Fourteen business functions, each researched three ways with function-wide sources: what AI can cut today, the eventual ceiling once every foreseeable lever is used, and how that ceiling was judged in 2023, the first post-ChatGPT year. Every figure is scoped to a whole function, not a single task.
Median low–high across each cell's function-wide and broad-scope sources (sub-task-only figures excluded from the median). Read the columns side by side to see the movement; the full evidence, with every report and link, is in the three tables below.
| Business function | Today | Eventual (2023) | Eventual (now) |
|---|---|---|---|
| 1. Production / Manufacturing | 18–28% | 30–44% | 50–60% |
| 2. Logistics, warehousing & distribution | 5–20% | 8–16% | 25–42% |
| 3. Field service, installation & maintenance | 12–30% | 26–34% | 32–48% |
| 4. Accounting operations | 12–20% | 25–40% | 35–50% |
| 5. Purchasing operations | 19–40% | 25–40% | 35–40% |
| 6. HR & workforce administration | 19–30% | 30–42% | 45–55% |
| 7. Order processing & back-office operations | 15–20% | 25–35% | 52–70% |
| 8. R&D, engineering & product design | 15–40% | 20–37% | 18–48% |
| 9. IT & software development | 10–22% | 15–35% | 45–50% |
| 10. Marketing & brand | 5–20% | 8–22% | 20–30% |
| 11. Legal, compliance & corporate strategy | 12–22% | 20–40% | 52–66% |
| 12. Sales & business development | 15–20% | 22–36% | 48–62% |
| 13. Customer service & success | 20–35% | 30–45% | 60–65% |
| 14. Executive leadership & general management | 15–30% | 25–40% | 30–50% |
Each row is one research report. The grey scope pill flags whether a finding covers the whole function (function-wide, broad) or only a sub-task (partial); partial figures are left out of the consensus. Hover a range for its basis. Confidence: High named report, explicit figure · Med converted from productivity/time · Low vendor or inferred.
Cost reduction from AI already deployable now, for the whole function
| Source (research report) | Yr | Major AI lever | Cost-reduction range | Conf. |
|---|---|---|---|---|
| 1 Production / Manufacturing | ||||
| World Economic Forum — Global Lighthouse Network 2025 (12 new sites) | 2025 | Full-factory AI transformation (computer-vision QC, predictive maintenance, digital twins, 30-40+ AI use cases per site) | 30-35% of manufacturing cost at leading sites broad | Med |
| McKinsey — AI/automation in operations (State of AI 2025 / operations insights) | 2025 | AI and automation across the production line (yield, energy, throughput optimization) | 20-30% of operational cost broad | Med |
| BCG — Executive Perspectives: Unlocking the Value Potential of AI in Manufacturing (June 2025) | 2025 | End-to-end AI across industrial operations (generative/agentic AI plus analytics) | ~15-25% of function operating/labor cost (converted from productivity) broad | Med |
| Accenture — Industrial AI / Physical AI Orchestrator (2025) | 2025 | Robotics at scale plus agentic AI for asset management and predictive maintenance | ~15-20% of production cost (robotics); 5-10% labor cost (predictive maintenance) broad | Med |
| Deloitte — Predictive technologies for asset maintenance (Industry 4.0 insights) | 2025 | AI/IoT predictive maintenance | 25-40% of maintenance cost partial | High |
| 2 Logistics, warehousing & distribution | ||||
| Oliver Wyman — How Logistics Operators Harness AI To Boost Efficiency | 2025 | AI across the operating pools — SG&A automation, last-mile and route optimization, sorting, warehouse management, plus service chatbots | 10-25% of the function's addressable operating cost (≈1-2pp EBIT uplift on 3-5% margins) function-wide | High |
| McKinsey — AI is transforming distribution supply chains, turning margin pressure into a competitive advantage | 2026 | AI-rewired supply chain: demand/inventory planning, network optimization, warehousing and transportation, frontline scheduling | 5-20% of logistics cost; ~20% of network cost for full AI rewiring broad | High |
| BCG — AI Is Already Moving the Logistics Industry Forward | 2026 | Agentic AI embedded into core daily logistics workflows (operations, pricing, planning) | AI leaders ~3× peers' cost reduction; ~2pp EBITDA uplift (≈5-15% of function operating cost) broad | Med |
| BCG — GenAI in Procurement: From Buzz to Bottom-Line Cost Reductions | 2025 | GenAI for sourcing/procurement: price-outlier detection, tender and contract generation, workflow automation | 15-45% of procurement cost (single sub-function, not the whole logistics function) partial | Med |
| 3 Field service, installation & maintenance | ||||
| BCG (Boston Consulting Group) — AI and the Next Frontier of Field Service | 2025 | Field Service AI (smart dispatching/scheduling plus in-field troubleshooting co-pilots and remote support) | 10-25% of the field service function cost function-wide | High |
| McKinsey & Company — Rewiring maintenance with gen AI | 2025 | Gen AI maintenance copilot for technicians/operators (root-cause diagnosis, knowledge retention) | 20-33% of maintenance labor cost broad | High |
| McKinsey & Company — From pilot to profit / From exploration to impact: AI in aftermarket, field services, and customer care | 2025 | Gen AI across aftermarket and field service (digital scheduling, technician support, automated interactions) | 15-30% of the function operating cost broad | Med |
| BCG — Executive Perspectives: The Future of Field Service with AI (EP13) | 2025 | AI-driven scheduling/dispatch plus troubleshooting co-pilots layered on continuous ops improvement | 20-30% of field productivity/cost broad | Med |
| 4 Accounting operations | ||||
| BCG Center for CFO Excellence — "How Finance Leaders Can Get ROI from AI" (survey of 280+ finance executives experienced with AI/GenAI) | 2025 | Generative-AI copilots, document AI and automation across the finance function | ~5–10% typical net (median 10% ROI, well below the ~20% targeted) function-wide | Med |
| McKinsey — "How finance teams are putting AI to work today" | 2025 | Generative-AI copilots and automation on data-heavy finance work (reporting, analysis, close support) | 20–30% less time on data-crunching activities (not total finance cost) broad | Med |
| Bain & Company — global AI cost-savings survey (951 companies) | 2026 | AI/GenAI and automation across corporate functions (incl. finance/back office) | ≤10% realized for most firms (40% land at ≤10%; only 10% reach 21–30%) broad | Med |
