AI roles for every department: the 2026 automation analysis
The defining shift of 2025-2026 is the move from assistance to agentic execution: AI agents that plan and execute multi-step workflows autonomously across multiple systems. Capability is no longer the bottleneck; workflow redesign and governance are.
What changed since 2024: from ai assistance to agentic execution
The 2024 edition was written in an era when AI assisted with tasks. Five structural shifts underpin every department update below.
Salesforce Agentforce, ServiceNow AI agents, and orchestration platforms (n8n, CrewAI, LangGraph, Lindy, Relevance AI, Zapier Agents) now execute end-to-end processes rather than single tasks.
OpenAI Operator was absorbed into ChatGPT Agent in mid-2025; Claude Computer Use and others navigate GUIs. Still unreliable for high-stakes commerce, they augment rather than replace.
GPT-5.x, Claude Opus 4.x, and Gemini 3.x dramatically improved multi-step reasoning, commoditizing capabilities that were vertical-specific in 2024.
Devin, Claude Code, OpenAI Codex, Cursor, and GitHub Copilot agent mode shifted software work from pair-programming to delegating whole tasks.
Sub-500ms latency made voice agents (Synthflow, Retell, Vapi, Bland, Sierra) production-grade for call centers.
How to read these numbers
There are two numbers per role and the gap between them is the whole point. The tier below measures what AI can produce with a person reviewing it. It does not measure how much runs unattended, and it is not a prediction that the job disappears. WEF Future of Jobs Report 2025: respondents estimate 47% of tasks are primarily human, 22% primarily technology, and 30% hybrid today, shifting toward an even split by 2030. Projected disruption equals 22% of jobs by 2030, with 170M new roles created and 92M displaced (net +78M). Augmentation dominates over replacement in most roles.
The second number, in the tables further down, is the autonomy ceiling: the share that can run with nobody watching. It is lower, often much lower, wherever the last step sends, spends, publishes, signs, grants access, or files with a regulator. Accounts payable is a clean example. AI can do 70-85% of the clerk's work, and the ceiling is 50%, because releasing the payment is irreversible and waits for a person. Anyone quoting you a single automation percentage is answering two different questions with one number.
AI can produce most of the work at acceptable quality with a person reviewing. How much runs unattended is lower wherever the final step sends, spends, publishes or signs.
AI handles substantial chunks of the work. Humans direct it and finalise the output.
AI accelerates and supports. Humans do the core work and own the decisions.
Predominantly human judgment, relationships and accountability. Automating here does not pay.
† Estimate. Percentages marked with a dagger are structured estimates based on the current tool landscape. Unmarked percentages cite a published source (McKinsey, WEF, Gartner, Ardent Partners, or vendor documentation); vendor-reported resolution rates vary widely by deployment quality.
10 departments · 212 role analyses
210 distinct roles graded here (212 analyses, because a few roles are scored under two departments). Separate from that, our operating catalog maps 199 distinct corporate roles. Each analysis lists its automation tier, the tools doing the work in production today, and what stays human. Percentages without a cited source are structured estimates. Expand a department to see the full table.
Executive & Strategy24 roles analyzed19122
What changed since 2024: Reasoning models plus deep-research agents now produce first-draft market analyses, competitive teardowns, and board decks in minutes. McKinsey notes strategy and corporate finance is among the functions where high performers are scaling agents. But executive accountability, stakeholder trust, and final judgment remain irreducibly human.
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| CEO | Human-led10-15%† | Not rated | ChatGPT AgentClaudeGemini Deep Research | Judgment, vision, accountability not automatable. AI augments synthesis and prep. |
| Chief Strategy Officer | AI-assisted20-30%† | Not rated | Deep research agentsAlphaSenseScenario modeling | AI drafts analyses; strategic choices and trade-offs stay human. |
| COO | AI-assisted20-30%† | Not rated | Agentforce OperationsProcess mining + agents | Back-office orchestration increasingly agentic; cross-functional judgment human. |
| Chief of Staff | AI-assisted35-45%† | 35% Partly reversible. A person approves comms to staff and board. | LindyChatGPT AgentMeeting-note agents | Scheduling, briefing prep, follow-up tracking now largely automatable. |
| Corporate Strategy Analyst | Substantial AI capability45-55%† | 55% A person approves publish to leadership. | Deep research agentsAlphaSenseHebbia | First-draft research and modeling automatable; framing and recommendations human. |
| Market Intelligence Lead | Substantial AI capability50-60%† | Not rated | AlphaSensePerplexityCrayonDeep research agents | Continuous competitive monitoring now agentic; interpretation human. |
| Executive Assistant | Substantial AI capability60-70%† | Not rated | LindyChatGPT AgentScheduling agents | Scheduling, inbox triage, travel, drafting heavily automated. |
| Investor Relations Manager | AI-assisted25-35%† | Not rated | AlphaSenseEarnings-call summarizersDrafting copilots | Drafting and data prep automatable; relationship and disclosure judgment human. |
| ESG & DEI Officer | AI-assisted25-35%† | Not rated | ESG data agentsReporting copilots | Data aggregation and reporting automatable; policy and culture human. |
| Business Analyst / Strategy Associate | Substantial AI capability50-60%† | Not rated | ChatGPT AgentHebbiaExcel/Sheets copilots | Modeling, research, deck-building heavily augmented. |
| Chief Revenue Officer (CRO) | Human-led12-18%† | Not rated | GongClariRevenue copilots | Forecast synthesis and pipeline analysis augmented; revenue accountability and exec relationships human. |
| Chief Growth Officer | AI-assisted20-30%† | Not rated | Deep research agentsExperimentation platformsMixpanel/Amplitude AI | Growth-model analysis and experiment design augmented; bets and trade-offs human. |
| Chief Innovation Officer | AI-assisted20-30%† | Not rated | Deep research agentsPatent/landscape AIIdea-management platforms | Trend scanning and tech scouting augmented; portfolio bets and culture human. |
| Chief Sustainability Officer | AI-assisted25-35%† | Not rated | ESG reporting agentsCarbon-accounting platformsWatershed/Persefoni | Disclosure data collection and reporting increasingly automated; strategy and stakeholder negotiation human. |
