Occupation · SOC 43-4021.00
Compose letters or electronic correspondence in reply to requests for merchandise, damage claims, credit and other information, delinquent accounts, incorrect billings, or unsatisfactory services. Duties may include gathering data to formulate reply and preparing correspondence.
Also called: Correspondence Clerk · Correspondence Coordinator · Correspondence Representative (Correspondence Rep) · Dispute Resolution Analyst · Chargeback Specialist · Claims Correspondence Clerk · Correspondent · Dispute Specialist · Office Technician (Office Tech) · Technical Clerk · Authorization Coordinator · Authorization Representative (Authorization Rep)
Job family: Office and Administrative Support Occupations
A source-stamped Markdown brief of this occupation — paste it into an agent, or fetch
/roles/role-43-4021-00/context.md directly.
A fast read on where AI already shows up in this occupation, where it stays a copilot, where humans remain in the loop, and what the labor market is doing. Built from observed Claude.ai conversations mapped to O*NET tasks and from published research — measures of usage and exposure, not advice or predictions that the job is going away.
Task areas most often handled directively in observed AI conversations — candidates to delegate with light review.
Task areas where people work with AI — iterating, learning, or checking — staying in the loop rather than handing the task off.
Task areas where a human was still judged necessary in a large share of observed conversations — not a safety ruling, an observed-need signal.
The capabilities O*NET rates most important for this occupation — the human ground the work is built on.
See all skills →Independent published positions, read together — not a forecast.
91st-percentile task overlap — yet about 700 openings a year (-5.6% projected, BLS), and observed AI use leans 5482% copilot, not hand-off (AEI) . What exposure means →
What today's research says about this occupation's exposure to AI, how AI is actually being used in it, and where employment is headed. These are positions within published studies — measures of exposure and usage, not predictions that this job will disappear.
Each study uses its own scale, so the raw scores are not comparable across rows — the percentile (this job's rank among all U.S. occupations with data) is the comparable figure, and sizes the bars.
| Measure | Rank vs all occupations | Percentile | Score |
|---|---|---|---|
| Overall AI exposure (Felten et al.) High | 81st | 1.1 | |
| LLM task exposure, γ (OpenAI / Eloundou) High | 89th | 1.0 | |
| AI assistant applicability (Microsoft) High | 86th | 0.3 |
OpenAI's exposure study scores tasks three ways: with a language model alone (α 1.0), with simple added tooling (β 1.0), and including AI-powered software (γ 1.0). Higher means more of the job's tasks could be done at least twice as fast — not that they will be automated away.
Most of this job's tasks can be done remotely (Dingel–Neiman), which tends to track with higher digital and AI exposure.
A pre-LLM (2013) estimate of how automatable this job is by computerization and robotics. Shown for historical context only — it is not part of any current AI ranking.
Frey–Osborne probability 0.9 · 72nd percentile among occupations · High
Among measured AI assistant conversations mapped to this occupation (Anthropic Economic Index, 2026-01-15), these task types came up most. These are shares of observed AI conversations — not shares of the job, of worker time, or of what could be automated.
| Compose letters in reply to correspondence concerning such items as requests for merchandise, damage claims, credit information requests, delinquent accounts, incorrect billing, or unsatisfactory service. | 21.7% | |
| Complete form letters in response to requests or problems identified by correspondence. | 2.6% | |
| Present clear and concise explanations of governing rules and regulations. | 1.0% | |
| Gather records pertinent to specific problems, review them for completeness and accuracy, and attach records to correspondence as necessary. | 0.6% |
Independent U.S. Bureau of Labor Statistics employment projection for 2024–2034 — a labor-market forecast, not an AI-impact forecast.
| Outlook | Declining · -5.6% by 2034 |
| Projected annual openings | 700 |
| Employment 2024 → 2034 | 6,900 → 6,500 |
“Annual openings” counts new jobs plus replacements for workers who leave the occupation, so it can be large even when growth is modest.
The ILO's 2025 global study scores generative-AI exposure on the international ISCO-08 occupation system, not US SOC. Bridged through the published (and approximate, many-to-many) IBS O*NET-SOC ↔ ISCO-08 crosswalk, this US occupation corresponds to the international 2 occupations below. Exposure here means how much of the work's tasks today's AI can attempt — task overlap, not automation, adoption, or jobs lost.
| International occupation (ISCO-08) | Task exposure (2025) | Most tasks fall in |
|---|---|---|
| Clerical Support Workers Not Elsewhere Classified · 4419 | 63% | Gradient 4 |
| Inquiry Clerks · 4225 | 57% | Gradient 3 |
Read the whole six-band gradient on the GenAI exposure gradient page. The crosswalk is approximate: a US occupation can map to several international ones, and the ILO scores describe the international occupation, not this exact US role.
