A financial analyst’s job has always involved more than crunching numbers. Someone has to decide which figures matter, question the assumptions behind them, explain what's changed, and stand behind their recommendations.
AI is getting better at helping with the work that leads up to those decisions. It can summarize documents, refresh models, surface variances, and accelerate recurring analysis. So, will AI replace financial analysts, or will it merely change how they work?
Current evidence points more strongly toward the second outcome. Bounded tasks with clean inputs are easier to automate, but complex workflows still rely heavily on expert oversight.
What is the current state of AI automation in finance?
While AI in finance and accounting is moving beyond rule-based automation, there's still a gap between automating individual tasks and replacing an analyst entirely.
Finance teams use generative systems to draft commentary, summarize performance, surface variances, and support scenario modeling. Agentic systems aim to carry work across several steps and tools.
In a 2025 survey of 102 CFOs, McKinsey reported that 44% used generative AI for more than five use cases, up from just 7% in the prior year's survey. But adoption alone doesn't prove end-to-end reliability.
Role replacement would require a system to own a broader workflow, including exceptions, review, communication, and consequences. Finance's sensitive data, audit requirements, and high cost of error make that final step much harder to achieve. Mercor's guide to AI use cases in finance shows how current applications are concentrated around performing defined tasks and workflows rather than fully replacing analysts.
Which financial analysis tasks can AI automate or accelerate?
Task acceleration reduces the time an analyst spends on work, while task automation completes defined activities with limited intervention. AI performs best when inputs are clear, instructions are explicit, outputs are measurable, and a reviewer can quickly determine whether results are correct.
AI can extract specific fields from structured documents, transform data into a defined format, and apply formulas when supplied with the required inputs and calculation method. It can also assemble first drafts of management commentary from approved source material, summarize sets of documents, or surface variances that deserve investigation. In recurring workflows, it can reproduce a report or analysis when the process, tools, and acceptance criteria are stable.
These capabilities can materially increase analyst capacity without making every output final. A first-pass variance explanation may overlook a business event that's absent from the source data, for example, while a calculation may be mathematically correct but built on the wrong period, unit, or assumption. The strongest workflow, therefore, pairs automation with a review step that's designed to catch the mistakes most likely to occur.
What makes financial tasks easier for AI to handle?
The tasks AI handles most consistently tend to share several characteristics. Clear inputs, defined instructions, and outputs that analysts can easily verify reduce the number of hidden decisions the system has to make:
- Clean, machine-readable inputs: Values scattered across charts, footnotes, and image-only pages are harder for AI to process. For example, AI can more reliably pull revenue figures from a structured spreadsheet than from a scanned investor presentation.
- Explicit instructions: Defining the operation, assumptions, period, units, and required output reduces ambiguity. A well-defined request might ask the system to calculate year-over-year growth using a specified reporting period and return the result in a standard format.
- Limited tool use: Fewer handoffs between documents, spreadsheets, and business systems make a workflow easier to automate. A recurring report built from one approved data source is easier to automate than a workflow that requires switching between several platforms and reconciling conflicting information.
- Clear calculations and standards for a correct result: Established reference values make errors easier to detect. Common use cases include applying a defined formula, checking whether totals reconcile, or comparing an output against an expected value.
- Easy verification: Analysts can more easily trace sources, reproduce calculations, and inspect changes when outputs are transparent. For instance, an AI-assisted variance analysis is easier to review when each figure links back to its source and the underlying calculations are visible.
Clear constraints like these mean a bounded financial analyst AI task may work well even when an entire analyst workflow does not.
Which financial analysis workflows does AI still struggle with?
AI's advantages begin to weaken when financial work involves more steps, is less predictable, and is more dependent on context. When workflows require more source materials, tools, exceptions, and judgment calls, the opportunities for error increase.
Complex financial documents and source materials
Filings, investor presentations, and spreadsheets often distribute evidence across tables, charts, footnotes, and visual layouts. In a Mercor finance evaluation, frontier models were tested on real-document tasks involving earnings reports, investor decks, and regulatory materials.
