US dataswitch to UK
Tax Examiners and Collectors, and Revenue Agents
collecting taxes from individuals or businesses according to prescribed laws and regulations, notifying taxpayers of any overpayment or underpayment and maintaining records for each case. If that's your week, this page is about your job.
The honest answer
This job is splitting in two: maintaining records for each case is work AI now does quickly and cheaply, and collecting taxes from individuals or businesses according to prescribed laws and regulations is work it can't touch.
Which half fills your week decides your exposure. That is more in your control than it sounds.
Your week, as this page understands it
Determine tax liability or collect taxes from individuals or business firms according to prescribed laws and regulations. The job title says “tax examiners”, “collectors” or “revenue agents”: officially one job, several names. The real job is the part underneath: collecting taxes from individuals or businesses according to prescribed laws and regulations. That is the thing someone has to be right about.
The exposed part of this job is specific, and we won’t pretend it is coming back. But tax examiners and collectors, and revenue agents is not one task. It is 21 scored ones, and they are nowhere near equally exposed. The most durable of them, on this evidence, is collecting taxes from individuals or businesses according to prescribed laws and regulations, and the ledger below shows exactly why.
Where the work sits, by task weight
- shifting to AI
- 46%
- changing shape
- 34%
- staying human
- 20%
These bars are tasks changing hands, not people being counted out. The ledger below shows which.
Whole-job exposure score 59 out of 100 (54–65 allowing for uncertainty): partial exposure, across 21 scored tasks. The number is the support for the sentence above it, not a headline about anyone’s future.
How we know this
What is measured: Every published task statement for tax examiners and collectors, and revenue agents is rated on five dimensions: can a model produce the output, does the work need a body in a room, does it need a legally accountable person, does it depend on a person being trusted in the moment, and how much data exists. A published formula turns those five ratings into the score; the model never writes the number.
How the bar is built: Each task’s share of the bar is its published importance weight, so a task you do all day counts for more than one you do twice a year.
Release: 2026-q4.1, scores computed 2026-08-05. Read the full method.
Your job, task by task
These are the official task statements for this occupation, in plain English, sorted by what the evidence says is happening to each one. The official wording sits under every line so you can check the rewrite against it.
- 2 of this occupation's scored task statements carry a score that was measured under a different occupation's context, because the statement is shared between jobs and has only been scored once so far. Each one names the occupation it was measured under in the free routes below; none is presented as a measurement for this job.
- Task clusters are not derived in this release, so the task-cluster field is empty and no Collab365 Space routing is attached to this occupation yet.
Shifting to AI
8 tasksTasks today’s tools can already do most of. This is the part we will not soften: where these rows are the bulk of your week, the week changes.
Notifying taxpayers of any overpayment or underpayment
This is reading one thing and writing another: taxpayers of any overpayment in, a record out. That is the shape today's tools are built for.
importance 4 · CoreSource: “Notify taxpayers of any overpayment or underpayment, and either issue a refund or request further payment.” (O*NET task statement)
How this row was scored
Exposure score: 75 out of 100 (71–79 allowing for uncertainty): high exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Working out an over- or underpayment is arithmetic against published rules, though releasing money still needs an authorized officer.
The five ratings: output a model can produce 4/4 · needs a body in a room 0/4 · needs an accountable person 2/4 · needs to be trusted in the moment 0/4 · how much data exists 4/4.
Maintaining records for each case
This is reading one thing and writing another: records in, a record out. That is the shape today's tools are built for.
importance 4 · CoreSource: “Maintain records for each case, including contacts, telephone numbers, and actions taken.” (O*NET task statement)
How this row was scored
Exposure score: 93 out of 100 (89–97 allowing for uncertainty): very high exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Keeping a case record of contacts and actions taken is exactly the sort of logging software does reliably.