| Redwood Software — Enterprise Automation Index 2025 (285 practitioners across industries incl. finance) | 2025 | AI-driven business-process automation (RPA + document AI) | ~25%+ for the minority that achieve it; most below that partial | Low |
| The Hackett Group — Digital World Class Finance 2025 (Gen-AI finance benchmark) | 2025 | Full gen-AI + automation stack (agentic finance, ~80% AP and ~99% journal-entry automation) | 45% lower cost — LEADER CEILING, not typical today function-wide | High |
| 5 Purchasing operations | ||||
| McKinsey & Company — Redefining procurement performance in the era of agentic AI | 2025 | End-to-end agentic AI (autonomous sourcing, spend analytics, supplier management) | 25-40% of the function's operating cost/effort function-wide | High |
| BCG — GenAI in Procurement: From Buzz to Bottom-Line Cost Reductions | 2026 | Generative AI across key procurement processes (sourcing, contracting, S2P) | 15-45% of overall procurement cost function-wide | Med |
| The Hackett Group — Digital World Class Procurement (2.6X ROI / 58% faster) | 2025 | Intelligence-driven operating model with Gen AI in sourcing, negotiation, contract management | ~19% lower cost as a share of spend (with 31% fewer FTEs) broad | Med |
| The Hackett Group — AI in Purchase-to-Pay (Digital World Class benchmark) | 2025 | AI/hyperautomation in the purchase-to-pay (P2P) transactional cycle | up to 74-81% lower P2P process cost/staffing (sub-function only) partial | Med |
| 6 HR & workforce administration | ||||
| BCG — Executive Perspectives: Unlocking Impact from GenAI and Agentic AI in HR (EP1) | 2025 | GenAI and agentic AI across HR administration, shared services, recruiting, onboarding, L&D | 20-30% of HR administration & shared-services cost; up to 30% productivity across the whole HR function function-wide | High |
| PwC Strategy& — Agentic AI in HR | 2025 | Agentic AI covering end-to-end HR processes (screening, scheduling, employee service, payroll) | 15-30% of administrative HR cost broad | Med |
| Deloitte — HR Reimagined: Agentic AI for HR | 2025 | Agentic AI workflow library automating HR process execution, service delivery, and employee inquiries | ~20-25% of HR staff capacity/cost freed broad | Med |
| Salesforce — Agentic AI Impact on Workforce Research (survey of global HR executives) | 2025 | Agentic AI for HR service, recruiting, and administrative workflows; workforce redeployment | ~19-25% of HR labor cost once agentic AI is fully deployed broad | Med |
| Gartner — Unlocking AI Value in HR and the Enterprise | 2025 | GenAI and agentic AI automating transactional/administrative HR tasks | ~15-30% of function cost indicatively (up to ~50% of HR tasks automatable) broad | Low |
| 7 Order processing & back-office operations | ||||
| The Hackett Group — 2025 US SG&A Cost Study and Scorecard (Agentic Strategy to Slash SG&A Costs) | 2025 | Agentic AI operating model across transactional SG&A / back-office functions (finance, procurement, HR, IT shared services) | ~15-20% of the SG&A / back-office cost base broad | Med |
| McKinsey & Company — The paradigm shift: How agentic AI is redefining banking operations | 2025 | Agentic AI automating end-to-end back-office / operations workflows (case handling, verification, approvals) | 15-20% net reduction of the operations cost base broad | High |
| Oliver Wyman — 3 Ways To Harness Generative AI For Better Claims Management | 2025 | Generative AI across the claims-handling function (intake, document processing, adjudication) | 5-25% early adoption; 20-40% with mature/aggressive deployment function-wide | Med |
| Deloitte — 2025 Global Business Services (GBS) Survey | 2025 | GBS/shared-services automation and GenAI (invoice management, chatbots/AI tools, analytics) | ~20%+ of shared-services / back-office operating cost function-wide | Low |
| McKinsey & Company — Agentic AI will shake up banking, shrinking global profit pools | 2025 | Agentic AI managing autonomous end-to-end back-office workflows (1 human supervising 20-30 agents) | 20%+ of operational costs (≈9-15% of operating profit) broad | High |
| 8 R&D, engineering & product design | ||||
| McKinsey & Company — The next innovation revolution, powered by AI | 2025 | AI-driven discovery, generative design and simulation across the R&D/engineering workflow | ~15-40% of R&D development cost (converted from 20-80% throughput acceleration) function-wide | Med |
| BCG — Executive Perspectives: AI-Powered R&D | 2025 | AI agents orchestrating R&D workflows and AI-driven design optimization | ~25-40% of targeted engineering effort broad | Med |
| BCG — The Widening AI Value Gap: Build for the Future 2025 | 2025 | Agentic AI concentrated in core value-creating functions including R&D and innovation | ~10-20% of function cost for AI leaders (indicative) broad | Low |
| Index.dev — AI Coding Assistant ROI: Real Productivity Data 2025 (aggregating GitHub, DORA, METR) | 2025 | Engineering copilots for software/product engineering (a sub-function of R&D) | ~10-30% of software-engineering time cost partial | Low |
| 9 IT & software development | ||||
| McKinsey & Company — The State of AI in 2025: Agents, innovation, and transformation | 2025 | Generative-AI coding assistants and agents across software engineering and IT | 10-20% of the function's operating cost (sometimes higher) broad | High |
| McKinsey & Company — The AI revolution in software development | 2025 | Gen-AI coding copilots and agentic development across the SDLC | ~11-20%+ of software-engineering cost for organizations already seeing reductions function-wide | High |
| BCG — Executive Perspectives: Driving Sustainable Cost Advantage with AI | 2025 | GenAI-assisted coding and generative engineering tools | 20-30% of the software-engineering function cost (5-10% at early/basic adoption) broad | Med |
| Gartner — Don't Limit AI in Software Engineering to Coding | 2025 | Ensemble of AI tools across the SDLC (code generation plus test, review, ops) | ~10% of engineering effort today from code-gen; up to 25-30% as tooling extends across the SDLC broad | Med |
| Faros AI — Do AI Coding Assistants Really Save Time, Money and Effort? (citing DORA, GitHub, Bain, METR, enterprise data) | 2025 | AI coding assistants in enterprise engineering teams | ~15-35% of development-activity cost (highly variable; some studies show near-zero or negative) partial | Low |
| 10 Marketing & brand | ||||
| BCG — How Leaders Build an AI-First Cost Advantage | 2026 | GenAI content generation and end-to-end marketing process redesign (content, campaign assets, agency work) | 20-30% of marketing content and agency spend (~95% on raw content generation as a sub-task) broad | Med |