| Chief Communications Officer | AI-assisted25-35%† | Not rated | Drafting copilotsMedia-monitoring AICrisis-simulation agents | Drafting, monitoring, and message testing augmented; reputation judgment and crisis calls human. |
| Chief AI Officer (CAIO) | AI-assisted25-35%† | Not rated | Agent orchestration platformsModel eval suitesGovernance tooling | Capability scanning and rollout tracking augmented; AI strategy, governance, and accountability human. |
| VP of Corporate Development / M&A | Substantial AI capability40-50%† | Not rated | HebbiaAlphaSenseDiligence agentsExcel copilots | Sourcing, screening, and diligence document review heavily augmented; deal negotiation and structuring human. |
| Strategic Planning Director | Substantial AI capability40-50%† | Not rated | Scenario-modeling agentsChatGPT AgentPlanning copilots | Plan drafting, scenario modeling, and OKR rollups augmented; prioritization and alignment human. |
| Corporate Secretary / Board Relations | Substantial AI capability45-55%† | Not rated | Diligent BoardsMinute-taking AIDocument agents | Board-pack assembly, minutes, and filings automatable; governance judgment and confidentiality human. |
| Competitive Intelligence Analyst | High AI capability60-70%† | 70% A person approves none for internal brief. | AlphaSenseGemini Deep ResearchCrayonWeb-monitoring agents | Continuous competitor monitoring, teardown drafting, and briefing generation largely automated; synthesis reviewed by humans. |
| Business Operations Manager (BizOps) | Substantial AI capability45-55%† | Not rated | ChatGPT AgentProcess-mining agentsBI copilots | Reporting, process documentation, and cross-functional analysis heavily augmented; prioritization and stakeholder work human. |
| Transformation Program Director | AI-assisted30-40%† | Not rated | Program-management copilotsStatus-rollup agents | Status synthesis and risk tracking automated; change management and executive alignment human. |
| Partnerships & Alliances Director | AI-assisted25-35%† | Not rated | CRM copilotsResearch agentsOutreach drafting | Partner research, briefing prep, and follow-up drafting augmented; relationship-building and negotiation human. |
| Enterprise PMO Director | Substantial AI capability40-50%† | Not rated | Project copilotsPortfolio dashboardsReporting agents | Portfolio reporting, dependency tracking, and status decks automatable; governance and trade-off calls human. |
Summary: Mostly AI-assisted, with substantial capability at the analyst and EA layer. Strategy work is being compressed at the analyst level while leadership stays human-led.
Finance & Accounting16 roles analyzed5722
What changed since 2024: One of the fastest-moving departments. Agentic finance arrived: BILL AI agents (Oct 2025), Ramp Agents for Controllers and AP (July 2025), Tipalti context-aware agents. FP&A copilots (Datarails FP&A Genius, Pigment, Vena Copilot, Cube) answer natural-language questions and draft variance analyses. Ardent Partners 2025: Best-in-Class AP teams achieve 49.2% touchless processing at $2.78 per invoice in 3.1 days, versus $12.88 and 17.4 days for others.
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| CFO | Human-led15-20%† | Not rated | FP&A copilotsBoard-deck agents | Capital allocation, investor trust, judgment human. |
| Financial Controller | Substantial AI capability40-50%† | Not rated | BlackLineRamp Agents for ControllersClose automation | Reconciliations and controls increasingly agentic; sign-off human. |
| FP&A Analyst | Substantial AI capability50-60%† | Not rated | Datarails FP&A GeniusPigmentVena CopilotCube | NL queries, variance analysis, forecasting drafts automated; narrative and strategy human. |
| Accounting Manager | Substantial AI capability40-50%† | Not rated | Vic.aiBILL AIClose-automation agents | Workflow orchestration automated; review and exceptions human. |
| Accounts Payable Clerk | High AI capability70-85% Ardent Partners 2025: Best-in-Class 49.2% touchless; vendor claims 60-89% | 50% Irreversible. A person approves payment release. | Vic.aiTipaltiBILL AI (W-9 Agent) | BILL reports >80% increase in fully automated bills since early 2025. Touchless processing the norm for high performers; exceptions human. |
| Accounts Receivable Specialist | High AI capability65-75%† | 40% Irreversible. A person approves escalation to collections. | Collections agentsBILLHighRadius | Dunning, matching, reminders automated; disputes human. |
| Payroll Specialist | Substantial AI capability50-60%† | 25% Irreversible. A person approves pay-run approval. | ADP AIUKG AIGusto | Calculation and filing automated; exceptions and compliance human. |
| Bookkeeper | High AI capability70-80%† | 65% A person approves ledger posting. | Vic.aiQuickBooks AIPuzzle | Categorization and reconciliation heavily automated. |
| Treasury Analyst | Substantial AI capability45-55%† | 55% A person approves investment and hedge actions. | Tipalti (Statement)Real-time cash agents | Cash positioning and forecasting automated; liquidity strategy human. |
| Audit & Compliance Analyst | Substantial AI capability40-50%† | Not rated | MindBridgeAgentic audit tools | Sampling and anomaly detection automated; judgment and sign-off human. |
| Tax Specialist | AI-assisted35-45%† | Not rated | CoCounsel (Thomson Reuters)Blue JTax copilots | Research and prep automated; planning and positions human. |
| Procurement Analyst / Manager | Substantial AI capability45-55%† | Not rated | Ramp procurement agentsTipaltiZip | Intake, due diligence, renewals increasingly agentic. Ramp claims ~16% annual savings and 46 hrs/month saved. |
| Finance Systems Administrator | AI-assisted30-40%† | Not rated | ERP copilotsIntegration agents | Config and reporting augmented; architecture human. |
| VP of Finance | Human-led15-22%† | Not rated | FP&A copilotsBoard-reporting agents | Reporting and variance analysis augmented; capital decisions and exec partnership human. |
| Cost Accountant | High AI capability60-70%† | 40% A person approves standard-cost changes. | ERP cost modulesReconciliation agentsExcel copilots | Cost allocation, variance calc, and reconciliation largely automatable; method judgment human. |
| Billing / Revenue Operations Analyst | High AI capability62-72%† | Not rated | Stripe BillingZuoraRevRec agents | Invoice generation, dunning, and revenue recognition heavily automated; exceptions and policy human. |
Summary: Transactional roles (AP, AR, bookkeeping) show the highest AI capability of any department, but the lowest ratio of capability to autonomy: money movement is irreversible, so the payment run itself still waits for a person. Analytical roles (FP&A, treasury, audit) sit mid-range and leadership is human-led.
