For two years businesses have been trying to keep up with AI’s evolution. The promise of AI was that it would free us from mind-numbing rote work allowing us, with the time saved, to unleash our creativity and become more productive. It has, up to a point. But the shift to agentic AI signals new opportunities for companies and serious threats to employees.
Two years ago, when ChatGPT moved into the mainstream, we were both delighted and terrified. Generative AI felt like a nascent game changer, and as disruptors go, it has changed the game. We just didn’t realise how fast and unpredictable the disruption would be.
Despite the hallucinations and glitches, text-to-text generative AI tools quickly morphed into text-to-image and then text-to-video. It even spawned a new job title – a prompt engineer. This fast-moving technology tore through multiple sectors, from advertising to healthcare to education and into the corporate world, where a new Gen Z workforce embraced time-saving AI tools, which in turn exacerbated generational conflict issues.
But now we’re feeling a seismic aftershock of that initial disruption. The transition from generative AI to agentic AI is going to be more than a mere tremor underfoot: AI “agents” (another term for agentic AI) have been unleashed, and these agents are not just handy, time-saving tools, but systems designed for business efficiency.
AI agents are programmed to make decisions without human intervention while performing predefined tasks. Unlike the LLM (large language models) on which generative AI functions, agentic AI can access information outside of the data it has previously encountered, analyse it, and then take appropriate action.
In essence, an AI agent can go beyond its pretrained knowledge and function less like a chatbot and more like a human. It is designed specifically to drive measurable improvements in core business metrics such as sales optimisation and customer support automation.
As an example, Delta recently launched its new AI agent called “Delta Concierge”, which is designed to intuitively handle everything from rebooking delayed flights to suggesting the fastest route to your departure gate at various airports.
The integration of AI agents into a company’s operating system will have a massive impact, not only assessing the internal operating system, but also reducing the need for human interaction in its quest for optimum efficiency.
Currently, AI agents are being deployed as “co-pilots” that work alongside human workers to streamline their tasks or workflow. This is a double-edged sword. Scientists working in the field of AI are already warning of “AI dependency” as knowledge workers shift from “task execution” to “task stewardship”. They warn that this “cognitive offloading” will make us all more lazy and stupid. Who is the bot and who is the boss?
Workers at a Philippines call centre are discovering first-hand the pros and cons of working in tandem with an AI agent.
On the one hand, the AI agent, functioning as a “co-pilot”, assists the call centre agents with verbal prompts when dealing with customers, personalising each conversation. Instantaneous mass customisation is one of the advantages of agentic AI.
On the other hand, to improve efficiency, the agentic AI can feel like an automated helicopter manager. Agentic AI will monitor how long each call takes and rate an agent’s performance. If the agent’s call goes over the prescribed time, or if he or she pauses, mispronounces words, or deviates from the official script, the agent can earn a demerit. A series of demerits can result in probation, loss of incentives and even contract termination.
One call agent lamented, “AI is supposed to make our lives easier, but now I just see it as my boss. It’s like we’ve become the robots.”
A virtual helicopter manager might be a worker’s nightmare but it’s still preferable to redundancy. As more companies adopt agentic AI, overpaid and unnecessary (mostly) white-collar jobs will come into the firing line. Either companies will start retrenching or start with the more subtle (and less expensive) approach of eliminating job titles and positions that can be replaced by AI agents: “reverse redundancy”, if you like.
Last year the David Game College in London launched a new teacher-less course for twenty GCSE students. Lessons are given via laptop in combination with virtual reality headsets. The AI programmes learn what each student excels in and what they need help with, then adapts lesson plans for the term.
The students are not left alone. Three ”learning coaches” are physically on site to oversee the learning system, monitor progress, and to teach subjects like art and sex education, which AI agents are not designed to handle. Yet.
In terms of collateral damage, a less subtle approach was taken by an American local news station. Twenty staff were made redundant by a new agentic AI system, which was able to automate all their roles. Only one manager was retained to oversee the new system.
Unlike age-based redundancies, which was the norm, AI redundancies won’t discriminate. One of the news station staff said, “For the last decade I've worked in local news and have garnered skills I thought I would be able to take with me until my retirement. Now, at 30, all of those job opportunities for me are gone in an instant.”
To imagine how this plays out, meet Boardy, an AI “super connector” (with its own LinkedIn account) who managed to raise an $8 million seed round on its own. Boardy called his “human in residence” (yes, you read that correctly) just before he was going into a meeting with investors – to try and raise $4 million – letting him know that the round was closed and that he should not to bother with the meeting.
These case studies corroborate the 2025 WEF Future of Jobs survey, which found that 41 percent of employers intend to downsize their workforce as agentic AI proves its efficiency.
Future-proofed skills? Think again
Agentic AI’s impact on jobs now is beginning to emerge but how will the ripple effect affect skills needed for the future? The timeline to that future has just been fast-tracked.
Salesforce, one of the world’s largest cloud-based software companies, announced recently that it will no longer be hiring coders and software engineers, attributing the decision to a 30 percent productivity increase it achieved by using AI tools. It’s another example of “reverse redundancies”, which will become more prevalent.
With one strategic business decision, young tech workers now find themselves on uncertain footing. They believed they had tech skills that would carry them for most of their careers. AI has changed the game and their career prospects.
Parents who have been encouraging their children to learn coding, because it was seen as a future-proofed skill that could always be used as a career back-up, now find themselves scrabbling for new career guidance.
Business owners struggle to “meet the velocity of change” of new technologies, but now the same applies to the workforce.
KEY TAKEAWAY
Be careful what you wish for, as the saying goes. In just two years the pendulum has swung from AI being viewed as apocalyptic, to a utopian tool that would free humanity to flex their dormant creativity. It’s now evolved into an embedded system that can run a company’s operations and make autonomous tactical decisions.
Business leaders will have to weigh up, and wield, this form of automation very carefully, as well as rethink the definition of “future skills” and how they fit into new ecosystems. This prized business efficiency will come at a human cost, and evidence of collateral damage is already emerging.