| Gartner — 2025 CMO Spend Survey | 2025 | GenAI across the marketing function; reduced reliance on agencies and overlapping labor | ~10-20% of function operating cost (modeled from reported efficiency + labor/agency cuts) function-wide | Med |
| Bain & Company — Your AI Budget Is Growing, Your Returns Aren't (2026 AI cost-savings survey, 951 companies) | 2026 | Enterprise GenAI/agentic deployment across functions including marketing | 0-10% actual today for most; 11-20% targeted broad | Med |
| Gartner — 2026 CMO Spend Survey | 2026 | AI investment across the marketing function, funded by agency cuts | ~0-10% net function cost reduction realized today (savings mostly reallocated, not cut) function-wide | Low |
| Averi.ai — State of AI in Marketing 2026 (benchmarks report) | 2026 | Generative AI content and creative production across formats | ~37-42% of content-production cost partial | Low |
| 11 Legal, compliance & corporate strategy | ||||
| Gartner — AI in the Legal Industry (trend brief) / Top 6 Use Cases for Generative AI in Legal Departments | 2025 | Generative-AI legal assistants (drafting, research, contract review) plus workflow automation across the legal department | 10-20% of the function's operating cost broad | High |
| Thomson Reuters Institute — Future of Professionals 2025 | 2025 | AI tools embedded in daily legal work (research, drafting, review, summarization) | ~10-15% of professional labor cost broad | High |
| McKinsey — AI in practice: Shaping the law firm of the future / Legal innovation and generative AI | 2026 | Generative-AI automation and augmentation of legal activities (research, review, drafting) | 15-25% of total net legal hours (cost proxy) function-wide | Med |
| Deloitte — Zora AI launch (agentic AI for the workforce) | 2025 | Agentic AI 'digital workers' for knowledge-work functions including legal, compliance and finance | ~25% of function cost (vendor target) broad | Med |
| Bloomberg Law / Sirion & CLOC contract-review benchmarks (2025-26 compilation) | 2025 | Contract-review and playbook-redlining LLMs | ~30-35% cost on contract workflows (70-85% time per contract) partial | Med |
| 12 Sales & business development | ||||
| The Hackett Group — 2025 US SG&A Cost Study and Scorecard | 2025 | Agentic, AI-enabled operating model across SG&A (which contains the sales/commercial function) | 15-20% of SG&A cost broad | Med |
| Gartner — Survey on AI and seller time (press release, May 2026) | 2026 | Sales copilots / AI assistants automating non-selling admin and research tasks | 10-12% of seller labor cost broad | Med |
| Bain & Company — Technology Report 2025: AI Is Transforming Productivity, but Sales Remains a New Frontier | 2025 | Agentic AI + sales copilots offloading the work surrounding selling | 15-25% of sales operating cost (converted from productivity) broad | Med |
| BCG — AI Leaders Outpace Laggards (press release, Sept 2025) | 2025 | Scaled/agentic AI across commercial and other functions | ~40% greater cost savings vs laggards (relative, not absolute) partial | Low |
| DigitalApplied — AI SDR Statistics 2026 (citing Bridge Group SDR Metrics 2026 vendor benchmark) | 2026 | AI SDR agents in hybrid AI+human prospecting pods | ~50-55% of cost per qualified opportunity (prospecting sub-task only) partial | Low |
| 13 Customer service & success | ||||
| ISG (Ventana Research) — "AI Cuts Costs by 30%, But 75% of Customers Still Want Humans" | 2025 | AI voice agents, chatbots and self-service deflection at scale in the contact center | 25-35% of the customer-service function operating cost function-wide | Med |
| Deloitte Digital — "The Future of Service: The Age of Intelligent Experience" (2026 Global Contact Center Survey) | 2026 | Agentic AI across service: agent-assist copilots, autonomous multi-step resolution, self-service | 20-30% of the contact-center function cost broad | High |
| BCG — "Agentic AI Is the New Frontier in Customer Service Transformation" | 2025 | Agentic AI resolving end-to-end interactions plus agent-assist, reducing headcount and cost per interaction | 15-25% of the function cost (labor-driven) broad | Med |
| McKinsey & Company — "From promising to productive: Real results from gen AI in services" | 2025 | Gen AI raising productivity of the customer-care function plus deflecting human-serviced contacts | 20-40% of current function cost function-wide | Med |
| Gartner — "Gartner Predicts Agentic AI Will Autonomously Resolve 80% of Common Customer Service Issues Without Human Intervention by 2029" | 2025 | Agentic AI autonomously resolving common service issues without human intervention | 25-35% of operational cost function-wide | Med |
| 14 Executive leadership & general management | ||||
| BCG — Corporate Functions of the Future Won't Look Like Functions at All | 2026 | Agentic AI coordinating workflows and de-layering middle management | 25-50% of the function cost broad | Med |
| BCG — Leading in the Age of AI Agents: Managing the Machines That Manage Themselves | 2025 | Agentic management: agents coordinate workflows, widening spans of control and flattening layers | 15-30% of the function cost (modeled) broad | Low |
| McKinsey Global Institute — The State of AI 2025 / Agents, robots, and us | 2025 | AI agents automating analysis, reporting and routine decision tasks in managerial work | 20-40% of the function cost (modeled) broad | Med |
| Gartner — CFOs Should Reset Expectations About AI's Impact on Workforce Productivity and Headcount | 2025 | AI decision support and productivity tools in knowledge/management roles | 0-10% of the function cost broad | Med |
| World Economic Forum — Future of Jobs Report 2025 | 2025 | AI augmentation and automation of management and decision tasks | 10-25% of the function cost (modeled) broad | Low |
Eventual whole-function reduction with all foreseeable levers (agentic AI, humanoid robots)
| Source (research report) | Yr | Major AI lever | Cost-reduction range | Conf. |
|---|---|---|---|---|
| 1 Production / Manufacturing | ||||
| Boston Consulting Group (BCG) — How the Factory of the Future Is Reshaping the Economics of Manufacturing Competitiveness | 2026 | Physical AI + agentic systems + advanced robotics (fully AI-run 'factory of the future') | 40-60% of manufacturing conversion cost function-wide | High |
