[{"data":1,"prerenderedAt":385},["ShallowReactive",2],{"post-\u002Fai-transformation-redesign-work-not-cut-roles":3},{"page":4,"translations":240,"nav":248,"related":368,"random":376},{"id":5,"title":6,"body":7,"categories":208,"category":211,"changeHistory":211,"date":212,"description":213,"disclosures":214,"draft":217,"extension":218,"firstLiveAt":211,"image":219,"imageAlt":220,"kind":221,"lang":222,"meta":223,"navigation":224,"omitGermanLocalizationDisclosure":217,"path":225,"publishedAt":211,"readingTime":226,"rights":211,"seo":227,"seoTitle":228,"slug":229,"sources":211,"stem":229,"tags":230,"translationKey":237,"type":238,"updated":211,"__hash__":239},"posts\u002Fai-transformation-redesign-work-not-cut-roles.md","AI Transformation Starts with Redesigning Work",{"type":8,"value":9,"toc":196},"minimark",[10,14,17,22,25,28,31,34,38,51,54,57,60,64,72,75,82,85,89,92,95,98,101,109,113,116,119,122,125,129,132,135,138,141,145,148,151,154,157,163,170,174,177,180,187,190],[11,12,13],"p",{},"Imagine a customer-support team starting to use AI to prepare replies. First-response times fall. Management calculates that a smaller team could handle the same volume of tickets. But are customers actually getting their problems resolved faster? Or do the quick replies create more follow-up messages, reopened tickets, and handoffs to other teams?",[11,15,16],{},"AI may have accelerated one task in that example. We cannot yet say the same about the work as a whole. To assess a transformation claim, we need to look at the process from preparing a reply through to resolving the customer’s problem.",[18,19,21],"h2",{"id":20},"a-headcount-decision-does-not-explain-how-work-has-changed","A headcount decision does not explain how work has changed",[11,23,24],{},"Reducing employee numbers while adopting AI does not, by itself, show that a company has redesigned its work. It may simply be trying to run the same process with fewer people. The difference becomes visible in the tasks that change, the steps that disappear, and the way responsibility is distributed.",[11,26,27],{},"AI can, of course, reduce the need for certain tasks. That can change workload, cost structure, and staffing needs. But setting a headcount target first and then asking technology to absorb the remaining work turns an untested assumption into an operating plan.",[11,29,30],{},"Sometimes that assumption will hold. Sometimes the burden of control, exception handling, and correction is simply transferred to the people who remain. The organisation chart gets smaller while the labour required by the work does not shrink to the same degree.",[11,32,33],{},"The first question should therefore be not how many people can be removed, but why the work is being done and how it could produce a better result.",[18,35,37],{"id":36},"an-ai-did-it-claim-needs-evidence","An “AI did it” claim needs evidence",[11,39,40,41,50],{},"In their Harvard Business Review article, ",[42,43,49],"a",{"href":44,"rel":45,"target":48},"https:\u002F\u002Fhbr.org\u002F2026\u002F08\u002Fai-transformation-requires-redesigning-work-not-cutting-roles",[46,47],"nofollow","noopener","_blank","“AI Transformation Requires Redesigning Work, Not Cutting Roles”",", Faisal Hoque, Tom Davenport, and Paul Scade argue that early layoffs attributed to AI may not deliver the expected returns. They also note that some announcements may be a form of “AI-washing”: decisions made for other reasons presented through an AI narrative.",[11,52,53],{},"That does not mean every workforce reduction is wrong or that every company is concealing its reasons. An announcement alone is not enough to establish cause and effect. Falling demand, earlier over-hiring, or broader cost pressure may also shape the decision.",[11,55,56],{},"The business question I take from the article is this: where is the operational change that demonstrates AI’s contribution?",[11,58,59],{},"If a company cannot show which step in the work changed, how many resources it consumed before the change, and what happened afterward, it has a management assumption, not