How people actually apply AI to this occupation's tasks, from Claude.ai (Free and Pro) conversations in the Anthropic Economic Index, 2026-01-15. This is one AI assistant's consumer sample — not all AI, not the whole workforce. Autonomy and the collaboration mix are model-rated estimates; figures below the sample floor are hidden.
| Augmentation vs. automation | 54.8% working with AI · 41.8% handed to AI |
| Most common way people use AI here | Iteration · you and AI go back and forth |
| Typical AI autonomy | 3.0 / 5 · higher = AI acts more independently |
| Used for work (vs. personal / coursework) | 67.9% |
The role's most common tasks in AI conversations, each tagged with how people work with the AI on it. “Usage” is the share of observed conversations, not of the job.
| Task | How | Usage |
|---|---|---|
| Compose letters in reply to correspondence concerning such items as requests for merchandise, damage claims, credit information requests, delinquent accounts, incorrect billing, or unsatisfactory service. | Iteration | 20.5% |
| Complete form letters in response to requests or problems identified by correspondence. | Iteration | 4.7% |
| Gather records pertinent to specific problems, review them for completeness and accuracy, and attach records to correspondence as necessary. | Directive | 1.9% |
| Compose correspondence requesting medical information and records. | Iteration | 0.6% |
| Type acknowledgment letters to persons sending correspondence. | Iteration | 0.4% |
Tasks where the model most often judged that a person remained necessary — a useful read on the current boundary, not a guarantee.
| Compose letters in reply to correspondence concerning such items as requests for merchandise, damage claims, credit information requests, delinquent accounts, incorrect billing, or unsatisfactory service. | 99.6% | |
| Complete form letters in response to requests or problems identified by correspondence. | 98.5% | |
| Type acknowledgment letters to persons sending correspondence. | 97.7% | |
| Compose correspondence requesting medical information and records. | 95.2% | |
| Gather records pertinent to specific problems, review them for completeness and accuracy, and attach records to correspondence as necessary. | 91.0% |
Example prompts phrased from the tasks people most often delegate to AI in this occupation (Anthropic Economic Index). Each shows the underlying measured task and its share of observed AI use. They are suggested phrasings of real tasks — starting points, not endorsed instructions.
Help me compose letters in reply to correspondence concerning such items as requests for merchandise, damage claims, credit information requests, delinquent accounts, incorrect billing, or unsatisfactory service. From: Compose letters in reply to correspondence concerning such items as requests for merchandise, damage claims, credit information requests, delinquent accounts, incorrect billing, or unsatisfactory service. · 20.5% of measured AI use · task iteration
Help me complete form letters in response to requests or problems identified by correspondence. From: Complete form letters in response to requests or problems identified by correspondence. · 4.7% of measured AI use · task iteration
Help me gather records pertinent to specific problems, review them for completeness and accuracy, and attach records to correspondence as necessary. From: Gather records pertinent to specific problems, review them for completeness and accuracy, and attach records to correspondence as necessary. · 1.9% of measured AI use · directive
Help me compose correspondence requesting medical information and records. From: Compose correspondence requesting medical information and records. · 0.6% of measured AI use · task iteration
All 21 tasks O*NET lists for this occupation, ordered by importance. Each links to its own page with AI-exposure and observed-use detail.
O*NET importance rating, from 1 (not important) to 5 (extremely important).
| Written Expression | 4.1 | |
| Written Comprehension | 4.0 | |
| Oral Comprehension | 3.8 | |
| Oral Expression | 3.6 | |
| Near Vision | 3.6 | |
| Problem Sensitivity | 3.4 | |
| Speech Clarity | 3.4 | |
| Deductive Reasoning | 3.3 | |
| Information Ordering | 3.3 | |
| Inductive Reasoning | 3.1 | |
| Speech Recognition | 3.1 | |
| Fluency of Ideas | 3.0 | |
| Category Flexibility | 3.0 | |
| Selective Attention | 3.0 | |
| Originality | 2.9 | |
| Number Facility | 2.8 |
| Writing | 4.0 | |
| Reading Comprehension | 3.8 | |
| Active Listening | 3.8 | |
| Speaking | 3.5 | |
| Critical Thinking | 3.4 | |
| Monitoring | 2.9 | |
| Mathematics | 2.8 | |
| Active Learning | 2.8 |
| Judgment and Decision Making | 3.3 | |
| Social Perceptiveness | 3.1 | |
| Service Orientation | 3.0 | |
| Time Management | 3.0 | |
| Coordination | 2.9 | |
| Complex Problem Solving | 2.9 | |
| Persuasion | 2.8 |
Skills employers ask for in job postings for this occupation (Lightcast), with whether each is a common or specialized skill.
