File Clerks
In short: today's best AI models could do about 50% 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 53% by the end of 2028 if progress keeps its long-run pace. The conversations, in-person and hands-on work, about 47% of the time, stays with people.
File correspondence, cards, invoices, receipts, and other records in alphabetical or numerical order or according to the filing system used. Locate and remove material from file when requested.
73,440 people do this job in the US. $43,600 a year.
Reviewed on October 1, 2026: what each task is made of, and what handing it over would ask.
A typical 40-hour week
- AI could take on12 h
- A person checks AI's work8 h
- Desk work AI can't do yet1 h
- Stays with people19 h
File Clerks, now: of a 40-hour week, AI could take on 12 hours, 8 hours become checking its work, 1 hours are desk work it can't do yet, and 19 hours stay with people.
The largest pieces: Improve filing systems (about 5 h, a person checks); Design filing forms (about 4 h); Log filed and removed materials (about 2 h).
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.
- 37%
- 16%
- 5%
- 1%
- 42%
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 50% 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, 2%. With Claude Mythos Preview (early), measured in April 2026, 30%. Today's estimate: 50%.
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, 0.17% went to this job's tasks in May 2026, the 112th largest of the 447 jobs that showed up at all; 4 of its 19 tasks showed up. Within each of Anthropic's methods, never across a change: 0.032% to 0.083% (released February and March 2025); 0.015% to 0.029% (August 2025 to February 2026); 0.18% to 0.17% (April to May 2026).
Each release
| Conversations | Share | Rank | Tasks seen |
|---|---|---|---|
| Released February 2025 | 0.032% | 354 of 708 | 4 of 19 |
| Released March 2025 | 0.083% | 188 of 703 | 3 of 19 |
| The week of Aug 4, 2025 (released September 2025) | 0.015% | 402 of 655 | 3 of 19 |
| The week of Nov 13, 2025 (released January 2026) | 0.019% | 394 of 672 | 3 of 19 |
| The week of Feb 5, 2026 (released March 2026) | 0.029% | 341 of 677 | 4 of 19 |
| April 2026 (released June 2026) | 0.18% | 108 of 434 | 4 of 19 |
| May 2026 (released June 2026) | 0.17% | 112 of 447 | 4 of 19 |
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 66% of this job's working time as physical work. For 64% of the working time, a robot has been shown doing most of the task somewhere: 64% in a purpose-built cell. 0.64 of 3, the 272nd highest of the 923 jobs.
- Not physical 34%
- No robot shown 1%
- In a purpose-built cell 64%
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
47% of the working time: the conversation, in-person and hands-on parts of the tasks.
- Do general office work, the hands-on work, 9%
- Inspect files for order, the hands-on work, 9%
- Improve filing systems, the hands-on work, 7%
- Purge outdated files, the hands-on work, 6%
- Collect materials for filing, the hands-on work, 4%
- File materials in storage, the hands-on work, 3%
What becomes checking the machine
21% 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 32% is desk work that is small if wrong, which can be handed over with lighter checks.
- Improve filing systems, costly if wrong, 15%
- Do general financial tasks, costly if wrong, 3%
- Purge outdated files, costly if wrong, 3%
What to learn
- The hands-on craft. About 42% of the time is hands-on work, which no model here does; robots are rated separately on this page.
- Briefing a model and checking its drafts. About 37% 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.
- 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.
- Running AI inside the business systems. About 16% 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.
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 . 19 tasks from .