Marketing & Sales25 roles analyzed51442
What changed since 2024: Content generation is now table stakes; the frontier moved to agentic execution and AI SDRs. HubSpot Breeze and Salesforce Agentforce embed autonomous agents (Customer Agent activated by 8,000+ customers, Prospecting Agent by 10,000+). Reality check: the fully autonomous AI SDR narrative peaked then corrected, the market is converging on human-in-the-loop. Clari merged with Salesloft (Dec 2025).
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| Digital Marketing Manager | Substantial AI capability40-50%† | Not rated | HubSpot BreezeJasper agentsAgentforce Marketing | Campaign orchestration agentic; strategy human. |
| Content Marketing Specialist | High AI capability60-70%† | Not rated | JasperChatGPTCopy.ai | Drafting and repurposing heavily automated; editorial judgment human. |
| SEO Specialist | Substantial AI capability55-65%† | 65% A person approves site push. | Surfer SEOFraseGEO/AEO tooling | AI-driven optimization; strategy and GEO (generative engine optimization) emerging. |
| PPC / Paid Media Analyst | Substantial AI capability50-60%† | Not rated | Google smart biddingMeta AIPecan | Bid management automated; budget strategy human. |
| Marketing Analyst / BI | Substantial AI capability50-60%† | Not rated | ChatGPT AgentBI copilots | Reporting and analysis automated; interpretation human. |
| Email Marketing Specialist | High AI capability60-70%† | 35% Irreversible. A person approves send approval. | HubSpot BreezeKlaviyo AI | Drafting, segmentation, send-time optimization automated. |
| Growth Hacker | Substantial AI capability45-55%† | Not rated | ClayExperimentation agents | Data orchestration automated; creative experiments human. |
| Graphic Designer | Substantial AI capability45-55%† | Not rated | Canva Magic StudioAdobe FireflyMidjourney | Production automated; brand and art direction human. |
| Social Media Manager | Substantial AI capability50-60%† | 25% Irreversible. A person approves post approval. | HubSpot Social AgentJasper | Scheduling and drafting automated; community and crisis human. |
| Sales Development Rep (SDR) | Substantial AI capability50-65% AI SDRs automate 60-80% of SDR tasks per vendor reviews | Not rated | 11xArtisanAiSDRQualified PiperClay | Prospecting and sequencing automatable, but fully autonomous SDR underdelivered; hybrid model winning. |
| Account Executive (AE) | AI-assisted25-35%† | 0% Irreversible. A person approves all commitments and pricing. | GongClariCall-summary agents | Prep and admin automated; relationships and closing human. |
| Customer Success Manager | AI-assisted30-40%† | 0% Irreversible. A person approves all commitments. | Gainsight AIHealth-score agents | Monitoring and outreach automated; relationships human. |
| Head of Marketing / CMO | Human-led15-20% Gartner survey of 402 senior marketing leaders, Aug-Oct 2025: 65% of CMOs say AI will dramatically change the CMO role within two years | Not rated | Strategy copilots | Strategy and brand stewardship human. |
| VP of Sales | Human-led15-22%† | Not rated | GongClariForecast copilots | Forecasting and deal-inspection augmented; coaching, hiring, and big-deal strategy human. |
| Sales Operations Manager (RevOps) | Substantial AI capability50-60%† | Not rated | HubSpot/Salesforce copilotsClayReporting agents | Territory modeling, pipeline hygiene, and reporting heavily automated; process design human. |
| Product Marketing Manager | Substantial AI capability45-55%† | 15% Irreversible. A person approves positioning and launch. | ClaudeJasperCompetitive-intel agents | Messaging drafts, battlecards, and launch collateral augmented; positioning strategy human. |
| Brand Manager | AI-assisted30-40%† | 0% Irreversible. A person approves all brand decisions. | Brand-asset AIMidjourney/FireflySocial-listening AI | Asset variation and monitoring augmented; brand strategy and guardianship human. |
| Field Marketing Manager | Substantial AI capability40-50%† | Not rated | Event copilotsEmail agentsCRM automation | Event logistics, invites, and follow-up automatable; on-the-ground relationships human. |
| Marketing Operations Manager | Substantial AI capability55-65%† | Not rated | Marketo/HubSpot AIn8nAttribution agents | Campaign ops, lead routing, and attribution heavily automated; stack strategy human. |
| Lifecycle / CRM Marketing Manager | High AI capability60-70%† | Not rated | BrazeCustomer.io AISegmentation agents | Journey building, segmentation, and send-time optimization largely automated; strategy and offers human. |
| Affiliate / Partner Marketing Manager | Substantial AI capability45-55%† | Not rated | Partner platformsOutreach agentsReporting copilots | Recruitment outreach, tracking, and payout reconciliation automatable; partner relationships human. |
| Account Manager | AI-assisted30-40%† | Not rated | CRM copilotsGongQBR-deck agents | Account research, QBR prep, and follow-up drafting augmented; relationships and expansion human. |
| Business Development Representative (BDR) | High AI capability60-70%† | Not rated | ClayApolloOutreach agentsAI SDR (e.g. 11x, Artisan) | Sourcing, enrichment, and first-touch sequencing heavily automated; live qualification calls human. |
| Copywriter | High AI capability65-75%† | 40% Irreversible. A person approves publish approval. | ClaudeChatGPTJasperSurfer SEO | First-draft copy across formats largely automated; voice, taste, and final edit human. |
| Video / Multimedia Producer | Substantial AI capability50-60%† | Not rated | HeyGenSynthesiaDescriptRunway | Editing, captioning, avatars, and repurposing heavily augmented; creative direction human. |
Summary: Content roles High; analytical and SDR roles Partial; senior and relationship roles Augmented to Low. The SDR role is the cautionary tale: capability outran reliability.
Customer Service19 roles analyzed5941
What changed since 2024: The most transformed department. Autonomous resolution is real: Intercom Fin 3 reports 67% average resolution (Dec 2025); Salesforce Agentforce claims 85%; Zendesk (post-Forethought) claims 80%+ end-to-end. Voice AI matured, Sierra processes hundreds of millions of calls, and voice surpassed text on its platform by Sept 2025. Consolidation: Moveworks→ServiceNow, Forethought→Zendesk, Aisera→Automation Anywhere, Observe.AI→Snowflake.