| McKinsey Global Institute — Agents, Robots, and Us: Skill Partnerships in the Age of AI | 2025 | AI agents + robots (incl. advanced/humanoid automation of physical machine-operator work) | ~50-60% of manufacturing labor cost (converted from work-hours automatable) broad | Med |
| Morgan Stanley Research — Humanoid Robot Market Expected to Reach $5 Trillion by 2050 | 2025 | Humanoid robots with advanced dexterity replacing manual factory labor | ~50-80% of manual manufacturing labor cost (long-run) broad | Med |
| Goldman Sachs Research — The Global Market for Humanoid Robots Could Reach $38 Billion by 2035 | 2025 | Humanoid robots (falling unit economics enabling autonomous physical work) | up to ~60-70% of automatable manual-task labor cost partial | Low |
| 2 Logistics, warehousing & distribution | ||||
| Oliver Wyman — How Logistics Operators Harness AI To Boost Efficiency | 2025 | AI across the whole logistics operation: demand/route optimization, warehouse management, sorting and last-mile automation | 10-25% of the function's operating cost base broad | High |
| Deloitte Insights — Driverless autonomous trucks lead the way | 2025 | Full-autonomy autonomous trucking (driver removal, near-24/7 utilization, fuel and safety optimization) | 30-42% of transport cost (transport is the largest logistics cost pool) broad | High |
| Symbotic — FY2025 results / warehouse automation benchmarks | 2025 | AI-powered warehouse robotics (autonomous case handling, storage/retrieval, sortation, pallet building) | 25-40% of warehouse operating cost per unit shipped broad | Med |
| McKinsey — Will embodied AI create robotic coworkers? (physical AI / humanoids) | 2025 | Humanoid robots with advanced dexterity and near-full autonomy substituting for manual warehouse labor | 30-50% of the warehouse manual-labor cost pool (modeled) broad | Low |
| World Economic Forum — Physical AI: Powering the New Age of Industrial Operations | 2025 | Physical AI: humanoids, mobile manipulators, quadrupeds and autonomous fleets across industrial and logistics operations | 25-45% of function operating cost at full deployment (modeled) broad | Low |
| 3 Field service, installation & maintenance | ||||
| McKinsey Global Institute — Agents, Robots, and Us: Skill Partnerships in the Age of AI | 2025 | Agentic AI plus physical/humanoid robots automating on-site technician labor | 40-57% of the function's labor cost (eventual ceiling) broad | Med |
| Morgan Stanley — Humanoid Robot Market Expected to Reach $5 Trillion by 2050 (Humanoid 100) | 2025 | Dexterous, fully autonomous humanoid robots replacing manual field/installation/repair labor | 40-60% of the function's labor cost (long-run ceiling) broad | Low |
| Pomeroy / Opsima — Agentic AI Frameworks & Agentic Workflows in Field Operations | 2026 | Agentic AI stack: predictive maintenance + smart scheduling/dispatch + AI technician support | 25-40% of the function's operating cost broad | Med |
| McKinsey — Predictive maintenance analyses (Rewiring maintenance with gen AI; utility field-force pilot) | 2025 | AI/IoT predictive maintenance and AI schedule optimization for the field force | 10-40% of maintenance cost (function sub-domain) broad | Med |
| PTC — State of Industrial Augmented Reality 2025 | 2025 | AR/AI-guided technicians (remote expert, step-by-step overlays) | 15-30% of on-site labor cost (single lever) partial | Med |
| 4 Accounting operations | ||||
| KPMG — The Future of Agentic Finance | 2025 | Autonomous/agentic finance: end-to-end agent coverage across close, reporting, AP/AR, FP&A (digital command center) | 40-60% of total finance costs function-wide | High |
| EY — AI in finance: the hidden advantage for tech companies | 2026 | End-to-end agentic AI transformation of the entire finance operation (vs piecemeal RPA/automation) | up to ~40% of finance function cost function-wide | Med |
| BCG — The AI-First Finance Function | 2026 | Agentic workflows across finance; near-zero marginal cost per transaction, humans only on exceptions | ~40% of operating cost, with G&A falling from ~8% to 3-5% of revenue (~40-60%) broad | Med |
| McKinsey — AI in finance: how finance teams are putting AI to work today | 2025 | Gen AI assistants / agentic automation across data crunching, planning, reporting, working-capital | 20-30% (time-saving proxy for labor cost) broad | Med |
| The Hackett Group — 2025 US SG&A Cost Study | 2025 | Agentic, AI-enabled operating model across SG&A (incl. finance) | ~$290M annual savings per $10B company (~a mid-double-digit % of finance/SG&A spend) partial | Low |
| 5 Purchasing operations | ||||
| McKinsey & Company — "Transforming procurement functions for an AI-driven world" (corroborated by "Redefining procurement performance in the era of agentic AI") | 2025 | Agentic AI / autonomous category and sourcing agents running end-to-end workflows (sourcing, negotiation, contract drafting, spend/performance monitoring) | 25-40% function-wide | High |
| The Hackett Group — Gen AI in Procurement 2025 research ("64% of Procurement Leaders Say AI Will Transform Their Jobs") | 2025 | Generative + agentic AI across purchase-order processing, spend analytics, contract management and e-procurement (Digital World Class operating model) | up to ~47% (report ~40-47%) function-wide | High |
| Bain & Company — "The Rise of Autonomous, Intelligent Procurement" | 2026 | Autonomous, intelligent procurement — AI agents monitoring demand/supplier/market signals and executing sourcing, negotiation and category optimization end-to-end | ~35-40% (converted from 60%+ productivity gain) broad | Med |
| 6 HR & workforce administration | ||||
| Gartner — The Future of AI in HR: Reinventing the Operating Model | 2025 | Agentic AI / AI agents autonomously executing HR activities across recruiting, service delivery, admin | 40-50% of the function's activities/cost function-wide | High |
| IBM (IBM Consulting / IBM Think) — Artificial Intelligence for Human Resources | 2025 | HR self-service chatbots plus agentic automation (AskHR on watsonx Orchestrate) removing manual HR tasks | 50-60% of HR service-delivery cost broad | High |
| IBM — AskHR case study: Transforming HR support with agentic AI | 2025 | Agentic HR virtual agent (AskHR + watsonx Orchestrate) automating 80+ HR tasks, 94% containment | ~40% of HR operational cost (realized) broad | High |
| McKinsey — Superagency in the workplace / Generative AI and the future of HR | 2025 | Generative AI automating administrative and transactional HR work; agentic HR workflows | 50-70% of the function's admin/transactional cost broad | Med |
