Data Entry Keyers
In short: today's best AI models could do about 80% of this job's working time (projected from METR's April 2026 measurement and Epoch's November 2026 index) if the work were set up for them, and about the same by the end of 2028 if progress keeps its long-run pace. The conversations, in-person and hands-on work, about 20% of the time, stays with people.
Operate data entry device, such as keyboard or photo composing perforator. Duties may include verifying data and preparing materials for printing.
127,080 people do this job in the US. $41,340 a year.
Reviewed on September 30, 2026: what each task is made of, and what handing it over would ask.
A typical 40-hour week
- AI could take on20 h
- A person checks AI's work12 h
- Desk work AI can't do yet0 h
- Stays with people8 h
Data Entry Keyers, now: of a 40-hour week, AI could take on 20 hours, 12 hours become checking its work, 0 hours are desk work it can't do yet, and 8 hours stay with people.
The largest pieces: Verify entered data (about 9 h); Key data from source documents (about 6 h, a person checks); Resolve garbled messages (about 6 h, a person checks).
Estimates for a typical week of this job, if progress keeps its long-run pace, counting work the best models finish four times in five. Every task is below.
of the working time. Now is projected from METR's April 2026 measurement and Epoch's November 2026 index.
- 21%
- 59%
- 3%
- 0%
- 17%
of the work: Anthropic's measure of what people already bring to Claude at work (2025). It counts help with part of a task, so it can be higher than the share within reach.
On the chosen pace and reliability. Estimate.
Share of the working time within reach, now to 2030
Change assumptionsHide assumptionsLong-run pace, four in five
What each choice means: , , , , . Every pace starts from .
At this date and pace, models finish tasks up to 2.4 work days long four times in five.
What handing it over would take, now
The 80% within reach splits by what a model would need. The dates, pace and reliability follow the controls above. Estimates.
of the working time. The material can be handed to a model and the result handed back.
of the working time. The model has to be connected to the system the work lives in.
is conversation short enough by length. This site leaves conversation with people, so it is not in the total.
How far this has come
With GPT-4, measured in March 2023, 0% of this job's working time was within reach at . With Claude 3.7 Sonnet, measured in February 2025, 8%. With Claude Mythos Preview (early), measured in April 2026, 75%. Today's estimate: 80%.
A backcast: today's task data with the best model METR had measured by each date. It shows how fast the models have moved, not what anyone forecast at the time.
Measured use over time
Of the Claude.ai conversations Anthropic sampled and tied to a work task, 0.085% went to this job's tasks in May 2026, the 181st largest of the 447 jobs that showed up at all; 4 of its 9 tasks showed up. Within each of Anthropic's methods, never across a change: 0.021% to 0.011% (released February and March 2025); 0.026% to 0.024% (August 2025 to February 2026); 0.079% to 0.085% (April to May 2026).
Each release
| Conversations | Share | Rank | Tasks seen |
|---|---|---|---|
| Released February 2025 | 0.021% | 439 of 708 | 3 of 9 |
| Released March 2025 | 0.011% | 498 of 703 | 3 of 9 |
| The week of Aug 4, 2025 (released September 2025) | 0.026% | 337 of 655 | 2 of 9 |
| The week of Nov 13, 2025 (released January 2026) | 0.023% | 371 of 672 | 2 of 9 |
| The week of Feb 5, 2026 (released March 2026) | 0.024% | 372 of 677 | 2 of 9 |
| April 2026 (released June 2026) | 0.079% | 183 of 434 | 4 of 9 |
| May 2026 (released June 2026) | 0.085% | 181 of 447 | 4 of 9 |
Measured, not estimated: Anthropic's samples of Claude.ai conversations, each tied by a model to an O*NET task, from the Anthropic Economic Index. It says where Claude's use goes, not how much of this job's working time it covers, so it is not on the same scale as the reach above. The releases differ in their sample, their classifier and the smallest share they report (the June 2026 file rounds each task to a hundredth of a percent, so a smaller task reads as none), which is why a change of method starts a new line. At least 96% of each older release's task-tied use matched this site's tasks, and all of the newest. Claude.ai only, so use through other tools and other companies' models is not in it. Every job ranked.
Robots and the physical work
Anthropic rates 10% of this job's working time as physical work. For 10% of the working time, a robot has been shown doing most of the task somewhere: 10% in a purpose-built cell. 0.10 of 3, the 514th highest of the 923 jobs.