| Example | Category | |
|---|---|---|
| Epic Systems | Medical software | Hot technology In demand |
| Microsoft Excel | Spreadsheet software | Hot technology In demand |
| Microsoft Office software | Office suite software | Hot technology In demand |
| Microsoft Outlook | Electronic mail software | Hot technology In demand |
| Microsoft Word | Word processing software | Hot technology In demand |
| Microsoft Access | Data base user interface and query software | Hot technology |
| Microsoft PowerPoint | Presentation software | Hot technology |
| Microsoft SharePoint | Document management software | Hot technology |
| Microsoft Windows | Operating system software | Hot technology |
| SAP software | Enterprise resource planning ERP software | Hot technology |
| Electronic data interchange EDI software | Enterprise application integration software | |
| Electronic health record EHR software | Medical software | |
| GE Healthcare Centricity EMR | Medical software | |
| Healthcare common procedure coding system HCPCS | Medical software | |
| Imaging software | Graphics or photo imaging software |
How characteristic each condition is of the job, on O*NET's 1–5 context scale (higher = more present in day-to-day work). Each condition links to how it varies across all occupations.
What to study: Business, Management, Marketing, and Related Support Services . Fields of study crosswalked to this occupation (NCES CIP–SOC), not a requirement.
Share of people in this occupation at each level of education.
| High School Diploma | 42.9% | |
| Associate's Degree (or other 2-year degree) | 28.1% | |
| Some College Courses | 17.4% | |
| Bachelor's Degree | 11.6% |
The interests and personal qualities O*NET associates with people who do this work.
| Conventional | 6.9 | |
| Enterprising | 3.4 | |
| Social | 2.7 | |
| Investigative | 2.5 |
| Office Work | 6.5 | |
| Accounting | 2.9 | |
| Finance | 2.2 | |
| Management/Administration | 2.0 | |
| Law | 1.8 | |
| Human Resources | 1.8 | |
| Information Technology | 1.7 | |
| Personal Service | 1.7 | |
| Sales | 1.6 |
| Attention to Detail | 2.5 | |
| Dependability | 2.4 | |
| Integrity | 1.7 |
U.S. · annual wages (BLS OEWS)
| 10th percentile | $33,530 |
| 25th percentile | $38,510 |
| Median (50th) | $46,740 |
| 75th percentile | $52,480 |
| 90th percentile | $62,200 |
| People employed | 6,260 |
Where these workers are employed, by number of jobs (national, BLS OEWS). Pay shown is the occupation's national median, not industry-specific.
| Industry | Workers | National median pay |
|---|---|---|
| Finance and Insurance · Sector | 1,230 | $44,400 |
| Real Estate and Rental and Leasing · Sector | 960 | $46,740 |
| Professional, Scientific, and Technical Services · Sector | 900 | $46,530 |
| Health Care and Social Assistance · Sector | 600 | $44,600 |
| Direct Health and Medical Insurance Carriers · National industry | 310 | $48,640 |
| Management of Companies and Enterprises · Sector | 310 | $44,550 |
| Retail Trade · Sector | 270 | $42,200 |
| Wholesale Trade · Sector | 140 | $55,850 |
| Manufacturing · Sector | 120 | $52,870 |
| Educational Services · Sector | 70 | $48,860 |
| Transportation and Warehousing · Sector | 50 | $51,470 |
| Temporary Help Services · National industry | 50 | $39,780 |
Industries where this occupation is far more common than in the economy as a whole. The location quotient is how many times more concentrated it is here (a value of 5 means five times its economy-wide share).
| Industry | Concentration | Workers |
|---|---|---|
| Direct Health and Medical Insurance Carriers · National industry | 17× | 310 |
| Real Estate and Rental and Leasing · Sector | 9.99× | 960 |
| Finance and Insurance · Sector | 4.87× | 1,230 |
| Management of Companies and Enterprises · Sector | 2.72× | 310 |
| Professional, Scientific, and Technical Services · Sector | 2.06× | 900 |
| Health Care and Social Assistance · Sector | 0.64× | 600 |
| Wholesale Trade · Sector | 0.57× | 140 |
| Retail Trade · Sector | 0.43× | 270 |
Part of the Financial Services career cluster.