| SHRM — The State of AI in HR 2026 Report | 2026 | HR chatbots, self-service platforms, auto-responders, and automation of routine/transactional HR tasks | 20-40% of the function cost (modeled from productivity) partial | Low |
| 7 Order processing & back-office operations | ||||
| BCG — Agents Accelerate the Next Wave of AI Value Creation | 2025 | Agentic back-office with full end-to-end process redesign (agents sequencing multi-step transactions, humans handling exceptions) | 60%+ (ceiling with end-to-end redesign) function-wide | High |
| BCG — The AI-First Finance Function | 2026 | Autonomous transaction processing across order-to-cash, procure-to-pay, expense, payroll executing autonomously end-to-end; humans only on flagged exceptions | ~60% (up to 90% on some sub-tasks) function-wide | Med |
| McKinsey — The paradigm shift: how agentic AI is redefining banking operations | 2025 | End-to-end redesign of back-office operations workflows with humans 'above the loop', pushing marginal transaction cost toward the cost of compute | up to 70% (ceiling for redesigned cost categories) broad | Med |
| NiCE — Achieving Cost Reduction with Autonomous AI Agents | 2026 | Autonomous AI agents resolving back-office transactions end-to-end with minimal human intervention | 40–60%, up to 70% broad | Low |
| 8 R&D, engineering & product design | ||||
| BCG — Reinventing the Operating System of Work with AI | 2026 | Agentic AI with end-to-end process redesign (engineering copilots + autonomous agent workflows) | 15-60% of the function's operating cost broad | Med |
| BCG — AI Leaders Outpace Laggards (press release + AI-First Cost Advantage) | 2025 | Enterprise-wide AI / agentic automation across core functions incl. R&D | 20-40% of operating cost broad | Med |
| McKinsey — The State of AI 2025 (QuantumBlack) | 2025 | AI embedded across the full software/product-development life cycle (engineering copilots, agents) | 16-45% of the engineering/software function's effort broad | Med |
| DataM Intelligence — AI-Driven Drug Discovery 2025 | 2025 | Generative design / AI molecule generation and simulation in R&D | 25-50% of preclinical R&D cost (modeled from -50% timelines) broad | Low |
| BCG — consumer-goods product-development agent case (in AI-First reports) | 2026 | Custom AI agents chaining concept-to-product design workflows | 25-40% of the covered design/development workflows partial | Med |
| 9 IT & software development | ||||
| McKinsey — The AI revolution in software development | 2026 | Agent factory: full end-to-end agentic software delivery (day/night cycle — humans direct/design, coordinated agents execute, iterate, test) | ~50% (half the cost, at 10x speed) function-wide | High |
| Milestone (gomilestone) — How AI Reduces Software Development Costs | 2026 | AI integrated across the full project lifecycle (prototyping, scaffolding/boilerplate, automated testing & documentation, compressed code review & debugging) | 40–50% of total project cost broad | Med |
| Belitsoft — Agentic AI Coding: What Still Remains Expensive Amid a 90% Drop | 2026 | Agentic AI code generation; the cap from non-compressible maintenance, operations, coordination and infrastructure | up to ~50% whole-project (coding labor alone drops up to 90%) partial | Med |
| AWS (Amazon) — How frontier teams are reinventing AI-native development | 2026 | Agent-first AI-native development redesign (agents as primary actors, near-continuous execution, self-correcting integration testing) | no direct cost %; 4.5–20x productivity frontier (implies ~78–95% delivery-labor cost reduction on greenfield) partial | Low |
| 10 Marketing & brand | ||||
| McKinsey & Company (Growth, Marketing & Sales) — "Reinventing marketing workflows with agentic AI" | 2025 | Agentic AI orchestrating end-to-end marketing workflows (content generation, synthetic audience testing, media planning, campaign execution) | ~40-65% of marketing function labor cost (long-run ceiling) function-wide | Med |
| BCG — "Moving the Agentic Marketing Transformation from Illusion to Reality" | 2026 | Multi-agent orchestration across the marketing system (brand intelligence layer, content, campaigns, agency work) | 15-30% of internal + agency marketing spend broad | High |
| BCG — "The Agentic Marketing Race Is On. CMOs That Move First Will Win." | 2025 | Agentic marketing operating model (data foundation + agent orchestration across content and campaigns) | 20-30% of marketing operating cost broad | High |
| Accenture — "AI Refinery: Smarter & Faster Marketing" (Song / SynOps marketing case) | 2025 | Generative + agentic AI for high-volume content production and campaign management automation | ~30-55% of campaign/content operations cost partial | Med |
| 11 Legal, compliance & corporate strategy | ||||
| BCG — Reinventing the Operating System of Work with AI (also 'How Leaders Build an AI-First Cost Advantage') | 2026 | Full end-to-end agentic redesign of a function's workflows (autonomous agent squads acting, not just assisting), applied across knowledge/support functions including legal and compliance | 60-70% function-wide | High |
| Network Law Review — Paul Ohm, 'Toward Compliance Zero: AI and the Vanishing Costs of Regulatory Compliance' | 2025 | Autonomous agentic compliance workflows executing regulatory obligations end-to-end, driving per-obligation compliance cost asymptotically toward zero | 75-95% function-wide | Med |
| LegalMind — Agentic AI-Driven Process Optimization and Cost Reduction in Legal Services Using DeepSeek (peer-reviewed/ResearchGate) | 2025 | Full agentic AI legal-services system averaged across contract analysis, litigation review, and regulatory compliance checks | 43-62% function-wide | Med |
| McKinsey — Seizing the Agentic AI Advantage | 2025 | Agentic reinvention of a full workflow (single integrated platform running agent squads) applied to support functions including contract review and legal/compliance operations | 30-50% broad | Med |
| 12 Sales & business development | ||||
| World Economic Forum — Future of Jobs Report 2025 | 2025 | Gen-AI / agentic task automation across the seller role (prospecting, outreach, follow-up, deal support) | 45-60% of the sales function's labor cost (long-run ceiling) broad | Med |
| BCG — How Leaders Build an AI-First Cost Advantage | 2026 | Agentic AI with end-to-end go-to-market/SG&A process redesign | 50-60%+ of the function's operating cost broad | Med |