- Not physical 90%
- In a purpose-built cell 10%
Anthropic's rating, not this site's estimate: Claude Opus 5, searching the web, rated each physical O*NET task by the kind of setting in which a robot it could cite (deployed, sold or demonstrated) does most of the task, weighted by its own estimate of each task's share of the time (What work can robots do?, September 30, 2026, data CC BY 4.0; every job ranked). A robot shown somewhere is not a robot in your building: cost, setup and safety decide that. The share within reach on this page still counts no body work. The robot behind each rated task is under the task below.
What changes for the people in this job
From the task-by-task review, by the end of 2028 on the long-run pace at four in five. It says which parts of the work stay with people and which change, not whether anyone keeps a job: that depends on decisions no data set can see.
What stays with people
20% of the working time: the conversation, in-person and hands-on parts of the tasks.
- Compile and sort data before entry, the hands-on work, 8%
- File completed documents, the hands-on work, 4%
- Select work materials, the hands-on work, 3%
- Correct or report entry errors, the conversation, 3%
- Load media into machines, the hands-on work, 1%
What becomes checking the machine
29% of the working time: desk work a model could do by then, in tasks where a mistake costs something or a licensed person signs. A person checks it. Another 51% is desk work that is small if wrong, which can be handed over with lighter checks.
- Key data from source documents, costly if wrong, 15%
- Resolve garbled messages, costly if wrong, 14%
What to learn
- Running AI inside the business systems. About 59% of the time is work inside business systems, which a model can do only once it is connected: knowing the systems well enough to set that up and watch it.
- Reviewing work you did not do yourself. The tasks under What becomes checking the machine, where a mistake costs something, stay a person's to check: spotting what a draft gets wrong is a skill of its own.
- Handling protected records. Much of the work touches protected personal records: knowing what may be shared with a model, and how, is part of the job.
- Briefing a model and checking its drafts. About 21% of the time is documents and data, the part models reach first: writing a clear brief, giving the right material, and checking what comes back.
- The hands-on craft. About 17% of the time is hands-on work, which no model here does; robots are rated separately on this page.
Your own version of this job: read it against your job description. The headline is free; the full personal report, every task with its timeline and what to learn, is $19 (a sample).
The tasks
Sorted by . 9 tasks from .
| Task | ||||
|---|---|---|---|---|
| Compare data with source documents, or re-enter data in verification format to detect errors.Entries compared with the source or re-keyed to check.Done hourly or more by most who do it. | 22% | 21 min8 min to 50 min | Systems 100% | 100%Half by now |
| Compile, sort, and verify the accuracy of data before it is entered.Documents sorted and checked before keying.Done daily by most who do it. | 21% | 25 min8 min to 1.3 h | Documents 60%, hands-on 40% | 60%Desk part half by now |
| Read source documents such as canceled checks, sales reports, or bills, and enter data in specific data fields or onto tapes or disks for subsequent entry, using keyboards or scanners.Data keyed from checks, reports and bills.Done hourly or more by most who do it. AI observed on this task. | 15% | 17 min4 min to 50 min | Systems 100% | 100%Half by now |
| Locate and correct data entry errors, or report them to supervisors.Errors fixed in the system or reported to the supervisor.Done hourly or more by most who do it. | 14% | 13 min4 min to 38 min | Systems 80%, conversation 20% | 80%Desk part half by now |
| Resolve garbled or indecipherable messages, using cryptographic procedures and equipment.Cryptographic procedures worked through on the equipment.Done yearly or less by most who do it. | 14% | 50 min13 min to 1.7 h | Documents 60%, systems 40% | 100%Half by now |
| Maintain logs of activities and completed work.Work completed logged.Done daily by most who do it. | 5% | 8 min3 min to 21 min | Systems 100% | 100%Half by now |
| Store completed documents in appropriate locations.Documents put away.Done hourly or more by most who do it. Robots: in a purpose-built cell, Ripcord robotic digitization workcell (sold). Anthropic's rating. | 4% | 4 min2 min to 13 min | Hands-on 100% | Stays with people |
| Select materials needed to complete work assignments.Gathering what the assignment needs.Done daily by most who do it. Robots: in a purpose-built cell, Moxi (Diligent Robotics) (sold). Anthropic's rating. | 3% | 4 min2 min to 13 min | Hands-on 100% | Stays with people |
| Load machines with required input or output media, such as paper, cards, disks, tape, or Braille media.Paper, cards, disks or tape loaded.Done daily by most who do it.Robots: in a purpose-built cell, baumannperfecta RC automatic cutting system (industrial robot + Multi-Paper-Tool gripper) (sold). Anthropic's rating. | 1% | 4 min2 min to 13 min | Hands-on 100% | Stays with people |
Within reach, per task: the share of the task's time whose instances are short enough for a model to finish at the chosen date, pace and reliability. Half by: the date half of that time comes within reach. Made of: how the task's time splits among five kinds of doing. Only the documents and systems parts can count as within reach, so a task that is 30% documents can be at most 30% within reach.
