I've been in enough boardrooms over the last year to notice a pattern.
Someone puts up a slide on cost pressures or growth targets, drops the phrase "we're investing heavily in AI", and somewhere in the room, the other question is already forming:
Can I fire my ops team?
AI has entered the executive vocabulary as both promise and pressure. Markets react sharply to AI announcements. Investors reward perceived efficiency. And layoffs, increasingly, are accompanied by references to automation.
Over the past year, organisations including Salesforce, Amazon, Citigroup and Meta have cited AI alongside restructuring efforts.
Sometimes the connection is direct. Sometimes it sits adjacent to wider economic or strategic shifts. Either way, the narrative to the outside world is clear enough: AI is coming for jobs.
Spend time inside those organisations, though, and the reality feels far more complicated. There are Copilot licences, pilot projects, half-built workflows and a scattering of enthusiastic individuals experimenting with tools. Curiosity and confusion in equal measure. What you do not see are fully autonomous systems replacing entire departments overnight. The lived experience feels transitional, exploratory, even messy.
Automation has existed for years. Tools such as Zapier and Microsoft Power Automate have long allowed teams to stitch systems together. What has changed is the cost and speed of experimentation. Large language models now introduce interpretation and reasoning into workflows. The API economy has matured to the point where nearly every SaaS platform can be connected to another. AI-assisted tooling can scaffold complex automations in hours rather than weeks.
When the cost of trying something collapses, behaviour changes. Problems that once required lengthy scoping exercises can now be prototyped in a day. The barrier to entry has fallen so far that experimentation often precedes formal strategy. The constraint moves away from technical feasibility and towards organisational maturity.
Against that backdrop, the operations function sits directly in the spotlight. Traditionally, operations teams have been responsible for monitoring signals, triaging requests, escalating issues and ensuring the machinery of the business continues to run. Much of that work involves structured decisions based on defined inputs. Automation platforms can now handle quite a lot of that flow. A support ticket can be summarised, categorised by urgency, analysed for sentiment and routed automatically. It can even draft a response or propose an action before a human intervenes.
The nature of the work itself begins to change.
Processing gives way to oversight. Execution gives way to orchestration. Modern platforms such as n8n increasingly incorporate human-in-the-loop patterns, where automation pauses for approval or guidance before continuing. The human becomes the governor of the system, shaping its rules and validating its outputs, rather than manually performing each step.
the demand for people who can design, govern and evolve automated systems grows.
This transition requires a change in how leaders define value. When execution becomes inexpensive and fast, the differentiator moves upwards. Judgement, prioritisation and system design grow in importance. The operations team's role shifts: the measure moves from how efficiently they process tasks to how intelligently they design and refine the workflows doing that processing. Their contribution stretches towards governance, optimisation and continuous improvement.
There is, however, a shadow forming alongside this opportunity. The ease with which automations can now be created introduces a new wave of decentralised experimentation.
Individuals spin up workflows. Teams create their own agents. Departments experiment in isolation. The pattern echoes earlier cycles of spreadsheet macros and low-code apps, where ingenuity flourished but oversight lagged behind. The difference today lies in speed and reach. An automation built in an afternoon can quickly underpin a mission-critical process.
Without clear ownership, such systems accumulate risk. Security vulnerabilities emerge. APIs are deprecated. Token costs rise unpredictably. The platform evolves while the workflow remains frozen in time. Governance must mature alongside experimentation.
Platform ownership, cost monitoring and security patching are the scaffolding that allows innovation to scale safely. Treat them as obstacles and the whole thing comes apart.
The token economics matter here too. AI-enabled workflows often rely on consumption from external model providers. In the early days of cloud computing, similar anxieties surfaced around elastic infrastructure costs.
Over time, forecasting models, monitoring tools and financial operations disciplines developed to manage that volatility. AI will do the same.
Token pricing may fluctuate. Providers may adjust commercial models. Organisations that approach the space with discipline will adapt more smoothly.
When leaders ask whether they can fire their ops team, they are reacting to a visible shift in what machines can now do. The temptation is to view automation as a direct substitute for human effort. In practice, the picture is more layered. Tasks may disappear. Entire categories of repetitive work may shrink dramatically. At the same time, the demand for people who can design, govern and evolve automated systems grows.
The organisations that thrive in this transition will be those that consciously migrate their people up the stack. Investing in training operations professionals to become system designers and workflow architects. Defining clear guardrails around experimentation.
Treating AI as a multiplier of human capability rather than a justification for reduction.
As execution accelerates and becomes cheaper, human contribution concentrates around areas where context, accountability and ethical judgement remain essential. The operations function does not vanish; it transforms. Whether that transformation strengthens or weakens the organisation depends largely on the choices leaders make now.
AI can be a scapegoat for cost cutting. It can also be a catalyst for structural improvement.
How do we redesign our organisation when execution is no longer the bottleneck?
Answering that well will determine who merely reacts to AI, and who uses it to build something more resilient.