Analyst Workflows With Claude: What Changes for Junior Bankers

None of this changes who signs off on a number that goes in front of a client or a credit committee. If a comp set has a stale multiple or a model assumption doesn't tie to the data room, the analyst who let it through is the one who answers for it, not the tool that drafted the first pass.

That's the real shift in investment banking analyst AI use: the work that used to prove you could be trusted with a live deal, hours spent formatting, keying in data, building the first version of a model, is partly automated now. What's left is the part that was always harder to teach: knowing which number looks wrong before anyone else catches it.

Banks differ a lot in how formally they've adopted AI tools. Some desks have a standardized workflow and clear rules about what needs a second check. Others leave it entirely to individual analysts, which means the same mistake, trusting an AI draft without verifying it against source documents, shows up at different banks for different reasons.

Entry-level analyst AI adoption is rewarding a different mix of skills than the one that defined the first few years of the job a decade ago. Raw modeling speed matters less when a tool can build the scaffolding; the ability to spot when that scaffolding is wrong matters more.

Judgment about which question to ask is becoming a bigger differentiator than the ability to answer a known question quickly. An analyst who can tell a managing director which three numbers in a model are the ones worth double-checking before a client call is doing work that AI for junior bankers hasn't automated.

Communication and context still sit entirely with the human. Knowing what a specific client cares about, what a specific sector's investors are nervous about this quarter, and how to phrase a disagreement with a senior banker tactfully: none of that comes from a transcript summary.

Modeling intuition, a feel for whether a number is in the right range before you've even checked the formula, is still built the slow way, by doing the work by hand enough times to develop a sense for it. AI tools that draft the first version of a model don't replace that; if anything, analysts who skip the hand-built reps too early arrive at the review stage with less intuition to apply.

For someone weighing whether to start a career in investment banking with AI already part of the toolkit, the honest answer is that the job is changing shape rather than disappearing. The hours are still long, the learning curve is still steep, and the path still runs through the same apprenticeship model it always has, just with different early tasks.

What's changing is what the first two years teach. A junior banker who learns to direct AI tools well, to know what to ask for, how to check it, and when not to trust it, is building a skill that transfers up the career ladder, since senior bankers spend their time reviewing other people's work too. The analysts most at risk aren't the ones using AI tools; they're the ones treating AI output as finished work instead of a first draft.

If you're choosing a bank or a group, it's reasonable to ask how they expect analysts to use AI tools and what their review standards look like. A desk with a clear answer to that question is more likely to be investing in training analysts for the job as it exists now, rather than for the job as it existed five years ago.