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| Customer Support Rep (Tier 1) | High AI capability65-85% Intercom Fin 67% avg resolution; Agentforce 85%; Zendesk 80%+ (vendor-reported) | Not rated | Intercom FinSierraDecagonAdaZendesk AI | Most routine tickets resolved autonomously; humans handle complex and emotional cases. |
| Technical Support Specialist (Tier 2) | Substantial AI capability40-50%† | Not rated | Agent-assist copilotsForethought (Zendesk) | AI drafts and diagnoses; complex troubleshooting human. |
| Live Chat Agent | High AI capability70-85%† | Not rated | Intercom FinSierraDecagon | Largely autonomous for common queries. |
| Call Center Agent | High AI capability60-80% Voice agents resolve 60-87% per vendor data; PolyAI reports 80-87% containment | Not rated | SierraRetellVapiBlandPolyAI | Voice agents now production-grade. |
| Help Desk Coordinator | Substantial AI capability50-60%† | Not rated | ServiceNow agentsMoveworks (ServiceNow) | Routing and triage automated. |
| Knowledge Base Manager | Substantial AI capability50-60%† | 60% A person approves publish to KB. | Fin FlywheelContent agents | Gap detection and drafting automated; curation human. |
| Customer Experience Analyst | Substantial AI capability45-55%† | Not rated | Observe.AI (Snowflake)Analytics agents | Conversation analytics automated; insight human. |
| Community Manager | AI-assisted30-40%† | 10% Irreversible. A person approves public replies. | Social agentsModeration AI | Moderation and drafting automated; community building human. |
| Customer Escalations Specialist | AI-assisted25-35%† | Not rated | Agent-assistSentiment AI | AI surfaces context; resolution and empathy human. |
| Onboarding Specialist | Substantial AI capability45-55%† | 35% A person approves account provisioning. | Onboarding agentsGuided flows | Routine onboarding automated; high-touch human. |
| Customer Success Manager | AI-assisted30-40%† | 0% Irreversible. A person approves all commitments. | Gainsight AIHealth agents | Monitoring automated; relationships human. |
| Call Center QA Analyst | High AI capability65-75%† | Not rated | Observe.AI Auto QACresta | 100% interaction scoring automated; calibration human. |
| Customer Insights Analyst | Substantial AI capability50-60%† | Not rated | VoC agentsAnalytics copilots | Theme extraction automated; strategic insight human. |
| VP of Customer Experience / CXO | Human-led15-20%† | Not rated | DashboardsAgent oversight | Strategy and org design human; emerging agent-supervisor duties. |
| Director of Customer Support | AI-assisted30-40%† | Not rated | Zendesk AIIntercom FinSupport dashboards | Volume forecasting, QA sampling, and reporting automated; team leadership and escalation calls human. |
| Workforce Management Analyst (Contact Center) | High AI capability60-70%† | Not rated | Forecasting agentsAssembledScheduling AI | Volume forecasting, scheduling, and intraday adjustments largely automated; exception calls human. |
| Customer Training Specialist | Substantial AI capability50-60%† | Not rated | SynthesiaCourse-authoring AIKnowledge agents | Tutorial generation, docs, and onboarding content heavily augmented; live enablement human. |
| Support Operations Manager | Substantial AI capability50-60%† | Not rated | Zendesk/Salesforce AIMacro agentsRouting automation | Macro maintenance, routing rules, and reporting automatable; tooling strategy human. |
| Voice of Customer (VoC) Program Manager | Substantial AI capability45-55%† | Not rated | Survey AIThematicSentiment agents | Feedback aggregation, theme extraction, and reporting automated; program strategy and action human. |
Summary: Tier 1, chat, voice and QA show the highest AI capability of any department, and because a support reply can be corrected, more of it genuinely runs unattended than anywhere else. Tier 2, insights and escalations sit mid-range; leadership is human-led.
Operations & Supply Chain20 roles analyzed21062
What changed since 2024: Salesforce launched Agentforce Operations (GA April 2026) to turn manual back-office processes into agent-executed tasks. Demand planning and inventory optimization increasingly run on agentic forecasting. Physical roles remain gated by robotics, not software.
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| Supply Chain Analyst | Substantial AI capability50-60%† | Not rated | o9KinaxisAgentic forecasting | Analysis and scenario planning automated; judgment human. |
| Inventory Manager | Substantial AI capability50-60%† | Not rated | Inventory-optimization agents | Reorder and optimization automated; exceptions human. |
| Logistics Coordinator | Substantial AI capability50-60%† | 25% Irreversible. A person approves carrier bookings. | Route/load agentsAgentforce Operations | Coordination automated; disruptions human. |
| Demand Planner | Substantial AI capability55-65%† | 50% A person approves plan commit. | Kinaxiso9ML forecasting | Forecasting heavily automated; assumptions human. |
| Warehouse Manager | AI-assisted25-35%† | Not rated | WMS AIRobotics | Software planning automated; physical ops gated by robotics. |
| Procurement Manager | Substantial AI capability45-55%† | Not rated | Ramp procurement agentsZip | Intake, due diligence, renewals agentic. |
| Operations Analyst | Substantial AI capability50-60%† | Not rated | Process mining + agents | Analysis automated; process redesign human. |
| Fleet Manager | AI-assisted30-40%† | Not rated | Telematics AIRouting AI | Optimization automated; physical and safety human. |
| Supplier Relationship Manager | AI-assisted25-35%† | 0% Irreversible. A person approves all supplier terms. | Risk agentsScorecards | Monitoring automated; negotiation and relationships human. |
| Quality Control Specialist | Substantial AI capability40-50%† | Not rated | Computer vision QC | Visual inspection automated; root-cause human. |
| Sustainability / ESG Officer | AI-assisted25-35%† | Not rated | ESG reporting agents | Data aggregation automated; strategy human. |
| Global Trade Compliance Manager | Substantial AI capability40-50%† | Not rated | Trade-compliance agents | Classification and screening automated; judgment human. |
| Production Scheduler | Substantial AI capability50-60%† | Not rated | APSScheduling agents | Scheduling automated; exceptions human. |
| Order Fulfillment Specialist | High AI capability60-70%† | Not rated | Agentforce OperationsOMS AI | Order orchestration heavily automated. |
| Facilities Manager | AI-assisted25-35%† | Not rated | IoT/building AI | Monitoring automated; physical human. |
| Operations Executive / VP | Human-led15-20%† | Not rated | Operations dashboardsAgents | Strategy and accountability human. |
| Director of Operations | Human-led18-25%† | Not rated | Process-mining agentsOps dashboardsChatGPT Agent | Reporting, process docs, and bottleneck analysis augmented; org decisions and accountability human. |
| Continuous Improvement (Lean/Six Sigma) Manager | AI-assisted30-40%† | Not rated | Process-mining (Celonis)Analysis copilots | Process discovery, root-cause analysis, and documentation augmented; change leadership human. |
| Transportation / Dispatch Manager | Substantial AI capability50-60%† | Not rated | Route-optimization AITMS agentsTelematics | Routing, load planning, and dispatch heavily automated; exceptions and carrier relationships human. |
| Returns & Reverse Logistics Coordinator | High AI capability60-70%† | Not rated | Returns-automation platformsRMA agentsWMS | RMA processing, routing, and disposition rules largely automated; fraud and exceptions human. |
Summary: Planning, analytical, and fulfillment roles Partial-to-High; physical roles Augmented (gated by robotics, not AI); leadership low.