The five ratings: output a model can produce 4/4 · needs a body in a room 0/4 · needs an accountable person 0/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Sending notices to taxpayers when accounts are delinquent
This is reading one thing and writing another: notices in, a record out. That is the shape today's tools are built for.
importance 5 · CoreSource: “Send notices to taxpayers when accounts are delinquent.” (O*NET task statement)
How this row was scored
Exposure score: 88 out of 100 (84–92 allowing for uncertainty): very high exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Spotting a delinquent account and issuing the standard notice is rule-driven paperwork that systems already generate automatically.
The five ratings: output a model can produce 4/4 · needs a body in a room 0/4 · needs an accountable person 1/4 · needs to be trusted in the moment 0/4 · how much data exists 4/4.
Answering questions from taxpayers and assisting them in completing tax forms
This is reading one thing and writing another: questions in, a record out. That is the shape today's tools are built for.
importance 4 · CoreSource: “Answer questions from taxpayers and assist them in completing tax forms.” (O*NET task statement)
How this row was scored
Exposure score: 61 out of 100 (57–65 allowing for uncertainty): high exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Tax rules and forms are published in enormous detail, so answering taxpayer questions is well suited to automated help.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 1/4 · needs to be trusted in the moment 1/4 · how much data exists 4/4.
Changing shape
8 tasksTasks where the machine takes the producing and a person keeps the checking, the deciding, or the answering-for-it. For most jobs this is the biggest group, and it is where "transformation, not termination" is literally visible.
Contacting taxpayers by mail or telephone to address discrepancies and to request supporting documentation
The software now makes the first pass at taxpayers, but the deciding part still needs a person. So the job becomes checking and deciding rather than producing.
importance 4 · CoreSource: “Contact taxpayers by mail or telephone to address discrepancies and to request supporting documentation.” (O*NET task statement)
How this row was scored
Exposure score: 56 out of 100 (52–60 allowing for uncertainty): partial exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Chasing missing documents by letter is easily automated; the phone calls add a human element but stay fairly routine.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 1/4 · needs to be trusted in the moment 1/4 · how much data exists 3/4.
Imposing payment deadlines on delinquent taxpayers and monitoring payments to ensure that deadlines are met
The software now makes the first pass at payment deadlines, but someone has to be answerable for the result, and it cannot be the software. So the job becomes checking and deciding rather than producing.
importance 4 · SupplementalSource: “Impose payment deadlines on delinquent taxpayers and monitor payments to ensure that deadlines are met.” (O*NET task statement)
How this row was scored
Exposure score: 56 out of 100 (49–63 allowing for uncertainty): partial exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Setting and tracking payment deadlines is straightforward system work, though the deadline itself is imposed under an officer's authority.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 2/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Examining and analyzing tax assets and liabilities to determine resolution of delinquent tax problems
The software now makes the first pass at tax assets, but someone has to be answerable for the result, and it cannot be the software. So the job becomes checking and deciding rather than producing.
importance 5 · SupplementalSource: “Examine and analyze tax assets and liabilities to determine resolution of delinquent tax problems.” (O*NET task statement)
How this row was scored
Exposure score: 56 out of 100 (49–63 allowing for uncertainty): partial exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Analyzing a taxpayer's assets and liabilities is financial analysis software does well, though the resolution is settled by an agent.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 2/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Staying human
5 tasksTasks that stay with a person, because they happen in the physical world, because the rules need someone accountable, or because the value is that a specific person does them.
Collecting taxes from individuals or businesses according to prescribed laws and regulations
The rules require a named, qualified person to answer for taxes, and that person cannot be a piece of software.
importance 5 · SupplementalSource: “Collect taxes from individuals or businesses according to prescribed laws and regulations.” (O*NET task statement)
How this row was scored
Exposure score: 23 out of 100 (19–27 allowing for uncertainty): low exposure, high confidence.
Why it sits in this group: the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Collecting tax is a legal power exercised by an authorized officer, sometimes in person, not simply a calculation.
The five ratings: output a model can produce 2/4 · needs a body in a room 1/4 · needs an accountable person 3/4 · needs to be trusted in the moment 1/4 · how much data exists 3/4.
