An investment bank analyst can spend a large part of a week searching filings, updating spreadsheets, preparing slides and checking a presentation for inconsistencies. AI can assist with each of those activities. It is therefore unsurprising that bankers are asking what comes next.
Yet a bank also depends on regulated decisions, confidential information, relationships and people who can defend a conclusion under scrutiny. The important distinction is between accelerating the preparation of work and taking responsibility for it.
Which banking tasks are changing first?
Research preparation is an obvious example. An approved AI tool can help structure a company profile, extract facts from public reports or compare business models.
Presentation work is another. A system may suggest a slide story, rewrite dense explanations or draft commentary around an analysis. A banker still has to establish whether the content is true, appropriate for the audience and consistent with the numbers.
In operational teams, classification, document processing, customer support and exception handling may change as well. This makes the impact relevant beyond front-office deal teams.
Will fewer analysts be needed?
Potentially, but the answer will differ between firms and teams. A group that can complete routine preparation faster may alter its staffing model or the amount of work it accepts. The result could be fewer hires, different responsibilities, more capacity or some combination.
The World Economic Forum's 2025 projections anticipate significant labour-market churn from several forces, including technology. Those estimates are not a forecast of specific redundancies at a particular bank.
A more reliable signal is what your own team is actually changing: which tools are approved, which processes are being redesigned, what quality standards apply and whether hiring plans have shifted.
The problem with blindly trusting an AI-generated model
Financial analysis is full of hidden assumptions. A projection might be mathematically correct but based on the wrong fiscal year, an inconsistent currency or a non-recurring cost mistakenly treated as permanent.
If AI produces a valuation narrative, trace every number to the workbook and every material external claim to an identified source. Verify formulas, units, dates, definitions and the logic between the analysis and the recommendation.
An authoritative paragraph without that work can be more dangerous than a visibly unfinished draft.
One useful exercise for an analyst or associate
Choose a listed company's public annual report and a permitted AI tool.
- Produce a source-backed business overview with exact page references.
- Create a list of three performance drivers and three questions the report does not answer.
- Check each claim against the original filing.
- Draft a one-page recommendation identifying evidence, assumptions and open points.
- Ask a more experienced colleague to challenge the reasoning, not just the formatting.
The benefit is not that you finished a slide in half the time. It is that you could spend more of the time interrogating what matters.
Where career value may move
People who can connect financial analysis with a real commercial decision will remain important. So will people who understand risk, stakeholder incentives, regulation and the practicalities of execution.
Junior bankers should still learn accounting, modelling and transaction mechanics. AI proficiency without those fundamentals leaves you poorly equipped to spot a confident mistake.
For adjacent sectors, see will AI replace accountants? and will AI replace consultants?.
