Gini Loh System Tools

AI Automation Index

Will AI Replace Correspondence Clerks?

Task-level automation exposure for Correspondence Clerks: 17 O*NET tasks scored against their work activities, plus the simulator.

E-E-A-T Verified
Developed & audited by Gini Loh

Automation Exposure

68% weighted task index

17 of 17 tasks scored

Median Salary

$46,800 / year

BLS OEWS, May 2025 — this exact occupation

Employment Level

Low

4,290 people hold this job — size, not growth

Projected 2025–35

-5.6% employment change

500 openings a year · BLS line item

Pay by state

BLS OEWS publishes 26 state estimates for this occupation. Each is its own survey estimate, not the national median adjusted for local prices.

$36,790 – $61,310

Highest medians

Connecticut $61,310
Oregon $61,160
California $53,640
Idaho $51,010
Nebraska $49,800

Lowest medians

Louisiana $36,790
Arizona $39,900
Pennsylvania $40,270
Michigan $40,780
South Carolina $41,710

State figures come from the 43-4021 row in the May 2025 state panel. States where the estimate was suppressed or too small to publish are absent.

What the score is made of

Each task is mapped to its work activities, scored for AI automation potential, and weighted by how important those activities are to the job. These are the components.

Automation potential 68%

Importance-weighted mean of the scored tasks. This is the headline index.

Physical work 5%

Share of weighted tasks in hands-on or machine-operating work activities.

Judgment & influence 35%

Share in analysis, decisions, teaching, directing, advising, or caring for others.

Routine administration 43%

Share in documenting, recording, and administrative activities.

Rewritten by O*NET 0%

Share of weighted tasks O*NET rewrote in a recent update. Reported, not folded into the score.

Consequence of error 0/100 Low

O*NET's own rating of how serious an uncorrectable mistake would be here — respondents, not us. Deliberately not in the index: the score measures what AI can do with the task content, and multiplying it by stakes would be our judgment dressed as data.

Constituent Task Mix

17 tasks for this occupation, weighted by work-activity importance. Shares below are of total task weight, not task count.

High exposure (≥ 60% of the task is automatable) 73%

12 tasks

Medium exposure (≥ 35%) 4%

1 tasks

Low exposure (below 35%) 22%

4 tasks

How would you survive?

Move your week away from the high-exposure tasks and toward the ones that need judgment. The simulator recalculates the index as you shift the mix.

Simulate Upskilling →

Constituent Tasks

#11276 · Core · 80% automatable

Maintain files and control records to show correspondence activities.

data analysis — Maintain operational records in structured systems

Maintain operational records.
#11273 · Core · 82% automatable

Read incoming correspondence to ascertain nature of writers' concerns and to determine disposition of correspondence.

text generation — Reading and deriving actions from text is core LLM

Read materials to determine needed actions.
#11277 · Core · 87% automatable

Gather records pertinent to specific problems, review them for completeness and accuracy, and attach records to correspondence as necessary.

data analysis — Detecting recording errors is deterministic pattern matching

Check data for recording errors. Compile data or documentation.
#11270 · Core · 78% automatable

Prepare documents and correspondence, such as damage claims, credit and billing inquiries, invoices, and service complaints.

text generation — Routine business correspondence fully draftable by AI

Prepare documentation for contracts, transactions, or regulatory compliance. Prepare business correspondence.
#11271 · Core · 85% automatable

Compile data from records to prepare periodic reports.

data analysis — Compiling data or documentation is largely automatable

Compile data or documentation.
#11280 · Core · 88% automatable

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.

text generation — Routine business correspondence fully draftable by AI

Prepare business correspondence.
#11279 · Core · 25% automatable

Route correspondence to other departments for reply.

physical manipulation — Routing physical mail requires sorting by hand

Route mail to correct destinations.
#11281 · Supplemental · 30% automatable

Ensure that money collected is properly recorded and secured.

physical manipulation — Preparing physical cash for deposit is hands-on

Prepare cash for deposit or disbursement.
#11289 · Supplemental · 5% automatable

Process orders for goods requested in correspondence.