Conferring with taxpayers or their representatives to discuss the issues
The rules require a named, qualified person to answer for taxpayers, and that person cannot be a piece of software.
importance 4 · CoreSource: “Confer with taxpayers or their representatives to discuss the issues, laws, and regulations involved in returns, and to resolve problems with returns.” (O*NET task statement)
How this row was scored
Exposure score: 30 out of 100 (23–37 allowing for uncertainty): low exposure, medium confidence.
Why it sits in this group: the same decision, made over and over; someone qualified has to answer for it; the value is that a specific person does it.
The rating behind it: Talking a taxpayer through a disputed return needs an authorized officer who can respond to argument and reach an agreement.
The five ratings: output a model can produce 2/4 · needs a body in a room 0/4 · needs an accountable person 2/4 · needs to be trusted in the moment 2/4 · how much data exists 3/4.
Securing a taxpayer's agreement to discharge a tax assessment or submitting contested determinations to other administrative or judicial conferees for appeals hearings
The rules require a named, qualified person to answer for a taxpayer's agreement, and that person cannot be a piece of software.
importance 4 · SupplementalSource: “Secure a taxpayer's agreement to discharge a tax assessment or submit contested determinations to other administrative or judicial conferees for appeals hearings.” (O*NET task statement)
How this row was scored
Exposure score: 22 out of 100 (15–29 allowing for uncertainty): low exposure, medium confidence.
Why it sits in this group: someone qualified has to answer for it; the value is that a specific person does it.
The rating behind it: Getting a taxpayer to agree to a settlement is a negotiation carried out under an officer's legal authority.
The five ratings: output a model can produce 2/4 · needs a body in a room 0/4 · needs an accountable person 3/4 · needs to be trusted in the moment 2/4 · how much data exists 2/4.
Show the other 11 tasks
Entering tax return information into computers for processing
shifting to AIThis is reading one thing and writing another: tax return information in, a record out. That is the shape today's tools are built for.
importance 4 · SupplementalSource: “Enter tax return information into computers for processing.” (O*NET task statement)
How this row was scored
Exposure score: 93 out of 100 (89–97 allowing for uncertainty): very high exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Keying return information into a processing system is routine data entry that software already handles end to end.
The five ratings: output a model can produce 4/4 · needs a body in a room 0/4 · needs an accountable person 0/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Checking tax forms to verify that names and taxpayer identification numbers
shifting to AIThis is reading one thing and writing another: tax forms in, a record out. That is the shape today's tools are built for.
importance 5 · SupplementalSource: “Check tax forms to verify that names and taxpayer identification numbers are correct, that computations have been performed correctly, or that amounts match those on supporting documentation.” (O*NET task statement)
How this row was scored
Exposure score: 88 out of 100 (84–92 allowing for uncertainty): very high exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Checking names, numbers and arithmetic against supporting documents is the clearest example of work software does faster and more accurately.
The five ratings: output a model can produce 4/4 · needs a body in a room 0/4 · needs an accountable person 1/4 · needs to be trusted in the moment 0/4 · how much data exists 4/4.
Maintaining knowledge of tax code changes
shifting to AIThis is reading one thing and writing another: knowledge of tax code changes in, a record out. That is the shape today's tools are built for.
importance 4 · SupplementalSource: “Maintain knowledge of tax code changes, and of accounting procedures and theory to properly evaluate financial information.” (O*NET task statement)
How this row was scored
Exposure score: 83 out of 100 (76–90 allowing for uncertainty): very high exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Tax law changes are published in full, so keeping up to date is well suited to automatic summarising.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 0/4 · needs to be trusted in the moment 0/4 · how much data exists 4/4.
Reviewing filed tax returns to determine whether claimed tax credits and deductions are allowed by law
shifting to AIThis is reading one thing and writing another: filed tax returns in, a record out. That is the shape today's tools are built for.
importance 4 · SupplementalSource: “Review filed tax returns to determine whether claimed tax credits and deductions are allowed by law.” (O*NET task statement)
How this row was scored
Exposure score: 62 out of 100 (58–66 allowing for uncertainty): high exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Deciding whether a claimed deduction is allowed is applying published tax law to figures, which software does well for ordinary cases.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 2/4 · needs to be trusted in the moment 0/4 · how much data exists 4/4.