| Task | ||||
|---|---|---|---|---|
| Modify or improve filing systems or implement new filing systems.The new system worked out on paper, then the files moved over.Done more than yearly by most who do it. Robots: no robot shown. Anthropic's rating. | 22% | 4.2 h1.7 h to 1.7 work days | Documents 70%, hands-on 30% | 60%Desk part half by now |
| Perform general office activities, such as typing, answering telephones, operating office machines, processing mail, or securing confidential materials.Typing, phones, mail and the office machines.Done hourly or more by most who do it. Robots: in a purpose-built cell, Ripcord robotic digitization workcells (sold). Anthropic's rating. | 12% | 25 min8 min to 50 min | Hands-on 40%, documents 30%, conversation 30% | 30%Desk part half by now |
| Design forms related to filing systems.Forms drawn up at the desk.Done hourly or more by most who do it. | 11% | 2.5 h50 min to 6.7 h | Documents 100% | 96%Half by now |
| Eliminate outdated or unnecessary materials, destroying them or transferring them to inactive storage, according to file maintenance guidelines or legal requirements.Checking what the schedule allows, then shredding or boxing the files.Done more than yearly by most who do it. Robots: in a purpose-built cell, Ripcord robotic digitization workcells (Hayward, CA factory) (sold). Anthropic's rating. | 9% | 1.3 h25 min to 3.3 h | Hands-on 70%, documents 30% | 30%Desk part half by now |
| Perform periodic inspections of materials or files to ensure correct placement, legibility, or proper condition.Going through the cabinets to check placement and condition.Done daily by most who do it.Robots: in a purpose-built cell, Tally / Simbe Vision (deployed). Anthropic's rating. | 9% | 1.3 h25 min to 3.3 h | Hands-on 100% | Stays with people |
| Keep records of materials filed or removed, using logbooks or computers and generate computerized reports.The log kept in the system and reports run from it.Done hourly or more by most who do it. | 5% | 25 min8 min to 1.3 h | Systems 100% | 100%Half by now |
| Complete general financial activities, such as processing accounts payable, reviewing invoices, collecting cash payments, or issuing receipts.Invoices reviewed and paid in the system, with some cash taken and receipts issued.Done daily by most who do it. | 4% | 33 min13 min to 1.7 h | Systems 60%, documents 20%, hands-on 20% | 80%Desk part half by now |
| Input data, such as file numbers, new or updated information, or document information codes into computer systems to support document and information retrieval.Entry into the records system.Done daily by most who do it. AI observed on this task. | 4% | 21 min8 min to 50 min | Systems 100% | 100%Half by now |
| Gather materials to be filed from departments or employees.Collecting the papers from each department.Done daily by most who do it. Robots: in a purpose-built cell, Ripcord robotic digitization workcell (SCARA-arm based) (sold). Anthropic's rating. | 4% | 25 min8 min to 50 min | Hands-on 100% | Stays with people |
| Sort or classify information according to guidelines, such as content, purpose, user criteria, or chronological, alphabetical, or numerical order.Sorting the papers by content and order.Done daily by most who do it. Robots: in a purpose-built cell, Ripcord robotic digitization workcell (sold). Anthropic's rating. | 4% | 21 min8 min to 50 min | Documents 50%, hands-on 50% | 50%Desk part half by now |
| Find, retrieve, and make copies of information from files in response to requests and deliver information to authorized users.Pulling the file, copying it and taking it to whoever asked, with the request logged.Done daily by most who do it. Robots: in a purpose-built cell, Ripcord robotic digitization workcell (sold). Anthropic's rating. | 4% | 21 min8 min to 50 min | Hands-on 80%, systems 20% | 20%Desk part half by now |
| Scan or read incoming materials to determine how and where they should be classified or filed.Reading each item to decide where it belongs.Done daily by most who do it. | 3% | 13 min4 min to 38 min | Documents 70%, hands-on 30% | 70%Desk part half by now |
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.
Where this job works
The industry groups that employ the most of it (BLS, May 2025).
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 File Clerks is within reach of AI now?
- An estimated 50% of the working time is within reach of AI models now and 53% 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 File Clerks 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: Improve filing systems (60% of the task's time now, 70% by the end of 2028); Design filing forms (96% of the task's time now); and Log filed and removed materials (100% of the task's time now).
- Which tasks of File Clerks stay with people?
- Tasks that are people work or body work are not counted as within reach at any date. The largest: Inspect files for order; Collect materials for filing; and File materials in storage. Together such tasks are about 18% of the working time, and the people and body parts of the other tasks stay with people too.
- What should File Clerks learn as AI takes on more of the work?
- By the job's own task-by-task review, the first things to learn are the hands-on craft; briefing a model and checking its drafts; and reviewing work you did not do yourself. By the end of 2028 on the long-run pace at four in five, about 47% of the working time stays with people, 21% becomes checking a model's work and 32% 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 File Clerks?
- In the Claude.ai conversations Anthropic sampled for May 2026 and tied to a work task, 0.17% went to this job's tasks, the 112th largest share of the 447 jobs that showed up; 4 of its 19 tasks appeared. Anthropic's labor market measure puts the share of the job's tasks seen in real Claude use at work at 16%. 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 File Clerks?
- Anthropic's robot exposure rating (September 2026) finds 66% of the working time is physical work, and that a robot has been shown doing most of the task, somewhere, for 64% of the working time. Its index is 0.64 of 3. One example: Ripcord robotic digitization workcells for "do general office work". 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 19 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 October 1, 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.