This continues the “Will AI cause unemployment?” series. This piece focuses on the high-risk group of functions: where tasks can largely be standardized and verified, and are therefore more easily eliminated or largely replaced; and on functions whose processes are being accelerated, and will err faster if the structure is not rewritten.
The enhance / accelerate / eliminate three-axis framework (shared across this series)
- Enhance: people still make the judgments and hold the relationships, but AI amplifies materials, simulation, and insight.
- Accelerate: processes are compressed and cycle times shrink; if you do not rewrite the structure, you will only make mistakes faster.
- Eliminate: tasks that can be standardized, verified, and written back are automated or largely replaced.
I. Higher risk of elimination or large-scale replacement
1. Customer Service
Of all the functions, customer service is the first to feel the impact of AI. AI is remaking how a company interacts with its customers, and is working its way into every detail of the job. The traditional phone-centered contact center is giving way to an omnichannel service platform — text chatbots, voice assistants, and support inside a mobile app, fused so that a customer gets a consistent, timely response whichever channel they use. The work is also moving from reacting after the fact to predicting: by analyzing a customer’s tone and emotion in the moment, a company can spot dissatisfaction before a problem breaks out, step in earlier, and offer more personal care. Agents are no longer only handling a single question; with AI tools, they can help manage a customer’s journey across channels, and keep each touchpoint coherent.
With that change, AI also gives service teams a real enhancement. Advanced language models let a chatbot hold a multi-turn conversation, pull an answer from a company knowledge base, and resolve common questions at once. Progress in NLP also helps a system see a customer’s actual intent and shape a response to it. If AI detects a drop in mood or an agitated tone, for example, it can route the person into a more careful flow, or notify a human to take over at once. Assistive tools can also generate a conversation summary, a service record, and a follow-up reminder after each interaction, and take a large share of the recording burden off a person. Through human–AI collaboration, a service team can put more of its attention on harder cases and on building a longer-term relationship.
As AI takes over a large volume of basic, repetitive service work, some traditional roles are at risk of being eliminated. Standard Q&A agents (junior staff who only read from an FAQ script) will be needed in much smaller numbers, because simple questions are now handled quickly and accurately by automated chatbots. The old process of transcribing every call and compiling a history by hand also becomes unnecessary once an AI system logs and analyzes the interaction. Even outsourced service teams that depend on large headcount are under pressure, because an AI solution can handle a large volume of basic inquiries at lower cost and with more consistency. For a company, this means the role of a service person has to be reset — no longer a “switchboard operator” who repeats answers, but a customer-relationship specialist who can use AI tools and concentrate on higher-level communication and an emotional connection.
2. Marketing
Marketing is also going through a sharp change under AI. Teams used to move brand visibility forward in a sequence of campaign cycles. The work is now moving toward an always-on, personalized model. A company no longer relies only on a single big idea to move the market; it uses semantic analysis and data insight to keep adjusting the strategy, and to offer more precise content to different groups at any moment. The center of gravity is also shifting: from producing a large volume of assets for its own sake, to making sure the message actually fits the audience’s needs and emotional resonance — that is, around communicative intent, not merely around output volume.
With AI, marketing also gains a strong set of enhancement tools. Generative AI acts as a creative assistant: it can produce a key visual, a script, product-page copy, or even a short video in one pass, so a marketer can generate several options in a short time and run A/B tests to find what works. Companies are beginning to build a brand semantic graph, so that AI can judge when, through which channel, and to whom a given piece of content should go — a more intelligent form of distribution and monitoring. At the same time, real-time social listening and sentiment analysis let a team see changes in preference and in the language people are using — AI can flag a topic or a reputation trend as it is forming, so that brand communication stays closer to the current pulse.
As basic production and analysis work is taken by AI, some traditional marketing duties face elimination. Demand for junior content producers (simple copy, retouching, basic video editing) falls sharply, because AI tools can already do a large volume of standardized creative work. The old habit of judging a trend by instinct, and betting on a hit from experience, is also fading — a strategist who cannot use data and AI insight will find themselves behind. Work that consists of compiling social data by hand, tagging posts, and generating reports is also being pushed to the edge by automated dashboards and AI analysis. Future marketing talent will put their weight on what AI cannot replace: integrating the insight behind mixed data, building a brand story that has both originality and human judgment, and holding the brand’s values and the strategic direction.
3. Finance
Finance is likewise being deeply affected, and the function’s position is changing. Finance people used to be centered on recording results after the fact and producing reports. AI is moving the work from recording an outcome to forecasting a process and warning of anomalies. A company no longer has to wait for a monthly or quarterly close to understand its financial position; it can have a live financial dashboard: an AI system can assemble revenue, cost, cash flow, and other key indicators, so that management can see the company’s health at any time. If an unusual trend appears (a spike in spending, or an abnormal drop in revenue), AI can raise an alert in time. Building a financial model by hand in Excel is also being displaced. An analyst can now ask a question in natural language and have AI run a scenario and a calculation at once. From a question such as “what a 10% drop in next quarter’s volume would do to profit and loss,” to a more complex analysis of capital expenditure and return, AI can produce a forecast quickly and help the team decide with more foresight.