Reviewing selected tax returns to determine the nature and extent of audits to be performed on them
changing shapeThe software now makes the first pass at selected tax returns, but someone has to be answerable for the result, and it cannot be the software. So the job becomes checking and deciding rather than producing.
importance 4 · SupplementalSource: “Review selected tax returns to determine the nature and extent of audits to be performed on them.” (O*NET task statement)
How this row was scored
Exposure score: 56 out of 100 (49–63 allowing for uncertainty): partial exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Selecting returns for audit and scoping the work is pattern analysis software handles well, subject to an agent's judgment.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 2/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Investigating claims of inability to pay taxes by researching court information for the status of liens
changing shapeThe software now makes the first pass at claims of inability, but someone has to be answerable for the result, and it cannot be the software. So the job becomes checking and deciding rather than producing.
importance 4 · SupplementalSource: “Investigate claims of inability to pay taxes by researching court information for the status of liens, mortgages, or financial statements, or by locating assets through third parties.” (O*NET task statement)
How this row was scored
Exposure score: 56 out of 100 (49–63 allowing for uncertainty): partial exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Tracing liens, mortgages and assets through public and third-party records is search work software does quickly and thoroughly.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 2/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Directing service of legal documents
changing shapeThe software now makes the first pass at service of legal documents, but someone has to be answerable for the result, and it cannot be the software. So the job becomes checking and deciding rather than producing.
importance 4 · SupplementalSource: “Direct service of legal documents, such as subpoenas, warrants, notices of assessment, and garnishments.” (O*NET task statement)
How this row was scored
Exposure score: 47 out of 100 (40–54 allowing for uncertainty): partial exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Preparing and directing service of legal documents is well documented, but issuing them is a legally restricted official act.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 3/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Examining accounting systems and records to determine whether accounting methods used were appropriate and in compliance with statutory provisions
changing shapeThe software now makes the first pass at systems, but someone has to be answerable for the result, and it cannot be the software. So the job becomes checking and deciding rather than producing.
importance 4 · SupplementalSource: “Examine accounting systems and records to determine whether accounting methods used were appropriate and in compliance with statutory provisions.” (O*NET task statement)
How this row was scored
Exposure score: 42 out of 100 (35–49 allowing for uncertainty): partial exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Examining accounting records against statute is document work software does well, though field audits mean visiting the business.
The five ratings: output a model can produce 3/4 · needs a body in a room 1/4 · needs an accountable person 2/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Determining appropriate methods of debt settlement
changing shapeThe software now makes the first pass at appropriate methods of debt settlement, but someone has to be answerable for the result, and it cannot be the software. So the job becomes checking and deciding rather than producing.
importance 5 · SupplementalSource: “Determine appropriate methods of debt settlement, such as offers of compromise, wage garnishment, or seizure and sale of property.” (O*NET task statement)
How this row was scored
Exposure score: 40 out of 100 (33–47 allowing for uncertainty): partial exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Choosing between a payment plan, garnishment or seizure follows documented criteria, but the decision is legally an officer's to make.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 3/4 · needs to be trusted in the moment 1/4 · how much data exists 3/4.
Preparing briefs and assisting in searching and seizing records to prepare charges and documentation for court cases
staying humanThis work happens in the physical world: briefs, in a real place. Software cannot follow it there.
importance 4 · SupplementalSource: “Prepare briefs and assist in searching and seizing records to prepare charges and documentation for court cases.” (O*NET task statement)
How this row was scored
Exposure score: 22 out of 100 (15–29 allowing for uncertainty): low exposure, medium confidence.
Why it sits in this group: the same decision, made over and over; work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Drafting the brief can be largely automated, but searching premises and seizing records is physical work with legal authority.
