Somewhere in a boardroom this year, a decision was made to give an AI agent a corporate credit card before figuring out what to do with the people who had been showing up, doing the work, and building that business for years. That is not a hypothetical. It is happening right now, and it is worth talking about honestly.
This Is Actually Happening
AI agents, software systems that can take autonomous actions like browsing the web, sending emails, booking meetings, and processing transactions, are increasingly being given direct access to company finances and systems. Tools like Anthropic’s Claude, OpenAI’s Operator, and various enterprise agent platforms are being provisioned with real spending authority inside real organisations. The infrastructure for this has moved fast. Faster, in many cases, than the conversations happening with the people sitting two floors below the boardroom.
At the same time, workforce reductions tied to AI are accelerating. Klarna, Duolingo, and a growing list of companies have publicly signalled that AI is replacing headcount. These are not fringe cases. The pattern is becoming visible enough that it warrants a direct conversation, not a think piece full of optimism about roles that do not yet exist.
So the sequence, in many organisations, has become, give the machine a budget, then work out what to do with the humans. That is backwards. And the businesses doing it that way are going to feel the consequences.
What the Sequence Reveals About Leadership Priorities
The order in which a business makes decisions tells you what it actually values, regardless of what the values statement on the website says.
Provisioning an AI agent with a corporate card requires procurement approval, IT security sign off, finance team integration, and executive sponsorship. It is a deliberate, multi step process. It does not happen by accident. Someone chose to prioritise that infrastructure.
Now compare that to the workforce transition planning happening in the same organisations. In most cases, it is minimal. Employees hear about AI through company wide emails or town halls after the tools are already purchased. Retraining programs, where they exist at all, are often bolt on afterthoughts rather than structured pathways. The people doing the work that AI is about to change are frequently the last to be brought into the conversation.
That gap is not a technology failure. It is a leadership failure. And it is the gap that causes 70% of AI adoption efforts to underperform or fail outright. The technology works. The human side is where things come apart.
The Real Cost of Getting the Order Wrong
There is a common assumption that moving fast on AI and moving carefully with people are in tension. That you have to choose one or the other. That assumption is wrong, and it is costing businesses more than they realise.
When you automate before you prepare your people, you do not just create uncertainty. You create active resistance. People who feel replaced rather than involved do not become advocates for the new way of working. They become friction. They find workarounds. They disengage. They leave, and they take institutional knowledge with them that no AI agent can replicate on a timeline that suits the business.
The businesses getting this right are not moving slower on AI. They are moving smarter. They are having the conversations with their teams before the tools arrive, not after. They are identifying which roles evolve, which workflows change, and what support people need to make that transition without it feeling like something being done to them. That preparation is not a delay. It is the thing that makes the technology actually work at scale inside a real organisation with real people in it.
What a Better Order of Operations Looks Like
The businesses that will look back on this period and feel good about how they handled it are the ones that started with a simple question, what does this mean for our people, and how do we bring them with us?
That does not mean slowing down AI adoption. It means sequencing it properly. Before you provision an agent with a credit card, you talk to the team members whose workflows that agent is going to touch. You explain what is changing and why. You identify what they are good at that AI cannot replicate, and you build a path toward more of that work, not less. You treat retraining as a real investment, not a checkbox.
This approach takes more thought upfront. It requires leaders to have conversations that are sometimes uncomfortable. But it produces a workforce that understands the change, trusts the direction, and actually uses the tools effectively rather than tolerating them.
The technology is the easy part. You can buy it today. The hard part is building the human conditions that make it worth buying. That is the work most organisations are skipping, and it shows.
The People Behind the Exit Packages
It is easy to talk about workforce transitions in aggregate. Headcount reductions. Restructuring. Operational efficiency. The language sanitises what is actually happening to real people whose work identities, financial security, and daily routines are changing in ways they did not choose and were not prepared for.
Every exit package represents a person. Someone who showed up. Someone whose experience and judgment and relationships were worth something to that organisation. Someone who deserved to be part of the conversation before a software subscription replaced their role.
This is not an argument against AI adoption. It is an argument for doing it with enough respect for the people involved that the order of operations reflects what you actually value. If the credit card came before the conversation, that tells you something about what was valued. The businesses that bring their people with them will still be standing and performing in five years. The ones that did not will be rebuilding something much harder to rebuild than a technology stack.
The technology is ready. The real question is whether your people are being prepared for what comes next, or just handed a package when it is already done.
Get in touch at www.xsiv.au/#form