The research documented two recurring failure modes: misreading values in dense visual documents and applying the wrong financial operation even when the relevant inputs were available.
Multistep analysis and spreadsheet workflows
Work with more steps requires a system to preserve context, choose the right operation, update the correct cells, reconcile outputs, and remain consistent across applications. The APEX-Agents investment banking benchmark tests this kind of multi-hour, cross-application work. It's a useful reminder that doing well on a limited task is different from producing work that's ready for professional review.
Assumptions, exceptions, and judgment
Analysts decide whether an assumption is reasonable, detect when an event makes a historical pattern less useful, and recognize when a standard method shouldn't apply. Those decisions depend on business context that may be incomplete, changing, or absent from the model's inputs.
Verification and accountability
Consequential work needs an owner. Someone must validate sources, audit formulas, challenge conclusions, explain uncertainty, and take responsibility for recommendations. Current systems can support those activities, but they don't remove the need for accountable review.
Which financial analyst roles are most likely to change?
Because AI handles some parts of financial work more reliably than others, its impact is likely to show up within roles before it eliminates them entirely. Tasks involving repetitive information processing are likely to be automated across the finance industry, while business judgment and accountability remain central.
As Mercor's July 2026 analysis of AI's impact on jobs explains, benchmark capability doesn't automatically equal job displacement. How the shift plays out depends on the mix of tasks within each role and how much of the work still requires judgment, context, and accountable review.
| Role | Tasks AI may accelerate or automate | Work that still requires human judgment | Likely workflow redesign |
|---|---|---|---|
| FP&A analyst | Report assembly, forecast refreshes, variance surfacing, first-pass scenarios | Assumption setting, business partnering, management recommendations | Less manual preparation; more validation and decision support |
| Investment banking analyst | Bounded research, valuation steps, document review, first-pass models | Deal context, spreadsheet integrity, reconciliation, review-ready output | Faster drafting with tighter expert review across the deal workflow |
| Credit and risk analyst | Data synthesis, monitoring, first-pass risk signals | Exceptions, explainability, policy interpretation, accountable approval | Automated triage with escalation to analysts |
| Treasury and corporate finance analyst | Cash-position reporting, sensitivity analysis, recurring market-data synthesis | Liquidity decisions, counterparty judgment, controls, novel shocks | More continuous monitoring with human-led decisions |
FP&A analysts
Financial planning and analysis (FP&A) work may shift most noticeably in recurring reporting and forecasting. AI can handle more of the preparation behind variance analysis, report assembly, forecast refreshes, and first-pass scenarios, giving analysts more time to investigate what's changed and why.
The analyst’s value remains in setting assumptions, interpreting business drivers, working with operating teams, and using the analysis to support management recommendations. Over time, the role may place greater emphasis on validation, explanation, and decision support.
Investment banking analysts
Bounded research, valuation steps, document review, and first-pass modeling are all areas where AI can reduce manual preparation work. Analysts can then spend more of their time checking model integrity, reconciling inconsistencies, and making sure each deliverable reflects the full deal context.
Deal-specific judgment and review-ready output still require close oversight. Automated drafting may speed up work, but it doesn't remove the need for analysts to connect each piece of the analysis to the transaction as a whole.
Credit and risk analysts
Automated triage may become one of the biggest changes in credit and risk work. AI tools can synthesize data, monitor activity, and surface first-pass risk signals across larger volumes of information.
Analysts can then focus more heavily on exceptions, policy interpretation, unusual cases, and decisions that require explanation or approval. Human judgment remains especially important when the available evidence doesn't fit neatly within standard rules.
Treasury and corporate finance analysts
Routine treasury work is easier to automate when inputs, calculations, and reporting requirements remain stable. AI can speed up cash-position reporting, recurring market-data synthesis, and sensitivity analysis under those conditions.