| BCG — How AI Agents Will Transform B2B Sales | 2025 | Augmented, assisted, and fully autonomous selling agents (AI SDRs, copilots, autonomous deal agents) | 40-60% of sales-function operating cost (long-run, autonomous-agent end state) broad | Low |
| McKinsey — AI in the workplace: A report for 2025 (Superagency) | 2025 | Gen AI plus autonomous agents automating knowledge/communication work | 50-65% of function labor time/cost (technical ceiling) broad | Med |
| Landbase / industry AI SDR benchmarks (GTM cost analysis) | 2026 | Autonomous AI SDR agents for prospecting and lead qualification | 60-70% of prospecting/SDR sub-function cost partial | Low |
| 13 Customer service & success | ||||
| BCG — Agentic AI Is the New Frontier in Customer Service Transformation / Unlocking Impact from Agentic AI in Customer Service | 2025 | Full end-to-end agentic redesign of the customer service function (autonomous agentic resolution + agent-assist + self-service deflection), zero-based process redesign at scale | 60%+ (ceiling with end-to-end redesign; <20% without redesign) function-wide | High |
| McKinsey — The future of customer experience: embracing agentic AI (and related customer-care AI insights) | 2025 | Agentic AI with zero-based / end-to-end redesign of customer-care journeys; augmentation + automation + demand reduction across the function | ~50–60% (cost-to-serve ceiling implied by automating up to 60% of addressable care volume via full redesign) broad | Med |
| Bluetweak — Customer Support Cost Reduction AI: How Support Teams Lower Cost to Serve in 2026 (analyst/benchmark synthesis) | 2026 | Full hybrid AI-human contact-center redesign at scale — autonomous AI resolution plus agent-assist and self-service containment | up to 60–70% operational cost reduction (redesign at scale; conservative 2–3yr layering is only 20–40%) broad | Low |
| 14 Executive leadership & general management | ||||
| The Hackett Group — 2025 US SG&A Cost Study and Scorecard | 2025 | Agentic AI-enabled operating model across SG&A/G&A (agentic management, automated coordination and reporting) | 15-20% broad | Med |
| McKinsey & Company — The agentic organization: Contours of the next paradigm for the AI era (People & Organizational Performance) | 2025 | De-layering and agentic management: one manager plus a 2-5 person team orchestrating 50-100+ agents, flatter org with fewer middle-management layers | 30-50% function-wide | Low |
| BCG — Driving Sustainable Cost Advantage with AI (Executive Perspectives) | 2025 | Agentic AI with operating-model redesign of management/overhead work (automated coordination, decision support, de-layering) | 30-60% broad | Low |
The same eventual ceiling as estimated in 2023, the first post-ChatGPT year
| Source (research report) | Yr | Major AI lever | Cost-reduction range | Conf. |
|---|---|---|---|---|
| 1 Production / Manufacturing | ||||
| McKinsey Global Institute, "The economic potential of generative AI: The next productivity frontier" | 2023 | Generative AI applied across manufacturing and supply-chain operations (root-cause/defect prediction, dynamic work instructions, automated planning), on top of existing physical automation | up to ~30-50% of manufacturing/supply-chain operating cost (long-run ceiling, modeled) function-wide | High |
| McKinsey Global Institute, "The economic potential of generative AI: The next productivity frontier" | 2023 | Generative AI as a cross-function productivity uplift (the headline per-function cost framing) | 30-45% of a function's current cost (cross-function ceiling; manufacturing sits at the low end) broad | High |
| World Economic Forum, "Future of Jobs Report 2023" | 2023 | Machine/algorithm and robotics automation of business tasks | ~34% of tasks machine-done today rising toward ~42-45% by 2027 (share of task hours automatable) broad | Med |
| Goldman Sachs, "The Potentially Large Effects of Artificial Intelligence on Economic Growth" (Briggs/Kodnani) | 2023 | Generative AI automation of occupational task content | ~15-25% of production labor content exposed (lower end; production is among the least genAI-exposed) broad | Med |
| Accenture, "A new era of generative AI: Work, workforce, workers reinvented" | 2023 | Language-based generative AI augmenting/automating working hours; manufacturing use cases (operations twins, technical docs, test generation) | ~40% of working hours augmented/automated economy-wide; 40-50% effort reduction in manufacturing pilot use cases partial | Med |
| McKinsey, "The state of AI in 2023: Generative AI's breakout year" | 2023 | AI adoption in manufacturing (analytics, computer vision, predictive maintenance) | cost decreases of 10%+ reported by 66% of manufacturing adopters (share reporting savings, not magnitude) partial | Med |
| 2 Logistics, warehousing & distribution | ||||
| Goldman Sachs Global Investment Research — The Potentially Large Effects of Artificial Intelligence on Economic Growth (Briggs/Kodnani) | 2023 | Generative and broader AI task automation (route/dispatch, documentation, planning tasks; physical moving excluded) | ~11-12% of the function's frontline moving-goods labor tasks automatable (labor-cost ceiling proxy) broad | Med |
| McKinsey / MGI — The economic potential of generative AI: The next productivity frontier | 2023 | Generative AI across supply chain, warehousing, transportation and support tasks | ~1-5% of function cost from generative AI use cases (function ranks among the lowest) broad | Low |
| World Economic Forum — Future of Jobs Report 2023 | 2023 | Automation of tasks by machines and algorithms (incl. AI) across work activities | ~34% of tasks machine-performed in 2023, rising to ~43% by 2027 (cross-function task-automation ceiling) broad | Med |
| McKinsey / QuantumBlack — The state of AI in 2023: Generative AI's breakout year | 2023 | AI adoption across supply-chain and inventory management (forecasting, optimization) | function-wide cost decreases reported, typically under ~20% of function cost function-wide | Low |
| 3 Field service, installation & maintenance | ||||
| Goldman Sachs Global Investment Research — The Potentially Large Effects of Artificial Intelligence on Economic Growth (Briggs & Kodnani) | 2023 | Generative AI applied to occupational task bundles (the automatable share of installation/maintenance/repair work) | ~4% of tasks exposed (≈2-6% of function labor cost) function-wide | High |