Jobs that share skills with this one
From O*NET's related occupations. Within reach now and by the end of 2028, long-run pace, four in five.
- File Clerks50% now, 53% by 2028
- Word Processors and Typists77% now, 77% by 2028
- Office Machine Operators, Except Computer11% now, 11% by 2028
- Mail Clerks and Mail Machine Operators, Except Postal Service13% now, 13% by 2028
- Office Clerks, General56% now, 56% by 2028
- Postal Service Mail Sorters, Processors, and Processing Machine Operators1% now, 1% by 2028
Where this job works
The industry groups that employ the most of it (BLS, May 2025).
- Employment Services11,940 people
- Accounting, Tax Preparation, Bookkeeping, and Payroll Services6,530 people
- Computer Systems Design and Related Services6,110 people
- Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services5,260 people
- Management, Scientific, and Technical Consulting Services5,260 people
Where the most of these jobs are
The metro areas with the most people in it (BLS, May 2025).
Questions about this job
- How much of the work of Data Entry Keyers is within reach of AI now?
- An estimated 80% of the working time is within reach of AI models now and 80% by the end of 2028, on METR's long-run pace at four in five. Within reach means a model could finish work that long at that reliability. It is not a forecast of jobs lost or of savings.
- Which tasks of Data Entry Keyers come within reach first?
- Weighted by the time each takes, the tasks with the most within reach now (projected from METR's April 2026 measurement and Epoch's November 2026 index) are: Verify entered data (100% of the task's time now); Key data from source documents (100% of the task's time now); and Resolve garbled messages (100% of the task's time now).
- Which tasks of Data Entry Keyers stay with people?
- Tasks that are people work or body work are not counted as within reach at any date. The largest: File completed documents; Select work materials; and Load media into machines. Together such tasks are about 9% of the working time, and the people and body parts of the other tasks stay with people too.
- What should Data Entry Keyers learn as AI takes on more of the work?
- By the job's own task-by-task review, the first things to learn are running AI inside the business systems; reviewing work you did not do yourself; and handling protected records. By the end of 2028 on the long-run pace at four in five, about 20% of the working time stays with people, 29% becomes checking a model's work and 51% could be handed over with lighter checks. It says which parts of the work change, never whether anyone keeps a job.
- Do people already use AI for the work of Data Entry Keyers?
- In the Claude.ai conversations Anthropic sampled for May 2026 and tied to a work task, 0.085% went to this job's tasks, the 181st largest share of the 447 jobs that showed up; 4 of its 9 tasks appeared. Anthropic's labor market measure puts the share of the job's tasks seen in real Claude use at work at 67%. These are measures of use, not of how much of the work AI could finish, and they cover Claude only.
- Can robots do the physical work of Data Entry Keyers?
- Anthropic's robot exposure rating (September 2026) finds 10% of the working time is physical work, and that a robot has been shown doing most of the task, somewhere, for 10% of the working time. Its index is 0.10 of 3. One example: Ripcord robotic digitization workcell for "file completed documents". A robot shown somewhere is not one in every workplace; this site's own estimates still count no physical work as within reach.
- How is this estimated?
- From the 9 tasks O*NET lists for the occupation, each with an estimated length checked against GDPval and a share of the working time, reviewed one task at a time on September 30, 2026 for what each is made of (documents, business systems, conversation, in person, hands-on), and METR's measurements of how long a task the best AI models can finish, carried forward to the date from the latest, Claude Mythos Preview (early) in April 2026. Data checked November 10, 2026.
This job in your state or metro area
Embed these figures on your site
A small card with the share of working time within reach now and at the end of 2027, 2028 and 2030, the date and a link back. Free under CC BY 4.0; the card carries the credit.
Your own version of this job
The page above is the occupation. Your job is your title and what you actually do: give both, and they are read against these tasks in your browser, nothing sent anywhere. The headline is free: how much of your working time is within reach now and by the end of 2028, and how many of the tasks your description names.
The full personal report is $19, to print or save as a PDF: every task with when it comes within reach, which ones your description names, the parts of your description the task list does not cover, what stays with you, what becomes checking the machine and what to learn.