Unusual market events, liquidity pressures, counterparty concerns, and novel shocks are much harder to standardize. Treasury and corporate finance analysts may therefore spend less time compiling routine information and more time interpreting changes, weighing trade-offs, and making financial decisions.
What skills will financial analysts need as AI changes the role?
As AI changes the work behind routine preparation, the skills that distinguish strong analysts will shift with it. The future analyst will be a hybrid professional: strong in finance and capable of directing, evaluating, and improving AI-assisted work. The following skills will be important to the role:
- Financial judgment: Challenge assumptions and recognize outputs that are technically polished but economically implausible.
- AI literacy: Understand how model, context, prompt, and tool choices affect performance and risk.
- Verification and evaluation: Check sources, calculations, citations, spreadsheet changes, and conclusions.
- Workflow management: Break complex work into smaller tasks and decide where automation, AI, human review, and escalation belong.
- Communication and business partnering: Use analysis to inform recommendations, explain uncertainty, and work with decision-makers.
- Exception handling: Identify situations where standard processes or AI-generated answers shouldn't be trusted.
Prompting can help, but it's not a core career strategy. Analysts create more value when they can define problems, test workflows, and judge whether outputs are fit for decisions.
Why do financial analysts remain central to financial decision-making?
The end goal of financial analysis is a decision, not an output. Analysts have to determine whether the numbers make sense in context, challenge assumptions, explain uncertainty, and connect findings to business recommendations.
Their responsibilities include:
- Validating source data
- Interpreting ambiguity
- Challenging assumptions
- Managing exceptions
- Understanding business context
- Balancing competing objectives
- Communicating recommendations to accountable decision-makers
Current AI systems can support many of these activities, but they don't reliably combine them across an end-to-end financial workflow. Human ownership remains necessary when work requires judgment, context, communication, and accountability for recommendations.
How should finance leaders prepare teams for AI?
If human ownership remains necessary, finance leaders need to decide where automation fits and where expert control is still needed. Preparation should focus on establishing evidence and setting controls rather than replacing job titles:
- Inventory tasks, not job titles: Map volume, cycle time, input format, judgment level, and cost of error. Prioritize tasks where the business value is high and the outcomes are measurable.
- Choose one bounded pilot: Select a repeatable internal workflow with a known expected outcome and a reversible failure mode. Success means a clear improvement without unacceptable correction burden.
- Build a workflow-specific evaluation set: Test representative documents, spreadsheets, exceptions, and edge cases, not only clean prompts. Measure whether the system meets the team's quality standards.
- Define review and escalation controls: Assign accountable owners, confidence thresholds, audit trails, data-access rules, and rollback paths before deployment.
- Redesign roles and train analysts: Shift time from manual assembly to validation, interpretation, stakeholder communication, and scenario judgment while preserving apprenticeship opportunities.
- Measure business value and failure rates together: Track time saved, correction burden, exception rate, and downstream decision quality before scaling.
Public evidence from the APEX family of professional AI benchmarks can inform what teams should test, but it's not a substitute for evaluation of a company's own data, standards, tools, and failure conditions.
The bottom line: Will AI replace financial analysts?
AI can take over specific financial-analysis tasks and significantly change how analysts work, but current evidence doesn't suggest that it can reliably perform the full range of responsibilities financial analysts handle.
The dividing line is human responsibility for the full workflow. AI can accelerate defined tasks with clear inputs and measurable outputs, but analysts still need to validate evidence, interpret unclear information, exercise judgment, communicate recommendations, and take responsibility for important decisions.
Some teams may spend fewer hours on manual preparation, and some roles will change substantially. But asking whether AI will take over financial analyst jobs ignores the more useful management question: Which tasks can be redesigned safely, and when must human expertise remain in the loop?
Evaluate AI on your real finance workflows
Before automating analyst work, test AI on representative tasks, documents, and exceptions. Mercor helps teams evaluate their finance workflows against required quality and cost thresholds using expert-designed tasks and criteria.
Build your finance eval