Product Development & Data19 roles analyzed11251
What changed since 2024: Autonomous coding agents transformed software work. Devin, Claude Code, OpenAI Codex, Cursor, and GitHub Copilot agent mode now run multi-step tasks and open PRs. McKinsey reports software engineering shows 10-20% cost reductions and is a top function for scaled agents.
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| Data Analyst | Substantial AI capability55-65%† | 65% A person approves publish metric. | ChatGPT AgentBI copilotsHebbia | NL-to-query and analysis automated; framing human. |
| Data Scientist | Substantial AI capability40-50%† | 40% A person approves model deploy. | DataRobot (agentic)Notebooks + Copilot | Model building accelerated; problem framing human. |
| ML Engineer | Substantial AI capability40-50%† | 45% A person approves prod deploy. | Coding agentsMLOps copilots | Pipeline code automated; architecture human. |
| BI Developer | Substantial AI capability50-60%† | 60% A person approves publish dashboard. | Power BI CopilotTableau AI | Dashboard generation automated. |
| UX/UI Designer | AI-assisted30-40%† | Not rated | Figma AIv0Galileo | Prototyping automated; research and judgment human. |
| Product Manager | AI-assisted25-35%† | 5% Irreversible. A person approves all prioritization and launch. | PM copilotsResearch agents | Docs and synthesis automated; prioritization and vision human. |
| Product Owner / Technical PM | AI-assisted30-40%† | Not rated | Backlog agentsJira AI | Story-writing automated; trade-offs human. |
| Quality Assurance Analyst | Substantial AI capability55-65%† | Not rated | Test-gen agentsCodex | Test generation and execution automated; strategy human. |
| Software Engineer | Substantial AI capability45-60% McKinsey 2025: 10-20% cost reduction in software engineering | 55% A person approves merge to main. | Claude CodeCursorGitHub CopilotDevinCodex | Routine coding, refactors, tests delegated; architecture and review human. |
| CTO / VP of Product | Human-led15-20%† | Not rated | Strategy/portfolio copilots | Technical strategy and accountability human. |
| Data Engineer | Substantial AI capability50-60%† | 50% A person approves prod deploy. | dbtClaude CodePipeline copilotsSnowflake Cortex | Pipeline scaffolding, transformations, and tests heavily augmented; architecture and reliability human. |
| Analytics Engineer | High AI capability60-70%† | 55% A person approves merge to prod. | dbt CopilotChatGPT AgentSemantic-layer AI | Model building, documentation, and metric definitions largely automated; modeling judgment human. |
| Data Governance Manager | AI-assisted35-45%† | Not rated | Catalog AI (Collibra/Atlan)Classification agents | Cataloging, lineage, and PII classification automated; policy and stewardship human. |
| UX Researcher | Substantial AI capability45-55%† | 60% A person approves publish findings. | Research-synthesis AI (Dovetail)Transcription AISurvey agents | Transcript synthesis, theme tagging, and reporting automated; study design and interpretation human. |
| Product Designer | Substantial AI capability45-55%† | Not rated | Figma AIv0UX PilotImage models | Wireframes, variations, and prototypes heavily augmented; craft, taste, and flows human. |
| Scrum Master / Agile Coach | Substantial AI capability40-50%† | Not rated | Jira AIStandup-summary agentsReporting copilots | Ceremony admin, status rollups, and metrics automated; facilitation and coaching human. |
| Technical Program Manager | Substantial AI capability40-50%† | Not rated | Project copilotsDependency-tracking agentsStatus agents | Cross-team status, dependency tracking, and docs automatable; alignment and escalation human. |
| Research Scientist (R&D) | AI-assisted30-40%† | Not rated | Gemini Deep ResearchLiterature agentsCode copilots | Literature review, experiment scaffolding, and analysis augmented; novel hypotheses and judgment human. |
| Growth Product Manager | Substantial AI capability45-55%† | Not rated | Amplitude/Mixpanel AIExperimentation agentsSQL copilots | Funnel analysis, experiment setup, and reporting heavily augmented; prioritization and bets human. |
Summary: Coding, QA, and data analysis roles Partial (rising fast); design and PM Augmented; leadership low. Software engineering is being restructured around managing AI engineers.
Legal & Compliance19 roles analyzed31042
What changed since 2024: Legal AI exploded. Harvey reached an $11B valuation (March 2026) with 25,000+ custom agents executing M&A, due diligence, and document review end-to-end; CoCounsel (Thomson Reuters) evolved into integrated workflows; Ironclad launched agentic contract review (March 2026; Intake Agent cut average submission time 50%). Frontier reasoning models commoditized legal reasoning.