The five ratings: output a model can produce 2/4 · needs a body in a room 2/4 · needs an accountable person 2/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Participating in informal appeals hearings on contested cases from other agents
staying humanThe rules require a named, qualified person to answer for informal appeals hearings, and that person cannot be a piece of software.
importance 4 · SupplementalSource: “Participate in informal appeals hearings on contested cases from other agents.” (O*NET task statement)
How this row was scored
Exposure score: 11 out of 100 (4–18 allowing for uncertainty): minimal exposure, medium confidence.
Why it sits in this group: someone qualified has to answer for it; the value is that a specific person does it.
The rating behind it: An appeals hearing is a live legal proceeding where an authorized officer argues the case and answers questions.
The five ratings: output a model can produce 1/4 · needs a body in a room 1/4 · needs an accountable person 3/4 · needs to be trusted in the moment 2/4 · how much data exists 2/4.
What this job pays, and how many people do it
- Median pay
- $62,370a year, the middle of the range, so half earn more and half earn less.bls-oews, 2025 · May 2025 estimates (national_M2025_dl.xlsx)
How we know this
Source: bls-oews
Reference period: May 2025 estimates (national_M2025_dl.xlsx)
Rounding: Shown as published.
- People doing this job
- 56,610in the US, 2025.bls-oews · May 2025 estimates (national_M2025_dl.xlsx)
What is deliberately not here: a forecast of how many of these jobs exist in ten years. Where an official projection exists for a market we publish it with its vintage; where it does not, we leave the space empty rather than borrow the other country’s number.
Why this is shifting
The reason is boringly specific. Most of what is shifting here is reading one thing and writing another: taxpayers of any overpayment in, a record out. The rows above are exactly that shape: maintaining records for each case and notifying taxpayers of any overpayment or underpayment. What it cannot do is be answerable: taxes need a named person the rules will accept, and software cannot be that person. Which is why this page talks about your tasks changing, not your job ending.
Your move
Over a pint: what I’d tell you if you were my friend
The exposed part of your job is the biggest part, and I am not going to dress that up: maintaining records for each case is work today's tools do quickly and cheaply, and that is not coming back.
So, given all that: 46% of this job's task weight sits in rows the software is already learning, 34% in rows that change shape rather than disappear, and 20% in rows it is nowhere near. That is the position, measured across 21 scored tasks. It is not a forecast about you.
What you have that the software does not is collecting taxes from individuals or businesses according to prescribed laws and regulations, plus the years of knowing when something looks wrong before you can say why. That is the raw material for everything below.
This week: one thing
Sit on the machine's side of the desk. Pick one real piece of taxpayers of any overpayment you would normally do yourself, let whatever software you already have take the first pass at it, and then go through what it produced and write down every single thing it got wrong. One evening this week. Do not fix anything yet. Just catch it.
- What you end up holding
- a written list of the machine’s mistakes, in your handwriting
- How long it takes
- an evening, or an hour if you pick one job rather than one client
If you can’t run software on your employer’s or your clients’ data: Do the same hunt on paper. Take one printed piece of taxpayers of any overpayment, work through it the way you always do, and mark every point where you made a call rather than followed a rule. Same evening, same list, nothing to log into and nobody to ask permission from. That list is the same artifact: it is the judgment written down.
Over the next 90 days
Change one sentence about what you do. Not on a CV. Out loud, to whoever pays you. From “my job is maintaining records for each case” to “I check what the software does and tell you what it means.” Your error list from this week is what makes that sentence true instead of a claim, so use it: show it once, to one person, inside the next ninety days. Same skills, priced as judgment rather than as typing.
Over the next 12 months
Walk toward the end of this job that answers for things, and get it recognised. Pick the one part of collecting taxes from individuals or businesses according to prescribed laws and regulations you are already best at, and spend the year making it formal: a qualification, a named responsibility, a specialism people ask for by name. Price it honestly: that is evenings, it is months rather than weeks, and the seats are competitive because everyone in your position is looking at the same door. Before you pay for anything, use CareerOneStop - Find local training. It is free, it is the Labor Department's own service, and it is listed below with the rest of the free routes.