| McKinsey Global Institute — The economic potential of generative AI: The next productivity frontier | 2023 | Generative AI plus prior automation technologies across all work activities (technical automation ceiling) | 60-70% of work time (economy-wide ceiling; physical/field work far lower) broad | Low |
| World Economic Forum — Future of Jobs Report 2023 | 2023 | Machine/algorithmic automation of business tasks (survey-based human-machine frontier) | 34% of tasks automated today rising to ~42% by 2027 (all functions) broad | Med |
| Goldman Sachs Global Investment Research — The Potentially Large Effects of Artificial Intelligence on Economic Growth | 2023 | AI automation across the whole labor force (aggregate substitution ceiling) | up to ~25% of current work tasks (all occupations) broad | Med |
| 4 Accounting operations | ||||
| Goldman Sachs Global Investment Research — The Potentially Large Effects of Artificial Intelligence on Economic Growth | 2023 | Generative AI task automation of knowledge/office work | 20-35% of function labor cost broad | Med |
| World Economic Forum — Future of Jobs Report 2023 | 2023 | Automation of information- and data-processing tasks (RPA + AI) | 25-40% of function cost broad | Med |
| McKinsey Global Institute — The economic potential of generative AI: The next productivity frontier | 2023 | Generative AI plus prior automation across work activities | 30-50% of function cost broad | Med |
| Accenture — Work, workforce, workers: Reinvented in the age of generative AI | 2023 | Generative AI automation and augmentation of language tasks | 20-40% of function cost partial | Low |
| 5 Purchasing operations | ||||
| The Hackett Group — "Generative AI Will Drive Profound Reductions in SG&A Cost and Staffing — Up to 40% Over the Next Five to Seven Years" | 2023 | Generative AI across end-to-end SG&A processes (procurement one of the analyzed functions): spend analytics, sourcing, contract/PO processing, supplier management | up to ~40% (ceiling over 5-7 years) broad | High |
| KPMG — "How generative AI will transform procurement as we know it" / "Unleashing the Power of Generative AI in Procurement" | 2023 | Generative AI across the whole procurement function: agentic/conversational sourcing, spend and category intelligence, autonomous P2P, supplier management | ~25-40% (conservatively converted from 50-80% of procurement work automatable/eliminable/self-service) function-wide | Med |
| McKinsey Global Institute — "The economic potential of generative AI: The next productivity frontier" | 2023 | Generative AI plus prior automation applied to knowledge/back-office work (spend analysis, sourcing, contract admin, supplier ops) | ~20-30% (conservatively converted, broad anchor; procurement not isolated) broad | Low |
| 6 HR & workforce administration | ||||
| McKinsey Global Institute — The economic potential of generative AI: The next productivity frontier | 2023 | Generative AI for HR administration: HR chatbots/virtual assistants, automated screening and candidate matching, and drafting/answering employee queries across the talent-and-organization (HR) function | 20-40% of the function cost function-wide | Med |
| Goldman Sachs Global Investment Research — The Potentially Large Effects of Artificial Intelligence on Economic Growth (Briggs/Kodnani) | 2023 | Generative AI automating routine document-, query- and record-handling tasks in office and administrative support work, the task category most representative of HR/workforce administration | ~46% of function tasks (task-share proxy) broad | Med |
| Accenture — A New Era of Generative AI for Everyone (Technology Vision / generative AI report) | 2023 | Language-based generative AI automating and augmenting knowledge-work tasks across corporate support functions, including HR administrative and employee-service work | ~40-44% of working hours in scope (automation + augmentation) broad | Low |
| McKinsey Global Institute — Generative AI and the future of work in America | 2023 | Generative AI accelerating automation of routine clerical and administrative activities (office support, administrative assistants), the core of workforce administration | ~30% of work hours automatable by 2030 (office-support-heavy) broad | Med |
| World Economic Forum — Future of Jobs Report 2023 | 2023 | Broad workplace automation (AI/algorithms) of reasoning, communicating and coordinating tasks that dominate HR and talent-acquisition roles | ~34% now to ~42% of tasks automated by 2027 partial | Low |
| 7 Order processing & back-office operations | ||||
| McKinsey Global Institute — "The economic potential of generative AI: The next productivity frontier" | 2023 | Agentic / generative AI plus intelligent document processing across a service-operations back office (case handling, customer/account servicing) | 30-45% of current function cost broad | High |
| McKinsey — "Ten systemic productivity levers to fuel corporate and investment banking growth" | 2023 | RPA plus digitization and automation of shared-services/support activities under zero-based redesign | 25-35% of function cost (converted from 40-50% of activities automatable) function-wide | Med |
| Bain & Company — "The $100 Billion Opportunity for Generative AI in P&C Claims Handling" | 2023 | Generative/agentic AI plus intelligent document processing across the full claims-handling case-management function | 20-25% of function operating cost broad | High |
| 8 R&D, engineering & product design | ||||
| McKinsey & Company / MGI, The economic potential of generative AI: The next productivity frontier | 2023 | Generative design and AI-assisted R&D (candidate generation, simulation, design optimization) | 10-15% of overall R&D costs function-wide | High |
| Goldman Sachs Global Economics Analyst (Briggs/Kodnani), The Potentially Large Effects of Artificial Intelligence on Economic Growth | 2023 | Generative AI automation/augmentation of engineering and design tasks | ~25-37% of engineering task-hours exposed (upper bound on labor-cost reduction) broad | Med |
| Accenture, A New Era of Generative AI for Everyone (Technology Vision 2023) | 2023 | Language-based generative AI augmenting/automating knowledge work | ~20-40% of working hours augmented or automated (cross-function labor ceiling) broad | Med |
| Bain & Company, The Talent Implications of Generative AI (Technology Report 2023) | 2023 | Engineering copilots / generative AI in software and product engineering | ~20%+ productivity gain expected by engineers (partial, software-engineering subset) partial | Low |