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| Contract Manager | Substantial AI capability50-60%† | 35% Irreversible. A person approves execution and renewal. | Ironclad AIHarveyCoCounsel | Drafting, review, redlining agentic; negotiation strategy human. |
| Paralegal | Substantial AI capability55-65%† | 60% A person approves filing and service. | CoCounselHarveyDocument agents | Research, review, drafting heavily automated. |
| Compliance Analyst | Substantial AI capability45-55%† | Not rated | Compliance agentsMonitoring AI | Monitoring and screening automated; judgment human. |
| Internal Auditor | Substantial AI capability40-50%† | 45% A person approves audit opinion. | Agentic auditMindBridge | Sampling and testing automated; sign-off human. |
| Legal Operations Manager | AI-assisted30-40%† | Not rated | Legal-ops copilotsSpend agents | Process and spend automated; vendor strategy human. |
| Chief Compliance Officer | Human-led15-20%† | Not rated | Compliance dashboards | Accountability and judgment human. |
| General Counsel | Human-led15-20%† | Not rated | HarveyStrategy copilots | Strategy, risk, accountability human. |
| Regulatory Affairs Specialist | AI-assisted35-45%† | 10% Irreversible. A person approves all submissions. | Regulatory research agents | Tracking and drafting automated; strategy human. |
| Corporate Counsel | AI-assisted30-40%† | Not rated | Thomson Reuters CoCounselHarveyDrafting copilots | Research, first-draft agreements, and memos augmented; advice, risk, and accountability human. |
| Litigation Counsel / Manager | Substantial AI capability40-50%† | Not rated | eDiscovery AI (Relativity aiR)CoCounselDeposition-summary AI | Document review, deposition summaries, and research heavily automated; strategy and advocacy human. |
| Data Privacy Officer (DPO) | Substantial AI capability40-50%† | Not rated | OneTrustPrivacy-mapping agentsDSAR automation | Data mapping, DSAR fulfillment, and assessments automatable; regulatory judgment human. |
| Contracts Analyst / CLM Specialist | High AI capability60-70%† | Not rated | Ironclad AICoCounselClause-extraction agents | Clause extraction, redlining against playbooks, and metadata heavily automated; escalations human. |
| Corporate Governance Manager | Substantial AI capability45-55%† | Not rated | Entity-management AIFiling agentsBoard tooling | Entity records, filings, and compliance calendars automated; governance judgment human. |
| Patent / IP Counsel | Substantial AI capability40-50%† | Not rated | Patent-search AIDrafting copilotsPrior-art agents | Prior-art search and draft claims augmented; prosecution strategy and filings human. |
| IP / Trademark Paralegal | High AI capability60-70%† | Not rated | Docketing AISearch agentsFiling automation | Docketing, searches, and renewal filings largely automated; judgment calls human. |
| Ethics & Compliance Officer | AI-assisted25-35%† | Not rated | Case-management AIPolicy agentsTraining automation | Policy drafting, training rollout, and case intake automated; investigations and ethics human. |
| Regulatory Compliance Manager | Substantial AI capability40-50%† | Not rated | Reg-change monitoringDrata/VantaReporting agents | Regulatory tracking, control mapping, and reporting automatable; interpretation and strategy human. |
| Legal Intake / Knowledge Coordinator | High AI capability60-70%† | Not rated | Intake agentsKnowledge AITriage copilots | Matter intake, triage, and knowledge retrieval largely automated; prioritization human. |
| Privacy / GDPR Compliance Analyst | Substantial AI capability50-60%† | Not rated | OneTrustConsent agentsAssessment automation | Consent tracking, RoPA maintenance, and assessments heavily automated; regulatory calls human. |
Summary: Document-heavy roles (paralegal, contract manager, compliance analyst) Partial-to-High; senior counsel low. The associate/paralegal layer is most compressed.
People & HR25 roles analyzed61153
What changed since 2024: High-volume recruiting automation matured (Paradox/Olivia acquired by Workday, Oct 2025; customers like Chipotle report 75% faster hiring). HR helpdesk shifted to agents (Moveworks→ServiceNow). WEF notes 85% of employers plan to upskill workers, with AI expected to augment roughly 50% of tasks in affected industries.
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| Talent Acquisition (Recruiter, Sourcer) | Substantial AI capability50-60%† | Not rated | Paradox/Olivia (Workday)EightfoldSeekOut | Sourcing, screening, scheduling automated; judgment human. |
| Employee Onboarding (HR Coordinator) | Substantial AI capability55-65%† | Not rated | Onboarding agentsServiceNow | Document and workflow automation heavy. |
| HR Helpdesk & Support (HR Generalist) | High AI capability60-70%† | Not rated | ServiceNow AI agentsMoveworks (ServiceNow) | Routine HR queries autonomously resolved. |
| Workforce Planning (HR Analyst) | Substantial AI capability45-55%† | Not rated | Planning copilotsAnalytics agents | Modeling automated; decisions human. |
| Benefits Administration (Benefits Manager, HRIS) | Substantial AI capability50-60%† | Not rated | Benefits agentsHRIS AI | Enrollment and queries automated; design human. |
| Performance Management (People Ops) | AI-assisted30-40%† | Not rated | Review copilotsFeedback AI | Drafting and synthesis automated; judgment human. |
| Learning & Development (L&D Specialist) | Substantial AI capability45-55%† | Not rated | Content-gen agentsLMS AI | Course creation automated; strategy human. |
| Employee Engagement (People Analyst) | Substantial AI capability45-55%† | Not rated | Survey analytics agents | Analysis automated; action human. |
| DEI & Culture (DEI Officer) | AI-assisted25-35%† | Not rated | AnalyticsBias-detection AI | Measurement automated; culture human. |
| Time Tracking & Attendance (HR Admin) | High AI capability60-70%† | Not rated | UKG AIADP AI | Tracking and exceptions heavily automated. |
| Exit Interviews & Offboarding (HRBP) | Substantial AI capability40-50%† | Not rated | Offboarding agentsSentiment AI | Process automated; sensitive conversations human. |
| Chief Human Resources Officer (CHRO) | Human-led12-18%† | Not rated | People-analytics copilotsBoard-reporting agents | Workforce analytics and reporting augmented; org strategy, culture, and accountability human. |
| VP of People | Human-led18-25%† | Not rated | Workday AIPeople dashboardsChatGPT Agent | Reporting, policy drafting, and headcount analysis augmented; leadership and judgment human. |
| Senior HR Business Partner | AI-assisted25-35%† | Not rated | Case copilotsPolicy agentsSentiment AI | Case research, policy lookup, and prep augmented; coaching and sensitive decisions human. |
| Compensation Analyst | High AI capability60-70%† | 20% Irreversible. A person approves all comp changes. | Comp benchmarking AIPaveExcel copilots | Benchmarking, banding, and modeling largely automated; philosophy and exceptions human. |
| Total Rewards Manager | Substantial AI capability45-55%† | Not rated | Benefits agentsComp modeling AIReporting copilots | Plan modeling, communications, and analysis automated; design and negotiation human. |