evidence of a completed transformation. An expectation can inform an investment decision. It should not be presented as an achieved gain.",[18,61,63],{"id":62},"meta-smaller-teams-do-not-automatically-mean-redesigned-work","Meta: smaller teams do not automatically mean redesigned work",[11,65,66,71],{},[42,67,70],{"href":68,"rel":69,"target":48},"https:\u002F\u002Fblog.pragmaticengineer.com\u002Fthe-pulse-meta-wanted-to-reduce-teams-by-60-because-of-ai\u002F",[46,47],"Reuters reporting, summarized by The Pragmatic Engineer",", described Project OT scenarios in which some Meta teams could be reduced by up to 60%; Meta confirmed that planning scenarios applied to some teams. This was not a 60% company-wide workforce cut. The first May layoff wave happened, while the planned second wave was cancelled.",[11,73,74],{},"Shrinking a team of 10 to 20 people to three to five people with AI support does not, by itself, redesign work. Domain knowledge, redundancy, on-call capacity, judgment, and the arrangements for intervention and accountability also have to be redesigned. AI transformation should be assessed through changed tasks, preserved controls, and measured customer or product outcomes, not headcount alone.",[11,76,77],{},[78,79],"img",{"alt":80,"src":81},"A smaller team still carries a tangled burden of control and exceptions, while a redesigned flow separates tasks and human intervention clearly","\u002Fimages\u002Finline-ai-work-redesign\u002Fsmaller-team-is-not-redesign.webp",[11,83,84],{},"It also leaves unanswered whether the remaining team can sustain the same demand and which decisions remain with people.",[18,86,88],{"id":87},"move-from-job-titles-to-tasks","Move from job titles to tasks",[11,90,91],{},"Redesigning work means looking at what people actually do inside a role. Labels such as “sales operations” or “customer support” are too broad to support an automation decision. The same role may involve gathering information, drafting text, verifying it, making decisions, and reaching agreement with other people.",[11,93,94],{},"Consider a proposal process. Understanding the customer’s need, reviewing earlier contracts, preparing a draft, approving the price, and committing to delivery are different kinds of work. Making the draft faster does not show that pricing authority or delivery responsibility can also be handed to the same system.",[11,96,97],{},"Looking at tasks creates another option: some steps may be removed entirely rather than automated. There is little business value in producing an unused report more quickly.",[11,99,100],{},"The purpose is not to produce a detailed task inventory and stop there. It is to see which outcome each step serves and decide where technology can improve that outcome.",[11,102,103,104,108],{},"I explored how this change in tasks affects the skills expected of employees and the design of their roles in ",[42,105,107],{"href":106},"\u002Fwhen-ai-changes-work-training-employees-is-not-enough","When AI Changes Work, Training Employees Is Not Enough",".",[18,110,112],{"id":111},"decision-rights-and-accountability-must-remain-clear","Decision rights and accountability must remain clear",[11,114,115],{},"When a system prepares a recommendation, it matters who approves it. When the system acts directly, its boundaries need to be clearer still.",[11,117,118],{},"In a hypothetical returns process, AI could classify a request, ask for missing information, and recommend the appropriate action. But the company still has to decide under which conditions a customer receives a refund, which exceptions require further review, and who is accountable if the action is wrong.",[11,120,121],{},"“A human makes the final decision” is not a sufficient design by itself. Does that person have time to review the decision, the information needed to do so, and the authority to stop the action? As the number of approved transactions rises, is the control still meaningful?",[11,123,124],{},"An arrangement in which an employee merely approves the system’s output may include human oversight on paper. Real