers or younger, less experienced workers will be most impacted by AI. There are also clear signs of the professions at most risk.
One downside that often gets little attention is the environmental impact of AI. AI already consumes 1-2% of global electricity and could reach 500 TWh by 2030 unless efficiency improvements are made.
Staying connected and human
Neuroscience reveals that perception, action, and emotion are intricately intertwined in ways that machines can’t yet replicate. Our brains intuitively see what actions an environment allows, while AI still struggles with this contextual judgement.
As we transition into a world where AI agents are an integral part of how we work, leaders will want to establish rituals that reinforce humanity.
- Firstly, design for dialogue. Pair every agent hand-off with a human check-in. Quick ‘stand-behind’ reviews keep relationships alive while letting the bot handle routine drafts.
- Slow down for sense-making. Encourage teams to pause after an agent delivers an output and ask: What assumptions sit underneath? What might we be missing? What context might we not have surfaced? This approach protects critical thinking and judgement.
- Re-centre empathy. Celebrate moments of human care and nuanced judgement that agents cannot provide. For example, coaching a distressed colleague rather than delegating sentiment analysis.
Helping your team members adapt
Whether you like it or not, agentic AI is here. The best approach is to build your and your team’s readiness. Here are five key steps to take.
First, boost AI literacy by creating safe environments, such as sandbox sessions, where employees can experiment, explore functionality and learn through trial and error. This is crucial because AI is not fail-safe, and team members need to check, challenge and verify. Practice builds confidence and judgement skills and sharpens the output.
Second, establish clear guardrails by setting principles for the responsible use of AI. These should include transparency about how systems operate, regular checks for bias, and mechanisms that ensure human oversight remains in place.
Third, govern access, not ideas. Rather than stifling innovation, provide employees with access to vetted and approved AI tools that are suitable for their specific needs. This reduces the risk of ‘shadow AI’ practices, where individuals use unauthorised tools that may expose the organisation to data breaches.
Fourth, measure both connection and output. As AI becomes more embedded in workflows, it’s vital to track not just productivity but also the quality of the human experience, including engagement, trust, and well-being. Without this balance, digital integration can silently erode culture.
Finally, model curiosity. When leaders openly engage with AI tools, asking questions, testing outputs, and reflecting on what works, they signal that experimentation is welcome and encouraged. This helps to normalise learning, lower anxiety, and encourages others to explore the possibilities without fear.
Also, remember that it’s not solely AI that’s changing the nature of how we work and live. As the World Economic Forum’s Top 10 Emerging Technologies of 2025 report reminds us that many innovations are powered by the convergence of different technologies.
A playbook for team members
As AI agents become part of the team, how you work with them matters. As a team member, here’s how to show up effectively and responsibly in this new dynamic.
Start by treating the agent as a partner, not just a prompt engine. Please don’t throw a quick prompt or command and hope for the best. Provide context, ask for the reasoning behind its output, and explore alternative perspectives. The more you engage thoughtfully, the more useful the combined contribution becomes.
Stay in the role of editor-in-chief. AI can draft, but it’s your job to ensure the final output is accurate, appropriately framed, and aligned with your organisation’s tone and values. Never delegate judgement.
Maintain good manners. While the AI may not care how you speak to it, people do and poor behaviour can become habitual. Others may read chat transcripts, emails, or shared documents. Using respectful language sets a tone for psychological safety and reinforces a culture of professionalism.
Invest in the capabilities that AI can’t replicate. Hone your judgement, deepen your emotional intelligence, and strengthen your influence with stakeholders. These human skills remain your competitive advantage and are increasingly valuable in an AI-enabled workplace.
Finally, act quickly when something seems off. If an agent produces incorrect, biased or questionable output, report it and contribute to the retraining process. Responsible use isn’t just about avoiding mistakes; it’s about building trust in how we collaborate with technology and with each other.
Towards cybernetic teamwork
Agentic AI is not a threat or a saviour; it is a force multiplier whose direction depends on us.
If we combine clear strategy, ethical guardrails, and genuine human connection, bots can free you and your team to do what humans do best: imagine, empathise, and lead.
Step ahead, step up. The bots are clocking in. Let’s ensure they make us better and wiser. And remember, this is a leadership test. Show up ready. Stay anchored in what matters so your influence positively lifts everyone around you.
Getting you ready for tomorrow, today®
Award-winning author and global workplace expert, Dr Michelle Gibbings, helps leaders, teams and organisations unlock strategic influence to accelerate progress.