| Recruiting Coordinator | High AI capability65-75%† | 55% Partly reversible. A person approves offer and rejection. | Scheduling AI (Paradox)ATS automationComms agents | Scheduling, candidate comms, and logistics heavily automated; candidate experience touch human. |
| Technical Sourcer | High AI capability65-75%† | Not rated | LinkedIn Recruiter AISourcing agentsOutreach automation | Search, list-building, and first-touch outreach largely automated; engagement and assessment human. |
| Talent Development Manager | Substantial AI capability40-50%† | Not rated | Skill-mapping AILMS copilotsCareer-path agents | Skills mapping, content curation, and pathing augmented; coaching and programs human. |
| Employee Relations Specialist | Human-led20-30%† | 0% Irreversible. A person approves all ER decisions. | Case-management AIPolicy retrievalDocumentation agents | Case documentation and policy lookup augmented; investigations and sensitive conversations human. |
| Organizational Development Consultant | AI-assisted25-35%† | Not rated | Survey AIOrg-analytics copilots | Diagnostics, survey analysis, and material drafting augmented; intervention design human. |
| Workplace Experience Manager | Substantial AI capability40-50%† | Not rated | Ticketing AIVendor agentsEvent copilots | Requests, vendor coordination, and events automatable; in-person culture work human. |
| HRIS Administrator | Substantial AI capability50-60%† | 30% Irreversible. A person approves config and permission changes. | Workday/BambooHR AIIntegration agentsReporting copilots | Config, data hygiene, and reporting heavily automated; system strategy human. |
| Instructional Designer | High AI capability60-70%† | Not rated | SynthesiaCourse-authoring AIAssessment agents | Course drafting, assessments, and media generation largely automated; learning design human. |
| Diversity Recruiting Partner | AI-assisted30-40%† | Not rated | Sourcing agentsBias-check AIPipeline analytics | Pipeline sourcing and bias auditing augmented; partnerships and strategy human. |
Summary: HR helpdesk, time tracking, and onboarding High-to-Partial; relationship and culture roles Augmented. High-volume transactional HR is heavily automated.
Technology & IT25 roles analyzed21643
What changed since 2024: Agentic IT and security operations arrived. ServiceNow + Moveworks deliver autonomous IT service desk; Microsoft Security Copilot shipped autonomous SOC agents (Security Alert Triage Agent GA Nov 2025, identifies 6.5x more malicious alerts). Gartner: 35% of security leaders expect AI to replace Tier-1 SOC analysts within three years.
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| IT Support Specialist | High AI capability60-70%† | Not rated | ServiceNow agentsMoveworks (ServiceNow) | Routine tickets autonomously resolved. |
| Help Desk Technician | High AI capability60-70%† | 50% Partly reversible. A person approves access grants. | Moveworks (ServiceNow)ServiceNow | Password resets, provisioning automated. |
| System Administrator | Substantial AI capability40-50%† | Not rated | Infra copilotsAutomation agents | Routine admin automated; architecture human. |
| DevOps Engineer | Substantial AI capability45-55%† | 20% Irreversible. A person approves all prod changes. | Coding agentsCI/CD copilots | Pipeline automation heavy; design human. |
| Network Engineer | AI-assisted30-40%† | 5% Irreversible. A person approves all changes. | Network AI agentsMonitoring | Diagnostics automated; design human. |
| Cloud Engineer | Substantial AI capability40-50%† | 15% Irreversible. A person approves all resource changes. | IaC agentsCloud copilots | Provisioning automated; architecture human. |
| Cybersecurity Analyst | Substantial AI capability45-55%† | Not rated | Microsoft Security CopilotCrowdStrike Charlotte | Triage and investigation increasingly agentic. |
| SOC Analyst | Substantial AI capability50-65% Microsoft 2025: phishing triage agent identifies 6.5x more malicious alerts | 55% A person approves containment actions. | Microsoft Security CopilotGoogle SecOps GeminiPalo Alto Cortex AgentiX | Tier-1 triage being automated; investigation supervision human. |
| IT Project Manager | AI-assisted25-35%† | Not rated | PM copilots | Admin automated; stakeholder management human. |
| Infrastructure Engineer | Substantial AI capability40-50%† | Not rated | IaC agents | Provisioning automated; design human. |
| Database Administrator (DBA) | Substantial AI capability40-50%† | 10% Irreversible. A person approves all schema and data changes. | Query optimization AICopilots | Tuning automated; architecture human. |
| IT Asset Manager | Substantial AI capability45-55%† | Not rated | Asset agentsDiscovery AI | Tracking and discovery automated. |
| CIO | Human-led15-20%† | Not rated | Strategy copilots | Strategy and accountability human. |
| IT Compliance Officer | Substantial AI capability40-50%† | Not rated | Compliance agents | Evidence collection automated; judgment human. |
| Identity and Access Manager | Substantial AI capability45-55%† | Not rated | Entra CopilotLifecycle agents | Joiner/mover/leaver automation; policy human. |
| AI/ML Ops Engineer | Substantial AI capability40-50%† | Not rated | MLOps copilotsMonitoring agents | Pipeline and monitoring automated; design human. |
| VP of Engineering | Human-led18-25%† | Not rated | Eng-metrics copilotsDelivery dashboards | Delivery metrics and reporting augmented; technical strategy, hiring, and accountability human. |
| Chief Information Security Officer (CISO) | Human-led15-22%† | Not rated | SOC copilotsRisk dashboardsGRC agents | Posture reporting and risk aggregation augmented; security strategy and accountability human. |
| Software Engineering Manager | AI-assisted30-40%† | Not rated | PR-summary agentsSprint copilotsCode-review AI | Review triage, status, and reporting augmented; people management and decisions human. |
| Frontend Developer | Substantial AI capability50-60%† | Not rated | CursorGitHub Copilotv0Claude Code | Component scaffolding, styling, and tests heavily augmented; UX judgment and architecture human. |
| Backend Developer | Substantial AI capability50-60%† | Not rated | Claude CodeCursorGitHub Copilot agent mode | Endpoints, refactors, and tests delegated; system design and review human. |
| Full-Stack Developer | Substantial AI capability50-60%† | Not rated | CursorClaude CodeOpenAI Codex | Feature scaffolding across the stack heavily augmented; architecture and integration calls human. |
| Site Reliability Engineer (SRE) | Substantial AI capability45-55%† | Not rated | Incident-response AIObservability copilots (Datadog Bits)Runbook agents | Alert triage, runbook execution, and postmortem drafts automated; reliability strategy human. |
| Solutions Architect | AI-assisted35-45%† | Not rated | Diagramming AIDesign copilotsResearch agents | Reference designs, diagrams, and docs augmented; trade-off decisions and stakeholder work human. |
| Application Security Engineer | Substantial AI capability45-55%† | Not rated | SAST/DAST AISnykSecure-code copilots | Scanning, triage, and remediation guidance heavily automated; threat modeling and judgment human. |
Summary: Help desk and Tier-1 SOC roles High-to-Partial (the deepest change); engineering Partial; architecture and leadership Augmented-to-Low.