accountability requires the accountable person to be able to intervene in the process. That is a question of capacity and authority as much as a job description.",[18,126,128],{"id":127},"measure-quality-and-rework-alongside-speed","Measure quality and rework alongside speed",[11,130,131],{},"Producing an initial output faster does not mean the total amount of work has fallen. Checking a draft, correcting false information, or explaining the matter to a customer again are all part of the same cost.",[11,133,134],{},"Return to the support example. If first-response time falls while the rate of reopened tickets rises, part of the speed gain may be disappearing in later stages. Looking only at the number of replies sent would create a misleading picture of success.",[11,136,137],{},"Not every error should be treated as equal either. Correcting a small wording issue has different consequences from making an incorrect commitment to a customer. Controls should reflect both the likelihood of an error and its impact.",[11,139,140],{},"Customer value is not the company producing more output. It is the customer’s problem being solved correctly, a promise being kept, or a service becoming easier to access. When an internal efficiency measure cannot be connected to those outcomes, the process can look better on an internal dashboard while the customer experience gets worse.",[18,142,144],{"id":143},"how-should-transformation-be-measured","How should transformation be measured?",[11,146,147],{},"Measurement begins by making the state before the change visible. Without knowing the end-to-end time, total labour required, correction burden, and customer outcome, it is difficult to assess a gain reliably.",[11,149,150],{},"The cost of a new arrangement is not limited to the model or software. Integration, control, maintenance, and the work required to handle exceptions are part of the cost too. It is also worth tracking whether saved time is moved to other work. Released capacity has value; assuming it becomes a direct cash saving is a separate claim.",[11,152,153],{},"Where possible, the change can be tested in a limited scope. Comparisons should account for demand volume, the difficulty of the work, and seasonal effects. Results from simple requests do not show that the full operation will perform in the same way.",[11,155,156],{},"A useful evaluation reads several measures together: Is the work completed faster? How does total cost change? What happens to errors and correction work? What difference does the customer see? When those answers conflict, the contradiction should lead to a redesign rather than be hidden.",[11,158,159],{},[78,160],{"alt":161,"src":162},"A manager compares elapsed time, human effort, rework, and customer outcome across the before-and-after workflow","\u002Fimages\u002Finline-ai-work-redesign\u002Fmeasure-outcomes-not-speed.webp",[11,164,165,166,108],{},"For a four-layer distinction that keeps model output, completed work, and business impact from being confused with one another, see ",[42,167,169],{"href":168},"\u002Fis-your-ai-system-delivering-business-results","Is Your AI System Actually Delivering Business Results?",[18,171,173],{"id":172},"a-more-useful-question-to-begin-with","A more useful question to begin with",[11,175,176],{},"Redesigning work will not always result in a smaller team. It may allow the same team to meet accumulated demand, expand the scope of service, or spend more time on work that needs greater care. Which result matters depends on the company’s actual need.",[11,178,179],{},"That is why it makes sense to choose one concrete piece of work and examine it end to end. Where does it wait? Where does it create errors? Which decisions remain uncertain? Which of those problems could AI address, and how would we know the solution worked?",[11,181,182,183],{},"This leads to a question worth taking into a management meeting: ",[184,185,186],"strong",{},"When we redesign this work, how will the result delivered to the customer improve, and what evidence will show that it has?",[11,188,189],{},"A