The roads out of here, and why I am not sending you down them
I looked at the obvious moves out of this job, and here is what I found.
I checked the 12 nearest US occupations to tax examiners and collectors, and revenue agents (nearest by the work that AI is not taking, not by job title), and none of them survived. The closest was tax preparers: only about 11% of its durable work is work you already do, it is under the same pressure this job is and it pays 11.9% less. Your own job splits about 46/54: that share of the list sits in the top exposure band and the rest does not. On this evidence the honest move is inside the job rather than out of it. Become the person who owns “contact taxpayers by mail or telephone to address discrepancies and to request…”, and let the exposed end go.
How that was checked: this job was compared against all 830 US occupations in this release on their official task statements, and the 12 nearest were examined one by one. A move that turns on an industry, an employer or a qualification rather than on the work itself will not show up in a check like that. And this release carries no licence register, so anything you are weighing needs that looked up separately.
3 moves I checked and rejected
These are the obvious-looking jumps. They are here with their reasons rather than quietly dropped, because the ones that fail are worth knowing about. It is one less thing to turn over at night.
Tax Preparers
Why it looked obvious: It came up as a near neighbour on the work AI is not taking: you already answer questions from taxpayers and assist them in completing tax forms, and their equivalent is to furnish taxpayers with sufficient information and advice to ensure correct tax form completion. Across both published task lists that is about 11% of the durable work in that job.
Why I am not recommending it: You would be starting most of it from nothing: about 11% of the durable work in that job is work you do today, and the rest you would be learning while the bills carried on. I will not move you off one melting floe onto another: 56% of its own task list already scores in the top exposure band (65/100 in this release), so the same software is eating it. It is a pay cut, in those words: $54,920 against your $62,370, 11.9% less (OEWS May 2025 (both)). Retraining to earn less is a decision, not advice.
Credit Analysts
Why it looked obvious: It came up as a near neighbour on the work AI is not taking: you already send notices to taxpayers when accounts are delinquent, and their equivalent is to contact customers to collect payments on delinquent accounts. Across both published task lists that is about 3% of the durable work in that job.
Why I am not recommending it: Almost none of it is work you already do: about 3% of the durable side of that job. That is a different job, not a next step. I will not move you off one melting floe onto another: 78% of its own task list already scores in the top exposure band (70/100 in this release), so the same software is eating it.
Securities, Commodities, and Financial Services Sales Agents
Why it looked obvious: It came up as a near neighbour on the work AI is not taking: you already examine and analyze tax assets and liabilities to determine resolution of delinquent tax…, and their equivalent is to interview clients to determine clients' assets, liabilities, cash flow, insurance coverage, tax status…. Across both published task lists that is about 3% of the durable work in that job.
Why I am not recommending it: Almost none of it is work you already do: about 3% of the durable side of that job. That is a different job, not a next step. I will not move you off one melting floe onto another: 56% of its own task list already scores in the top exposure band (62/100 in this release), so the same software is eating it.
What I’d stop worrying about
A friend tells you what not to spend fear on. This is that list.
The headline number you read somewhere
The big “X% of jobs” figures are about the whole economy, not about you. The number that describes your job is on this page: 46% of its task weight, across 21 scored tasks. Every row behind it is printed above with the source. Check ours; ignore theirs.
“It’s too late for me to become something else”
You are not starting from zero, and the page shows why: collecting taxes from individuals or businesses according to prescribed laws and regulations is work the software cannot do and you already do it. The move above is a repricing of what you know, not a new career. Nobody who has just left college has that.
“I should learn to code”
Almost certainly not. The value in your job is moving toward checking, deciding and answering for the output, not toward writing the software. A weekend of Python will not change your position; the error list from this week will.
The “obvious” next job everyone suggests
I checked the obvious moves and most of them did not survive. The reasons are printed with the routes above, including the pay and the gate. A move that fails on the numbers is worth knowing about so you can stop turning it over at night.