Risk & Audit20 roles analyzed21053
What changed since 2024: Agentic audit and risk monitoring emerged; Salesforce cites audit work that took a team of auditors four hours now taking an AI agent minutes, with a full audit trail. Reasoning models improved anomaly detection and control testing. Accountability and regulatory judgment remain human.
| Role | AI can do, with review | Can run without you | AI tools / technologies | Notes |
|---|---|---|---|---|
| Internal Auditor | Substantial AI capability40-50%† | 45% A person approves audit opinion. | MindBridgeAgentic audit | Testing and sampling automated; sign-off human. |
| Risk Manager | AI-assisted30-40%† | Not rated | Risk-monitoring agents | Monitoring automated; judgment human. |
| Third-Party Risk Analyst | Substantial AI capability45-55%† | Not rated | Vendor-risk agents | Due diligence and screening automated. |
| Fraud Investigator | Substantial AI capability45-55%† | Not rated | Fraud-detection AI | Detection automated (BILL stopped 8M fraud attempts in FY25); investigation human. |
| Compliance Risk Analyst | Substantial AI capability45-55%† | Not rated | Compliance agents | Monitoring and screening automated. |
| SOX Compliance Specialist | Substantial AI capability40-50%† | Not rated | Controls-testing agents | Control testing automated; judgment human. |
| ERM Program Manager | AI-assisted25-35%† | Not rated | Risk dashboardsAgents | Aggregation automated; strategy human. |
| Ethics & Whistleblower Officer | Human-led15-25%† | Not rated | Case-management AI | Intake and triage automated; investigation and ethics human. |
| Business Continuity Planner | AI-assisted30-40%† | Not rated | Scenario/planning agents | Scenario modeling automated; decisions human. |
| Chief Risk Officer (CRO) | Human-led15-22%† | Not rated | Risk dashboardsERM copilotsScenario agents | Risk aggregation and reporting augmented; risk appetite and board accountability human. |
| Chief Audit Executive (CAE) | Human-led15-22%† | Not rated | Audit-analytics copilotsReporting agents | Audit-universe analytics and reporting augmented; audit strategy and independence human. |
| VP of Internal Audit | AI-assisted25-35%† | Not rated | Audit-management AI (AuditBoard)Analytics agents | Planning analytics, workpaper review, and reporting augmented; judgment and sign-off human. |
| Audit Manager | Substantial AI capability45-55%† | Not rated | AuditBoard AISampling agentsWorkpaper copilots | Sampling, testing, and workpaper drafting heavily automated; opinions and judgment human. |
| IT / Cyber Risk Analyst | Substantial AI capability45-55%† | Not rated | GRC agents (Drata/Vanta)Vuln-scanning AI | Control mapping, evidence collection, and scanning automated; risk interpretation human. |
| Operational Risk Analyst | Substantial AI capability45-55%† | 40% A person approves risk decisions. | Risk-monitoring agentsRCSA copilots | Risk-register maintenance, KRI monitoring, and reporting automated; assessment human. |
| Credit Risk Analyst | High AI capability60-70%† | Not rated | Credit-scoring AIFinancial-spreading agents | Spreading, scoring, and monitoring largely automated; exposure decisions human. |
| Model Risk / Risk Modeling Analyst | Substantial AI capability45-55%† | Not rated | Model-validation copilotsBacktesting agents | Validation testing, documentation, and backtesting automated; methodology judgment human. |
| Insurance / Captive Risk Manager | AI-assisted30-40%† | Not rated | Claims analytics AIPolicy-review agents | Policy review, claims analytics, and renewals augmented; coverage strategy human. |
| Controls Testing Specialist | High AI capability60-70%† | Not rated | Controls-testing agentsEvidence-collection AI | Control testing, evidence gathering, and exception flagging largely automated; conclusions human. |
| Forensic Auditor | Substantial AI capability40-50%† | Not rated | Fraud-analytics AIeDiscovery agentsTransaction-mining | Transaction mining, anomaly detection, and document review automated; investigation and testimony human. |
Summary: Testing-heavy roles (internal audit, fraud, compliance risk) Partial; strategic and ethical roles Augmented-to-Low.
The tool landscape consolidated fast
Recommending current owners matters: several 2024-era leaders were acquired or merged. We re-verify tool currency quarterly before any client delivery.
Methodology and honest caveats
Automation ≠ job elimination
These tiers measure task and workflow automatability. The WEF Future of Jobs Report 2025 projects 170 million new roles created and 92 million displaced by 2030, a net increase of 78 million jobs. Augmentation dominates over replacement in most roles.
The pilot-to-scale gap is real
Roughly two-thirds of organizations remain in pilot mode, and only ~6% capture material EBIT impact. McKinsey's high performers rework processes at almost three times the rate of others, capability does not equal realized value. Workflow redesign and governance are the bottleneck.
Why most agentic projects stall, and ours don't
Industry research keeps citing the same cancelation drivers: escalating costs, unclear value, and weak risk controls. We treat agents like junior employees, scoped, supervised, and measured, with approval gates and live proof on the stack we run ourselves.
Sources, flagged transparently
Cited percentages trace to McKinsey, WEF, Gartner, Ardent Partners, or vendor documentation. Percentages marked † are structured estimates from the current tool landscape. Vendor resolution rates are self-reported and vary by deployment quality. Valuations are point-in-time snapshots.
Get your custom 2026 automation roadmap
See exactly which workflows in your company AI agents can execute today, anchored to the benchmarked numbers above, not the hype.