staffing decision can then rest on firmer ground. The value of AI transformation appears first in the result the work delivers, not in the name of the tool or the smaller organisation chart.",[11,191,192],{},[193,194,195],"em",{},"Evidence boundary: The HBR article is the starting point for this discussion; the views presented here are not treated as causal evidence that applies to every company. The support, proposal, and returns examples are hypothetical and do not report a real customer experience or measured outcome.",{"title":197,"searchDepth":198,"depth":198,"links":199},"",2,[200,201,202,203,204,205,206,207],{"id":20,"depth":198,"text":21},{"id":36,"depth":198,"text":37},{"id":62,"depth":198,"text":63},{"id":87,"depth":198,"text":88},{"id":111,"depth":198,"text":112},{"id":127,"depth":198,"text":128},{"id":143,"depth":198,"text":144},{"id":172,"depth":198,"text":173},[209,210],"ai","business",null,"2026-09-19","AI transformation is not proven by a smaller headcount. It begins by redesigning tasks, decision rights, controls, and business outcomes.",{"aiUse":215,"aiNote":216},"ai-assisted","This article is based on Evren Bal’s editorial assessment. AI-assisted tools were used during translation and editorial development.",false,"md","\u002Fimages\u002Fhero\u002Fai-work-redesign-before-headcount-cuts.avif","Task cards move through AI-assisted drafting, human approval, quality inspection, and a customer outcome checkpoint","Analysis","en",{},true,"\u002Fai-transformation-redesign-work-not-cut-roles",8,{"title":6,"description":213},"AI Transformation: Redesign Work Before Cutting Roles","ai-transformation-redesign-work-not-cut-roles",[231,232,233,234,235,236],"artificial-intelligence","business-transformation","work-design","process-management","operational-efficiency","organizational-transformation","ai-work-redesign-before-headcount-cuts","post","Jl3c8KxgI0j4eWKbE3uygPjmpjzAvSlIjI7QsNvJUac",{"en":241,"tr":242,"de":245},{"path":225,"title":6},{"path":243,"title":244},"\u002Ftr\u002Fyapay-zeka-donusumu-isi-yeniden-tasarlamakla-baslar","Yapay Zekâ Dönüşümü, İşi Yeniden Tasarlamakla Başlar",{"path":246,"title":247},"\u002Fde\u002Fki-transformation-arbeit-neu-gestalten-statt-rollen-streichen","KI-Transformation beginnt mit der Neugestaltung von Arbeit",{"prev":249,"next":211,"others":252,"lucky":367,"readingTime":226},{"path":250,"title":251},"\u002Fai-will-spread-across-turkiyes-businesses","AI in Türkiye: What Adoption Figures Leave Out",[253,256,259,262,265,268,269,272,275,276,279,282,285,288,291,294,297,300,303,306,309,312,315,318,321,324,327,330,333,336,337,340,343,346,349,352,355,358,361,364],{"path":254,"title":255},"\u002Fai-turned-a-refactor-i-wouldnt-do-into-a-one-hour-job","With AI, a Refactor I Wouldn't Bother With Took One Hour",{"path":257,"title":258},"\u002Fbeyond-the-bot-lessons-from-building-a-chat-system-for-global-patients","What We Learned Building a Healthcare Chatbot for International Patients",{"path":260,"title":261},"\u002Feu-ai-act-after-risk-classification","EU AI Act: What Should You Do Once You Know the Risk Level?",{"path":263,"title":264},"\u002Ffrom-asking-questions-to-delegating-work","From Asking Questions to Delegating Work: What the Codex Study Shows",{"path":266,"title":267},"\u002Fchatbot-data-third-party-ai-provider","What Happens When Chatbot Data Is Sent to a Third-Party AI Provider?",{"path":250,"title":251},{"path":270,"title":271},"\u002Fis-buying-backlinks-worth-it","Would We Still Buy This Backlink If It Sent No Ranking Signal?",{"path":273,"title":274},"\u002Fai-value-is-not-always-more-work","AI's Value Is Not Always More Work",{"path":168,"title":169},{"path":277,"title":278},"\u002Fai-assisted-rest-api-development","Preserving API Quality in AI-Assisted Development",{"path":280,"title":281},"\u002Fhow-to-do-content-pruning-a-real-world-case-study","How to Do Content Pruning: A Real-World Case Study",{"path":283,"title":284},"\u002Fgoogle-generative-ai-data-ai-citation-timing","AI Visibility Dropped Before Search: The Google Data That Changed My