You are reading the United States figures
The United Kingdom splits this work across more than one official group, of which Taxation experts is the closest. The pay and employment figures are not directly comparable, and we do not average them together.
Switch to the United Kingdom page →partial match
The other groups this work is counted across:
In UK official statistics this job is counted as Taxation experts and Financial administrative occupations n.e.c.. Pay is shown separately for each of those groups (medians cannot be averaged together), while the task list and the scores on this page are for this group only.
Your route through this
Two honest options, and no deadline on either
Free, and complete
The moves above cost nothing. These are the real services that go with them: public, government-funded, and free at the point of use. Nothing on this page is behind an email address or a payment.
Anywhere in the US:
CareerOneStop - Find local training
Search what's running near you, from the Labor Department's own database, before anyone sells you a course.
Free to search; individual programs vary, and some are funded
Anywhere in the US:
An American Job Center will sit down with you for free. Find yours by ZIP code.
Free
Anywhere in the US:
CareerOneStop - Licensed occupations finder
Check what your state actually requires before you pay for anything.
Free
Anywhere in the US:
Free
A nearby route
There's no Space built for tax examiners and revenue agents yet.
Collab365 Spaces is built by the same people as this site. We find the problems that AI and automation are creating inside one kind of work, then solve them as short courses, briefings and Blueprints. Each Space is the community too, so the research and the people doing your job are in the same place.
The closest match is The AI Authority, a community for non-technical managers and domain experts turning one-off AI prompts into workflows a team can trust. It overlaps with the part of your job that is growing: checking AI-drafted assessments and write-ups before they carry your name, and deciding what can safely be put into an AI in the first place. It covers no tax law and nothing about the systems you assess in. If that overlap isn't you, the free route below covers the same ground.
- Problem: “I can use AI, but I can’t turn it into a workflow my team can trust”
- Problem: “I can’t hand off AI work without it falling apart”
7 days free, no card needed. Explore up to 2 Spaces before you choose a plan: you pick a plan later, not now.
After the trial it is a paid community, and you get identical data either way. If the overlap above is not your job, the moves above cost nothing and stand on their own.
Noted, and thank you. We’ll email you if a Space for tax examiners and revenue agents launches. Nothing else.
That did not look like an email address, so nothing was saved. Have another go below.
We could not save that. The fault is ours, not yours, and your address was not stored. Please try again later.
No deadline on any of this. The page will still be here, and the data is refreshed on a published schedule rather than when someone wants a headline.
Questions people ask about this job
- Will AI replace Tax Examiners and Collectors, and Revenue Agents?
- Not as a job, but it is already doing parts of the work. Across the 21 official task statements scored for Tax Examiners and Collectors, and Revenue Agents (United States, SOC 13-2081), 46% of the importance-weighted core work is made of tasks today's AI could already do most of. The overall exposure score is 59 out of 100 (range 54–65, band: partial). That is a statement about tasks, not about headcount: this measures what AI could do, not whether any employer adopts it, whether the law allows it, or whether doing the routine parts faster creates more demand for the human parts. Figures are from release 2026-q4.1.
- Which tasks in “Tax Examiners and Collectors, and Revenue Agents” can AI already do?
- The highest-scoring tasks in release 2026-q4.1 are: “Maintain records for each case, including contacts, telephone numbers, and actions taken” (93/100, very high); “Enter tax return information into computers for processing” (93/100, very high); “Send notices to taxpayers when accounts are delinquent” (88/100, very high). Each score comes from five published 0–4 ratings turned into a number by a published formula, and each carries the model's one-sentence reason on the page.
- Which tasks in “Tax Examiners and Collectors, and Revenue Agents” stay human?
- About 20% of this job's task weight sits in work that scores low for AI exposure. The lowest-scoring tasks in release 2026-q4.1 are: “Participate in informal appeals hearings on contested cases from other agents” (11/100, minimal); “Prepare briefs and assist in searching and seizing records to prepare charges and documentation for court cases” (22/100, low); “Secure a taxpayer's agreement to discharge a tax assessment or submit contested determinations to other administrative or judicial conferees for appeals hear…” (22/100, low). Low scores usually mean the task needs a body in a room, a legally accountable human, or trust built in real time. Those are the three things the scoring rubric treats as gates rather than obstacles.