These changes give finance a real enhancement. AI can assist with cash-flow forecasts, profit estimates, and simulations of a hedging plan. A machine-learning model can, from historical data and market movement, forecast cash over the next few quarters and suggest a corresponding strategy; or, when markets move, automatically assess how different hedges would affect the company’s finances. Tedious document work is also lightened by automation: AI can parse invoices, receipts, and bills, reconcile them, and produce reports and filings that meet tax rules. More important, timely and precise analysis lets financial decisions sit closer to the business strategy. Finance is no longer only a support unit that delivers numbers after the fact; it can offer insight of its own and take part in a strategic discussion as a core partner.
As AI takes over a large volume of repetitive calculation and record-keeping, some traditional finance roles are gradually being eliminated or redefined. Demand is falling for junior bookkeepers and cashiers who used to sit with the ledgers (manual bookkeeping, sorting vouchers), because automated bookkeeping software and AI image recognition can finish that work quickly and without error. The finance assistant whose job is to compile static reports and draw charts is also becoming rarer; a manager can now query a dashboard for what they need, without waiting for a manual compilation. Even traditional audit-support roles (jobs that consist of a large volume of mechanical checking) face reduction, because AI can watch transactions for anomalies over a long period, compare changes in the moment, and check that rules are being met. Facing this, finance professionals in a new era need to turn toward a more strategic orientation: reading the insight AI produces, assessing the financial effect of a decision, and working closely with other functions so that the company can grow in a healthy way on the back of data.
5. Data & Analytics
Driven by AI, a company’s data and analytics function is also being remade. Analysts used to have to write SQL or a script themselves to produce a report. That model is quickly being replaced by “a semantic question + automatic insight.” A business user can ask the system a question in natural language; AI can then extract the relevant information from a large volume of data and produce an insight, without an engineer writing code step by step to dig out the answer. The center of gravity is also shifting: from “collect and integrate the data” to building a closed loop of “data → model → intent” — making sure that what is collected can enter a machine-learning model in time, and serve a decision-maker’s specific intent (answering a business question, or forecasting an indicator). The way the data team is organized is also becoming more embedded: a company may once have had a central data team serving the whole firm; the trend now is to place analytics talent inside each business unit, as partners who understand both the business and AI, and can support a front-line need in the moment.
With these changes, AI gives data and analytics a strong enhancement: a large volume of basic work that used to consume an analyst’s time is being automated. AI tools can, for example, clean and integrate messy data from different systems automatically, and find outliers or a key trend, so an analyst no longer has to put so much effort into an ETL (extract, transform, load) flow. AI can also fuse and analyze multimodal data — text, voice, image — and mine insight from unstructured sources. It can, for instance, analyze a service transcript (text), a customer call recording (voice), and a product photograph (image) at the same time, and give the company a more complete view. Senior executives benefit as well — they can now ask a dashboard a complex business question in natural language, and AI will produce a visual answer and a suggestion at once, shortening the wait before a decision.
Under this pressure, many traditional, junior data-analysis tasks are being eliminated or redefined. The analyst-assistant role that only drew charts is fading, because self-service BI tools and AI reporting systems can generate a chart for anyone to look up themselves. Demand is also falling sharply for people whose only job is to assemble numbers on a schedule and submit a report — weekly and monthly packs of this kind can often be completed and sent by an AI platform on its own. People who repeatedly write similar analysis reports face pressure to change: if the analysis is only a template applied to data, a machine can replace it entirely. That does not mean data talent has no place; it raises their position. Future data professionals will put more time into higher-value work: designing a better analytical question, checking the quality of the data and the reliability of a model, and explaining what AI has produced and turning it into a concrete business recommendation.
15. Sales
The AI wave is also reshaping sales. Traditional selling was often treated as an art of relationships: a representative won a deal on trust built over time, and on face-to-face visits. In an AI era, the sales process is moving from a heavy reliance on personal relationships toward a better use of data and automation. AI can now, from a large list, automatically screen the prospects most likely to close (lead scoring), and finish the first filter and the first contact more efficiently, so that a salesperson can concentrate on the most valuable accounts and on the closing skill that still has to be human. Sales activity is also moving from mainly offline visits to a mix of online and offline: teams increasingly rely on digital tools to track a customer’s online behavior, and interact through email, social media, and content, building a continuous content loop that guides a decision. Salespeople used to record customer information in a notebook or a spreadsheet; an AI-driven CRM can now assemble touchpoint information automatically, forecast a conversion rate, and remind the rep when to follow up, and with what, so that selling becomes less purely experience-driven and more precise.