| BCG, The CEO's Roadmap on Generative AI (Executive Perspectives, March 2023) | 2023 | Generative AI across R&D, engineering and product development workflows | R&D flagged among highest-impact functions (qualitative, no single function-wide %) partial | Low |
| 9 IT & software development | ||||
| McKinsey Global Institute — The economic potential of generative AI: The next productivity frontier | 2023 | Generative-AI coding assistants and agentic software engineering (code generation, correction/refactoring, root-cause analysis, system design) | 20-45% of the software-engineering function's annual spend function-wide | High |
| Goldman Sachs (Briggs & Kodnani) — The Potentially Large Effects of Artificial Intelligence on Economic Growth (Global Economics Analyst) | 2023 | Generative AI automating task content of computer/programming occupations | ~15-29% of the function's labor cost (ceiling ~29% of tasks automatable) broad | Med |
| McKinsey Digital — Unleashing developer productivity with generative AI | 2023 | AI pair-programming across code generation, refactoring, and documentation | ~10-30% of software-development (coding) effort/cost broad | Med |
| Accenture — A New Era of Generative AI for Everyone / Technology Vision 2023 | 2023 | Language-based generative AI augmenting and automating knowledge-work tasks, including software development | ~15-40% of working hours augmented or automated (software dev among most-exposed) broad | Low |
| GitHub / Microsoft Research — The Impact of AI on Developer Productivity: Evidence from GitHub Copilot (arXiv 2302.06590) | 2023 | AI code-completion (GitHub Copilot) on a coding task | ~20-55% faster on the measured task (single-task, not function-wide) partial | High |
| 10 Marketing & brand | ||||
| McKinsey Global Institute — The economic potential of generative AI: The next productivity frontier | 2023 | Generative content, personalization, and campaign/analytics automation across the marketing function | 5-15% of total marketing spend function-wide | High |
| McKinsey — How generative AI can boost consumer marketing (Growth, Marketing & Sales) | 2023 | Gen-AI content creation, hyper-personalization, and marketing-data analysis | 5-15% of total marketing spend function-wide | High |
| BCG — Benefits of Generative AI in Marketing (survey of 200+ CMOs) | 2023 | Gen-AI content generation, insight generation, and segmentation across marketing work | up to ~30% of the function's productivity/effort (proxy for cost) broad | Med |
| Accenture — A New Era of Generative AI / Work, workforce, workers reinvented in the age of generative AI | 2023 | Language-based gen-AI automating and augmenting knowledge-work tasks | ~20-40% of function labor cost (share of hours automatable, subset of the 40% augmented/automated) broad | Low |
| Goldman Sachs Research — The Potentially Large Effects of Artificial Intelligence on Economic Growth (Briggs/Kodnani) | 2023 | Gen-AI automation of exposed task content in marketing/creative occupations | ~25-50% of workload of exposed marketing occupations partial | Med |
| 11 Legal, compliance & corporate strategy | ||||
| Goldman Sachs Global Economics — 'The Potentially Large Effects of Artificial Intelligence on Economic Growth' (Briggs & Kodnani) | 2023 | Generative-AI legal-research and contract-review LLMs automating drafting, review and research tasks | ~22–44% of legal-function labor cost function-wide | Med |
| Accenture — 'A New Era of Generative AI for Everyone' (LLM work-impact study) | 2023 | Large-language-model automation and augmentation of language-based legal and compliance work | ~20–40% of legal-function cost broad | Low |
| Thomson Reuters Institute — 'Future of Professionals Report' (August 2023) | 2023 | Generative-AI assistants for legal research, drafting and review across the legal function | ~10–12% of legal-function cost function-wide | Med |
| 12 Sales & business development | ||||
| McKinsey Global Institute — The economic potential of generative AI: The next productivity frontier | 2023 | Generative AI sales copilots and automation across the sales cycle (lead identification, outreach, deal support) | 3-5% of the sales function's cost function-wide | High |
| Goldman Sachs Global Investment Research — The Potentially Large Effects of AI on Economic Growth (Briggs/Kodnani, Mar 2023) | 2023 | Generative AI automation of sales work tasks (prospecting research, drafting, follow-up) | ~25-31% of the sales function's labor cost (share of tasks automatable) broad | Med |
| Accenture — A New Era of Generative AI for Everyone (Technology Vision 2023) | 2023 | Language-based generative AI augmenting/automating customer-facing sales work | ~20-40% of sales working hours automatable/augmentable (only part is net cost) broad | Low |
| World Economic Forum — Future of Jobs Report 2023 (May 2023) | 2023 | AI/automation of routine sales and administrative sales-support tasks | ~34-42% of sales-related tasks automatable by 2027 (labor-cost proxy) broad | Low |
| 13 Customer service & success | ||||
| McKinsey Global Institute / McKinsey & Company — "The economic potential of generative AI: The next productivity frontier" | 2023 | Agentic/generative AI agents, agent-assist copilots and self-service deflection across customer operations | 30–45% function-wide | High |
| Boston Consulting Group (BCG) — "How Generative AI Is Already Transforming Customer Service" | 2023 | Generative AI agent-assist and conversational/self-service automation in customer service operations | 30–50% function-wide | Med |
| McKinsey & Company (Operations Practice) — "The next frontier of customer engagement: AI-enabled customer service" | 2023 | AI-led self-service, virtual agents and agent assist reducing live service interactions | 20–40% broad | Med |
| 14 Executive leadership & general management | ||||
| McKinsey Global Institute — The economic potential of generative AI: The next productivity frontier | 2023 | Agentic management and decision support automating management/people-management activities | ~25–40% broad | Med |
| The Hackett Group — Generative AI Will Drive Profound Reductions in SG&A Cost and Staffing (Up to 40% Over 5–7 Years) | 2023 | Hyper-automation plus digital assistants across general-management and administrative work (de-layering) | ~25–40% broad | Med |
| Goldman Sachs Global Economics Analyst — The Potentially Large Effects of Artificial Intelligence on Economic Growth (Briggs/Kodnani) | 2023 | Task automation and decision support across management occupations | ~20–32% broad | Med |