Theory",{"path":286,"title":287},"\u002Fwho-captures-ai-productivity-gains","Who Captures the Productivity Gains from AI?",{"path":289,"title":290},"\u002Fwhat-ai-can-automate-in-cro-research","What Can AI Actually Automate in CRO Research?",{"path":292,"title":293},"\u002Fcio-cto-it-director-difference","CIO, CTO, and IT Director: What the Title Misses",{"path":295,"title":296},"\u002Fbringing-english-back-into-my-working-day","Bringing English Back Into My Working Day",{"path":298,"title":299},"\u002Fstart-with-the-business-problem-not-the-ai-model","Start With the Business Problem, Not the AI Model",{"path":301,"title":302},"\u002Fwhat-makes-an-ai-system-specific-to-your-business","What Makes an AI System Specific to Your Business?",{"path":304,"title":305},"\u002Fai-visibility-illusion-bing-citation-share","The AI Visibility Illusion: What Bing's Citation Share Data Actually Reveals",{"path":307,"title":308},"\u002Fwhen-is-one-ai-model-enough","When Is One AI Model Enough?",{"path":310,"title":311},"\u002Fwhen-does-enterprise-ai-need-rag","When Does Enterprise AI Actually Need RAG?",{"path":313,"title":314},"\u002Fwhen-process-automation-actually-needs-ai","When Does Process Automation Actually Need AI?",{"path":316,"title":317},"\u002Fwhen-do-you-actually-need-an-ai-agent","When Do You Actually Need an AI Agent?",{"path":319,"title":320},"\u002Fhow-can-companies-create-value-from-ai","How Can Companies Create Value From AI?",{"path":322,"title":323},"\u002Fwhere-should-people-be-in-the-loop-while-ai-does-the-work","Where Should People Be in the Loop While AI Does the Work?",{"path":325,"title":326},"\u002Fbuild-in-public-2-0","Build in Public in the AI Era: What to Share and What to Keep Private",{"path":328,"title":329},"\u002Fdoes-enterprise-ai-really-need-fine-tuning","Does Enterprise AI Really Need Fine-Tuning?",{"path":331,"title":332},"\u002Fhow-ai-changes-the-experience-gap","Can a Junior Who Uses AI Well Outperform a Senior Expert?",{"path":334,"title":335},"\u002Faccessing-know-how-is-not-the-same-as-creating-it","Accessing Know-How Is Not the Same as Creating It",{"path":106,"title":107},{"path":338,"title":339},"\u002Fdo-ai-visibility-tools-really-work","Do AI Visibility Tools Really Work? What They Actually Measure",{"path":341,"title":342},"\u002Fproductlog-the-platform-i-built-for-myself-first","ProductLog: The Platform I Built for Myself First",{"path":344,"title":345},"\u002Fwhy-im-building-rankextension-making-google-search-console-actually-make-sense","What Happened to RankExtension?",{"path":347,"title":348},"\u002Fdont-write-to-be-recommended-by-ai","Getting Your Brand Recommended by AI Is the Wrong Goal. What Is the Right One?",{"path":350,"title":351},"\u002Ftesting-a-button-treating-an-entire-website-redesign-as-a-sure-thing","Testing a Button, Treating an Entire Website Redesign as a Sure Thing",{"path":353,"title":354},"\u002Fwhy-we-trust-ai-judgments","Why We Trust AI Judgments and What That Fails to Prove",{"path":356,"title":357},"\u002Fredar-ai-powered-summaries-for-kap-disclosures-and-open-sources","Redar: AI-Powered Summaries for KAP Disclosures and Open Sources",{"path":359,"title":360},"\u002Fprotecting-organizational-memory-during-ai-transformation","Protect Organizational Memory During an AI Transformation",{"path":362,"title":363},"\u002Fis-your-data-ready-for-ai","Is Your Data Ready for AI? Start With the Decision It Must Support",{"path":365,"title":366},"\u002Fturkeys-first-real-time-mystery-shopping-reporting","Turkey's First Real-Time Mystery Shopping Reporting Platform",{"path":307,"title":308},[369,371,373,375],{"path":359,"title":360,"date":370},"2026-09-11",{"path":106,"title":107,"date":372},"2026-09-08",{"path":322,"title":323,"date":374},"2026-09-13",{"path":331,"title":332,"date":370},[377,379,383],{"path":341,"title":342,"date":378},"2026-06-23",{"path":380,"title":381,"date":382},"\u002Fwhy-the-same-ai-model-produces-different-results","Why the Same AI Model Produces Different Results Across Applications","2026-09-12",{"path":319,"title":320,"date":384},"2026-09-04",1789771929299]