With AI tools, a sales team gains a rich set of enhancements for efficiency and customization. First, an intelligent agent can act as a sales assistant, mining the best prospect list from a large market database, and suggesting a personalized approach for each lead (which product value to emphasize, which case study is most likely to resonate). Second, AI can automatically generate sales materials: a proposal deck tailored to a customer’s needs, a quotation, a follow-up email — all of these can have a first draft produced quickly, for the representative to adjust. After a sales meeting, AI can also, from a transcript, generate minutes and a summary of the main points at once, identify the questions the customer cared about and the follow-ups that were promised, and write that information back into the CRM. In that way, a rep does not have to spend long hours on administration, and can put more attention on the relationship and on negotiation strategy.
As the sales process becomes more intelligent, some traditional sales-support and low-value activities are being eliminated. First, the old “spray and pray” cold call is fading — precise marketing can warm a lead in advance, and a rep can then follow up in a more targeted way, which is more efficient than an undifferentiated call; telemarketers who live by calling strangers will have to change. Second, support roles that exist only to polish a bid deck or make a proposal in PowerPoint will be reduced, because AI can quickly produce a professional proposal template and greatly lower the barrier to that kind of purely documentary work. Likewise, demand falls for assistants who type data into a CRM and update a progress report by hand; information gathered from customer interactions can now flow into the system automatically, and generate the analysis a manager needs. Facing this, salespeople will concentrate more on the parts of human interaction that cannot be removed: building trust, understanding a customer’s particular pain and need, and showing insight and persuasion at the table. By working with AI on the chores, the strongest sellers will play more of an advisor’s role, and offer a value proposition built for that customer — a genuine two-sided gain.
II. Accelerate: functions whose cycle time is being compressed
4. Operations
Once AI is involved, operations is going through a shift from experience-driven to data-driven work. Production scheduling and resource allocation used to rely mainly on a senior manager’s judgment from experience; they are gradually moving toward optimization by live data and intelligent algorithms. The old linear, single-path model of process control is being replaced by a more flexible, complex network of coordination — each step is tightly linked through a digital platform, information moves quickly, and an operations decision is no longer a single line, but a network that can be adjusted as it runs. Operations is also moving from “react after the fact” to “predict in advance”: a company can use an AI model to forecast a swing in demand, a maintenance window, or a potential bottleneck, adjust early, and deal with a problem while it is still small. That kind of predictive operations makes a company more resilient, and reduces the loss that comes from being caught unprepared.
AI gives operations teams many kinds of enhancement, and raises both execution efficiency and precision. First, in scheduling, an AI system can adjust a production plan automatically from live data: dynamically ordering the work, reallocating inventory and logistics routes in the moment, even setting a staffing roster from a forecast of peaks and troughs in demand. Second, in monitoring, AI can collect signals from several systems and sensors at once, quickly identify an anomaly (an early sign of a line fault, or a delay risk in a supply-chain step), and raise an alert in time. Further still, some AI-driven operations platforms can “self-optimize” — through a continuous loop of automated tests and feedback, they find a better process configuration. A system can, for example, try different scheduling strategies in parallel, adjust from the result, and put operations decisions into a fast cycle of trial and iteration.
In that setting, many traditional, basic operations roles are gradually being replaced or reduced by AI systems — the eliminate part of the story. Operators who only execute a fixed process from a form are seeing their value overtaken by automation — when an SOP can be executed precisely by a machine, people can be released for more valuable work. Likewise, roles such as data transcribers and order chasers have been fading as data systems are integrated and RPA (robotic process automation) tools spread. Even some coordinators who judged risk and scheduling purely from experience become less indispensable once an AI risk model is there to help. For a modern operations manager, the center of gravity will move from personally “fighting fires” and watching a process, to designing and supervising an AI-driven operations system, tuning algorithmic parameters, and continually improving the process under human–AI collaboration. In other words, people will put more of their effort into strategic questions: how to set a better metric, how to make sure AI is reading the right business intent, and how to handle the special cases AI cannot understand.
8. Information Technology (IT)
A company’s IT function is also being remade in its core work and its position. IT people used to be seen as maintainers of the architecture, concentrated on keeping systems stable and handling day-to-day technical problems. They are increasingly becoming integrators of an AI ecosystem. That is, an IT team not only has to manage the traditional hardware and software estate; it also has to bring various AI solutions (from cloud services to machine-learning platforms) into the company’s technical map without friction. The pattern of software development is changing too: engineers are moving from writing most of the code by hand toward playing more of a “prompt engineer” role, assembling the needed function from existing AI models and modules. The center of IT’s value is also shifting, gradually, from back-end support to front-end enablement — by building internal platforms and tools that package data and AI capability into an interface ordinary employees can use, so the whole organization can benefit from a technical change without having to depend on IT for every operation.
