EXPOSED
This catch-all bucket — treasury analysts, financial risk specialists, pricing and compliance analysts, benefits and grants finance staff — is dominated by spreadsheet modeling, reconciliation, variance reporting, and memo-writing, all of which current AI drafts at usable quality with a human reviewing. The modal worker holds no license that legally requires a signature, so the surviving work is the part where someone assembles ambiguous evidence, makes a capital, pricing, or risk call, and answers for it to a committee or regulator. Employment concentrates upward: fewer analysts producing packets, more specialists owning positions and defending assumptions.
Headcount grew steadily across the period.
This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.
So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing.
BLS projection, 2024–2034
+3.1% 137,100 → 141,400 on the projections basis
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +3.1% more of these jobs by 2034. Demand for the output is growing faster than the work is being automated away — the mechanism BLS gives for software developers, and the combination people most often misread as an error.
Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.
~10,300 openings a year on average, including replacing people who leave.
PurserAnalystAdjusterBondsmanBail AgentStructurerShip PurserData AnalystEscrow AgentInvestigatorRate AnalystBail BondsmanEscrow CloserFraud AnalystRate EngineerEquity AnalystFraud ExaminerFutures TraderBail Bond AgentFinance AnalystFraud SpecialistResearch AnalystEquity StructurerFinancial Analyst
The BLS uses Financial Specialists, All Other for work that doesn't fit any named occupation, so it covers roles that have little in common with each other. Two consequences worth knowing before you read anything below:
Holding it up: judgment & accountability . Weakest point: embodiment .
Mixed — a routine tier and a judgment tier Pulling bank feeds into a cash position, tying subledger to GL, building a rate-sensitivity table, and drafting the variance narrative are all things a model does at first-draft quality today — what holds the score at 8 rather than 4 is the recurring work of chasing down why a counterparty confirmation doesn't match, arguing an assumption with a business unit head, and deciding what the packet should say when the data is late or wrong.
Fully desk- and screen-based The job is a laptop, a market data terminal, and a recurring committee meeting; the only physical element is walking a signature page or sitting in the room when the treasurer approves a hedge, which is why this is 3 and not 0.
Certification preferred, not legally required Nothing in this bucket requires a state license to perform — the CFA, CTP, or FRM your employer put on the job posting is a hiring filter, not a statutory one, and when a hedge accounting treatment or a grant cost allocation gets challenged it is the CFO, controller, or an external auditor whose name is on the attestation, not yours.
Meaningful discretion Setting a transfer price, sizing a liquidity buffer, calling an allowance assumption, or deciding a cost is unallowable under 2 CFR 200 are genuinely contestable calls you defend in a meeting — but at 11 rather than 16 because a policy, limit framework, or delegated authority almost always caps your discretion and someone above you signs the final position.
The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.
Your task mix speaks to task resistance (8/20 here) — how much of the day's work current AI already does. That is the dimension the boxes above are about.
It cannot move the other three. Liability shield (5/20) is whether the law requires a licensed human to sign. Trust premium (9/20) is whether buyers specifically pay for a person. Judgment and accountability (11/20) is whether the role exists to own consequential calls. Those are facts about the occupation's standing, not about which tasks are in your week — a paralegal who does only trial exhibits still holds no licence. Together they are 25 of this occupation's 36 points (69%).
Embodiment (3/20) is also a property of the work rather than the worker, but we don't tag individual tasks as physical or not, so the picker can't tell you anything about it. That's a limit of this tool, not a claim.
Did we get the list right? Tell us what's missing — the tasks are written from the outside, and you're reading this from the inside.
The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 49/100, still EXPOSED.
Bank/insurer model-risk regimes extending named-individual accountability to AI-generated analytics: e.g. an SR 11-7 / OCC model risk management update or a state DFS circular (NY DFS already issued 2024 guidance on AI in insurance underwriting) requiring a named model owner and independent validator to personally attest to each AI-assisted pricing or capital model before use. Also Sarbanes-Oxley 302/404 sub-certification cascades: if audit committees require a named specialist to sub-certify AI-produced reconciliations and reserve estimates, the signature becomes a job function.
Task-mix shift plus formal ownership: as drafting and reconciliation are automated, the residual role is defending assumptions to ALCO, pricing committees, rating agencies, or a regulator's exam team. This rises further if institutions formalize 'AI output challenger' roles — a documented human dissent record required in model validation files, as EU AI Act Art. 14 human-oversight obligations demand for high-risk credit and insurance systems.
Genuine two-tier structure: routine tier (variance packets, reconciliations, grant drawdown schedules) automates; residual tier (novel instrument valuation with no comparables, distressed counterparty judgment, reserve setting under litigation uncertainty, defending a transfer-pricing position under audit) resists because inputs are non-public, contested, and adversarial. Score rises mechanically as the routine tier disappears from the day, not because AI got worse.
Narrow route only: grant-funded and public-sector finance where funders (federal agencies under 2 CFR 200, foundations) require a named human preparer for cost allocation and single-audit certifications. Outside that, buyers of internal analytics are employers, not clients, and there is no realistic route to paying extra for a human.
The limit. Embodiment has no route; this is screen work. The structural problem is that this SOC bucket is defined by residual, not by a profession — there is no board, no exam, no title protection to hang a liability shield on, so any gain has to come from firm-level or supervisory attestation rules that name individuals. Those rules concentrate accountability in a smaller number of senior specialists rather than protecting headcount, so the occupation's score can rise while employment falls.
| New York-Newark-Jersey City, NY-NJ | 11,950 | $109,580 +35% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 5,820 | $120,580 +49% |
| Chicago-Naperville-Elgin, IL-IN | 4,630 | $81,740 +1% |
| Dallas-Fort Worth-Arlington, TX | 4,440 | $72,330 -11% |
| Los Angeles-Long Beach-Anaheim, CA | 4,100 | $80,840 +0% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 3,050 | $84,680 +4% |
| San Francisco-Oakland-Fremont, CA | 2,880 | $78,410 -3% |
| Denver-Aurora-Centennial, CO | 2,820 | $88,090 +9% |
| Lexington Park, MD | 720 | $132,620 +64% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 5,820 | $120,580 +49% |
| Chambersburg, PA | 40 | $111,530 +38% |
We have no reported case of a named organisation automating this occupation. Not one deployment, not one announcement.
That is worth saying out loud next to a score of 36. The verdict above is about what the work exposes — what current AI could do to these tasks. It is not a claim that anyone has done it. For this occupation those two things have come apart completely: the capability argument is on this page, and the evidence column is empty.
Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.
Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.