- What should someone working in “Tax Examiners and Collectors, and Revenue Agents” do about AI?
- Start from the ledger rather than the headline: 46% of this job's weighted core work is exposed, and roughly 20% is not. The practical move is to spend more of your week on the tasks that score low, the ones above, and to get fluent at directing AI through the tasks that score high, because those are the parts that change whether or not you are ready for them. This page does not predict your job, and nothing here is career advice tailored to you: the score describes the occupation, not the person.
- How is the AI exposure score for Tax Examiners and Collectors, and Revenue Agents calculated?
- Each official task statement for the occupation is rated on five published 0–4 dimensions (output replicability, physical embodiment, licensed accountability, real-time human trust, and data availability) by claude-opus-5 using scoring prompt task_scoring_v1.0. The model never writes the score; a published formula turns the five ratings into a 0–100 number, so every score can be recomputed by hand. The occupation figure is the importance-weighted mean across 21 scored tasks. The prompt, the rubric, the formula and the full dataset are published at https://futureproof.collab365.com/method and https://futureproof.collab365.com/data/2026-q4.1 under CC BY 4.0.
Where these numbers come from
Worth knowing about these figures
- The match between this job and its counterpart in the other country is partial, so the two pages count slightly different groups of people.
- 2 of this occupation's scored task statements carry a score that was measured under a different occupation's context, because the statement is shared between jobs and has only been scored once so far. Each one names the occupation it was measured under in the free routes below; none is presented as a measurement for this job.
- Task clusters are not derived in this release, so the task-cluster field is empty and no Collab365 Space routing is attached to this occupation yet.
- Task statements
- onet-dbProcessing: catalogue-bridge → onet-im-rt-weighting → task-scoring → score-aggregation
- Task weights
- onet-db (im-rt)
- Scores
- Rubric task_scoring_v1.0, model claude-opus-5, computed 2026-08-05.
- Pay and employment
- bls-oews (May 2025 estimates (national_M2025_dl.xlsx))bls-oews (May 2025 estimates (national_M2025_dl.xlsx))
Figures on this page come from release 2026-q4.1, published 2026-08-05. Every release keeps its own permanent address, so a figure you cite in March is still there, unchanged, in November.
The plain-English wording on this page is assembled directly from the task statements and the published ratings, not written by hand for this occupation. That is why it is specific, and it is also why we say so.
The routes and free resources further up are today’s, not the release’s (last reviewed 2026-08-05). A route is an offer, not a historical fact, so it moves on its own clock.
Using these figures?
Cite this
Everything on this site is published under CC BY 4.0. Quote it, chart it, sell something built on it. Just say where it came from, and cite the dated release rather than the site, so the figure you quote stays checkable.
Plain text
Collab365 (2026). Collab365 Futureproof: task-level AI exposure for US and UK occupations, release 2026-q4.1 (methodVersion 2.0.0, promptVersion task_scoring_v1.0). https://futureproof.collab365.com/data/2026-q4.1. Licensed CC BY 4.0. Built with O*NET data (USDOL/ETA, CC BY 4.0); ONS data (Open Government Licence v3.0); GAISI task framework (arXiv:2507.22748, MIT); BLS data (public domain).
BibTeX
@misc{collab365futureproof2026q41,
title = {Collab365 Futureproof: task-level AI exposure for US and UK occupations, release 2026-q4.1},
author = {{Collab365}},
year = {2026},
url = {https://futureproof.collab365.com/data/2026-q4.1},
note = {Release 2026-q4.1, methodVersion 2.0.0, promptVersion task_scoring_v1.0, CC BY 4.0}
}Data as of release 2026-q4.1, published . Releases never change after publication; when the figures move, a new dated release is published beside this one and this one stays exactly where it is.