AI gives IT professionals a rich set of enhancements, and raises the efficiency and quality of technical service. First, automated testing and deployment tools become more intelligent with AI’s help: a system can generate test cases, run software tests, and quickly locate and fix common errors (some tools can even propose a patch from an error log). CI/CD (continuous integration / continuous deployment) reaches a higher degree of automation, and a new release can go live faster and more reliably. At the same time, large language models are taking code generation and debugging into a new range — an engineer can describe a need through a conversational interface, and AI can produce a corresponding code example or find a problem in existing code. For internal users, IT is also introducing AI-driven knowledge bases and chat assistants, gradually replacing the traditional helpdesk: an employee can ask an internal chatbot how to set up email, connect a printer, or reset a password, and get guidance, which substantially reduces the burden on first-line technical support.
With these changes, some traditional IT roles face elimination or a reshuffle. Highly repetitive development work is being replaced by automation — for example, demand falls for programmers who only build a simple front-end page or a basic application function, because a low-code platform and AI-generated code can finish that work faster. Labor-heavy support roles (installing software machine by machine, handling password-reset requests by hand) also shrink as automated scripts and AI agents come in. For a purely technical person who only knows a traditional stack and has no ability to integrate AI, competitiveness will fall noticeably — what companies will need is hybrid talent that understands software development and can also use AI tools with some flexibility. IT practitioners will therefore have to upgrade their skill mix if they are to keep creating value: acting as a bridge between AI systems and the existing infrastructure, assessing the strategy for bringing in a new technology, and making sure people can work safely and efficiently with these intelligent tools.
9. Procurement & Supply Chain
Once AI is involved, the center of gravity in procurement and supply-chain management is gradually moving toward higher-level strategy and risk forecasting. A procurement manager used to be seen mainly as a keeper of supplier relationships, controlling cost through a regular buying plan and a negotiation. The role is now closer to a strategy optimizer and a cost forecaster. The traditional cyclical buying model is giving way to immediate, dynamic, multivariable scheduling — AI can adjust a procurement plan at any time from sales trends, inventory levels, and supply and demand, instead of placing a fixed order by quarter or by year. Buyers used to sign a fixed contract, agree a price for the year, and rest easy; in a more volatile market, AI helps a company move toward a dynamic contract: terms can adjust automatically from the market or from performance, and AI can even generate contract text in the moment and suggest a revision, so that an agreement is more flexible and reflects the current situation.
AI gives procurement and supply-chain teams a new kind of enhancement, and puts decisions on a more complete information base. First, AI tools can track raw-material prices, supplier inventory, geopolitical risk, logistics, and even carbon indicators around the clock, and gather that scattered data into a live dashboard. A procurement manager can see a price trend and a supply-risk warning at a glance, and prepare earlier. Second, multi-source price comparison and risk scoring become far more efficient — an AI platform can automatically collect quotations, lead times, and quality records from several suppliers, and, from weights the company has set (price, reliability, sustainability, and the like), calculate a composite score to help the team choose a supply strategy more objectively. An intelligent agent can even help draft a procurement contract, filling in key details from past clauses and the company’s preferences; and if an anomaly appears in execution (a delayed shipment, a quality issue), the system can send an alert at once so that the right people can step in.
As procurement becomes more digital and more intelligent, some traditional procurement and supply-chain roles are being eliminated or repositioned. Ways of working that depend on calling for a quote and updating a buying list by hand in Excel are being replaced by supply-chain systems and e-procurement platforms — AI can take the latest price from an online platform and update the system, without manual work. Demand falls for basic negotiation-support roles (assistants who only prepare tender documents or simple bargaining materials), because much of that preparation can be finished quickly by AI. People who managed an RFP process by hand also face pressure; from writing a tender notice, to screening a supplier’s qualifications, to scoring a proposal, there are now intelligent tools that can help automate the work. In future, procurement and supply-chain professionals will concentrate more on what AI cannot replace: building a strategic partnership with a supplier, seeing risk in the round and setting a backup plan, and, from the information AI provides, making the final judgment and the decision in a negotiation.
The judgment, restated
My judgment is that a high-risk function is not a whole job disappearing. It is standardized tasks being pulled out. The people who remain have to change what they manage: exceptions, intent, and experience.
Three things this week
- Pick the three tasks this week that are most easily automated, and write them as an SOP plus an acceptance check.
- Hand one of them to a tool or an agent; people only handle escalation and exceptions.
- Reset the SLA with customers, or with internal customers: is the speed buying a better experience, or a complaint?
Return to the series overview and the hub directory, or go to Part C for the functions being enhanced and the manager’s handbook. Subscribe to Future Insights Lab.