physical manipulation — Manual packing for shipping

Package objects for shipping.
#11272 · Supplemental · 62% automatable

Present clear and concise explanations of governing rules and regulations.

information retrieval — AI explains policies; human handles live questions

Explain regulations, policies, or procedures.
#11275 · Supplemental · 87% automatable

Review correspondence for format and typographical accuracy, assemble the information into a prescribed form with the correct number of copies, and submit it to an authorized official for signature.

text generation — Proofreading text for accuracy is core LLM strength

Compile data or documentation. Proofread documents, records, or other files to ensure accuracy.
#11283 · Supplemental · 82% automatable

Compute costs of records furnished to requesters, and write letters to obtain payment.

data analysis — Cost calculation from structured inputs

Calculate costs of goods or services.
#11290 · Supplemental · 85% automatable

Compile data pertinent to manufacture of special products for customers.

data analysis — Compiling data or documentation is largely automatable

Compile data or documentation.
#11274 · Supplemental · 88% automatable

Type acknowledgment letters to persons sending correspondence.

text generation — Routine business correspondence fully draftable by AI

Prepare business correspondence.
#11278 · Supplemental · 88% automatable

Complete form letters in response to requests or problems identified by correspondence.

text generation — Routine business correspondence fully draftable by AI

Prepare business correspondence.
#11287 · Supplemental · 48% automatable

Confer with company personnel regarding feasibility of complying with writers' requests.

interpersonal — Real-time coworker coordination is relational work

Confer with coworkers to coordinate work activities.
#11285 · Supplemental · 30% automatable

Prepare records for shipment by certified mail.

physical manipulation — Preparing physical mail requires hands-on assembly

Prepare outgoing mail.
Embed this calculator on your site ↓

Bloggers & authors: Embed this interactive calculator directly in your articles or comparison guides. It loads dynamically via a responsive iframe (auto-resizing to prevent double scrollbars) and provides a direct, high-value tool for your audience.

<div class="giniloh-calculator-embed" data-calculator="correspondence-clerks"></div> <script src="https://giniloh.com/embed.js" async></script>

Community Sentiment

Do you agree?

Do you agree with the AI automation risk score calculated for this occupation?

Method

How this was scored

  1. 1. Tasks. O*NET task statements for this occupation, from a release whose newest record is dated 08/2026.
  2. 2. Work activities. Each task maps to 1–5 Detailed Work Activities; each activity belongs to one of 41 Generalized Work Activities that O*NET rates for importance (1–5) in this occupation.
  3. 3. Automation potential. Every work activity carries one score for how far current AI can perform it end-to-end. Tasks inherit the mean of their activities.
  4. 4. Weighting. Task weight = 0.35 + 0.65 × (importance − 1)/4, scaled by how often the task is performed and by 0.7 for Supplemental tasks. Importance is the task's own O*NET rating where it exists (18,369 of 18,838 tasks); the rest inherit their work activities' importance. The index is the weighted mean.

Check: across the 910 occupations where O*NET also rates how far machine speed sets the pace, this index moves opposite to it (Spearman ρ = -0.49). The more a job is paced by equipment, the less of its task content AI can address — the relationship you would expect if the scoring is working.

Sources · O*NET 31.0 (task records to 08/2026) · BLS OEWS May 2025 · BLS projections 2025-2035 · rubric v2.1 · tasks for O*NET, not a forecast

See the full method →

Related Article

Engineer at work

Best AI Proof Jobs in a Changing Market

Worried robots are coming for your paycheck? You’re not alone—but the real story is far more hopeful. This guide cuts through the scare headlines to reveal exactly where smart, sustainable career opportunities exist right now. We’ve curated a list of true ai proof jobs—roles in healthcare, skilled trades, and cybersecurity that not only survive automation but thrive because of human judgment and empathy. You’ll get specific, money-making data on projected growth (the BLS says wind turbine technicians, for instance, will grow 44% by 2032), real salary ranges from active job listings, and a simple self-assessment quiz to identify your own most resilient path. Plus, we name companies hiring today and the niche job boards they use. Stop worrying and start planning—your secure career is closer than you think.

Read Article ->