Side-by-side comparisons place two occupations’ pay, preparation, skills, and AI exposure on the same page — same data, same scale, no forecast.
Options the data surfaces for Correspondence Clerks — not advice or a forecast. Each is a real cross-link you can follow into the evidence.
Capabilities this work builds that are used across many other occupations.
Occupations O*NET rates as related — the nearby moves on the map.
How people typically prepare for this work.
On the global GenAI exposure gradient this work sits around the 97th percentile of 427 international occupations.
Correspondence Clerks show 91st-percentile AI task overlap — and about 700 annual U.S. openings
Correspondence Clerks show 91st-percentile AI task overlap — and about 700 annual U.S. openings • Correspondence Clerks rank in the 91st percentile (High band) for AI task overlap across U.S. occupations — a measure of how much of the work today's AI can attempt, not how much is automated. (Eloundou et al. (GPTs are GPTs) + Felten AIOE) • The occupation is projected to see about 700 U.S. job openings per year (2024–34), counting growth and replacement — a labor-demand projection made independently of AI. (BLS Employment Projections 2024–34) • BLS projects employment to be declining (-5.6%) from 2024 to 2034. (BLS Employment Projections 2024–34) • Median annual pay is $46,740, across about 6,260 U.S. workers. (BLS OEWS (May 2024)) • Of the AI use actually observed for this work, 55% looks like augmentation (drafting, iterating, checking) rather than hands-off automation — from a Claude.ai usage sample, not a census. (2026-01-15-v4-plus-2025-03-27-v2) Source: Singulariki — "Correspondence Clerks". https://singulariki.com/roles/role-43-4021-00 Note: AI task overlap measures what today's AI can attempt, not automation, job loss, or a forecast.
AssetsShare imageMethodology & sourcesPress & newsroomThe newsroom
Every line is built only from figures this page already shows and cites. AI task overlap means what today's AI can attempt — not automation, job loss, or a forecast.
Every figure above traces to a named public dataset and the exact release below — not hand-written opinion. See the full methodology for what each measure does and does not mean.
Data compiled June 2, 2026. Figures are estimates, not advice.
Singulariki. "Correspondence Clerks." Singulariki: a source-backed encyclopedia of work. Built from O*NET 30.3; BLS Occupational Employment and Wage Statistics (OEWS) May 2024; BLS Employment Projections 2024–2034; Anthropic Economic Index v4 (2026-01-15) + v2 (2025-03-27); Microsoft “Working with AI” working-with-ai; “GPTs are GPTs” (Eloundou et al.) arXiv 2303.10130; AI Occupational Exposure (AIOE) Felten, Raj & Seamans; ILO / Gmyrek et al. GenAI exposure gradient 2025; IBS O*NET-SOC ↔ ISCO-08 occupation crosswalk 2022; Frey & Osborne (2013) frey-osborne-automation; Dingel & Neiman (2020) dingel-neiman-workathome. Accessed June 7, 2026. https://singulariki.com/roles/role-43-4021-00
Singulariki. (2026). Correspondence Clerks. Singulariki: a source-backed encyclopedia of work. Retrieved June 7, 2026, from https://singulariki.com/roles/role-43-4021-00
@misc{singulariki-role-43-4021-00,
title = {Correspondence Clerks},
author = {{Singulariki}},
year = {2026},
note = {O*NET 30.3; BLS Occupational Employment and Wage Statistics (OEWS) May 2024; BLS Employment Projections 2024–2034; Anthropic Economic Index v4 (2026-01-15) + v2 (2025-03-27); Microsoft “Working with AI” working-with-ai; “GPTs are GPTs” (Eloundou et al.) arXiv 2303.10130; AI Occupational Exposure (AIOE) Felten, Raj & Seamans; ILO / Gmyrek et al. GenAI exposure gradient 2025; IBS O*NET-SOC ↔ ISCO-08 occupation crosswalk 2022; Frey & Osborne (2013) frey-osborne-automation; Dingel & Neiman (2020) dingel-neiman-workathome. Accessed June 7, 2026},
url = {https://singulariki.com/roles/role-43-4021-00}
} Citations name the underlying public dataset releases — they reflect what this page is built from, not just the URL.