The Business Analyst and AI: The Bridge Is the Target

The business analyst exists because business stakeholders and technical teams don't speak the same language. Someone has to sit in the middle: gather what the business actually needs, translate it into requirements an engineering team can build against, document the processes, write the user stories, and keep both sides aligned as the project changes. It's a translation layer, and translation layers are exactly what large language models were built to collapse.

Business analysts are rated Moderate risk, with a horizon toward 2030, one of the nearer-term horizons in this series. The exposure isn't a single tool. It's three separate pressures arriving at once, each removing a different piece of what the role has traditionally gatekept.

LLMs now draft requirements documents and user stories directly. Self-service analytics tools let stakeholders query data in plain English without an analyst in between. And AI-assisted no-code platforms let business users build working solutions without filing a requirements request at all. None of these replaces a senior analyst's judgment. Together, they shrink the volume of work that judgment used to be attached to.

What a Business Analyst Actually Does

The role varies by company, but the core is consistent: elicit requirements from stakeholders, often people who don't fully know what they want; analyze current business processes and identify where they break; document requirements, workflows, and acceptance criteria precisely enough that an engineering team can build to them; and manage the gap between what was asked for and what gets delivered as scope, constraints, and priorities shift. It is roughly equal parts documentation, analysis, and stakeholder management.

What LLMs Do Well

The documentation and drafting portion is the most exposed. Studies of LLM-generated requirements artifacts have found the models producing user stories, acceptance criteria, and first-draft requirements documents at a quality level that needs editing rather than rewriting. Atlassian's Rovo, generally available from late 2025, includes an agent that generates structured product requirements documents in Confluence from linked Jira work and other context, a task that used to be a meaningful share of a junior analyst's week. LLMs are also effective at summarizing stakeholder interviews and meeting transcripts into structured notes and open questions.

Where the Job Still Holds

The International Institute of Business Analysis' 2025 global survey found a majority of practitioners viewing AI's impact on the role positively, and identifying soft skills, communication, problem- solving, stakeholder management, as the growing differentiator. The parts of the job that don't automate are the ones that were never really about documentation: negotiating between stakeholders whose interests genuinely conflict, exercising judgment about which of fifty stated requirements actually matter, facilitating a workshop where the real problem surfaces only after an hour of the wrong conversation, and managing the organizational change that determines whether anything built actually gets used.

There's also a new function emerging from the same tools creating the pressure: as citizen developers build more of their own solutions, someone has to set standards, prevent a sprawl of ungoverned apps, and keep the requirements coherent across a business that can now build faster than it can plan. That's recognizably business analysis, aimed at a different target.

The Labor Market Reality

The BLS projection for management analysts is 10% growth through 2035 with a median wage around $101,860, which reads as healthy, but the category is broad and includes a lot of strategy consulting that isn't what a corporate business analyst does day to day. The more telling signal is the absence of one: unlike data analysts, customer support, or junior software roles, business analysts have not been the named target of a wave of AI-attributed layoffs. Whether that holds as the three pressures compound is the open question, and SHRM has cautioned that some AI-attributed cuts elsewhere are cost reductions wearing an AI label.

How to Use AI as a Business Analyst Now

For documentation: use LLMs to draft requirements, user stories, and interview summaries, then apply the judgment about completeness, ambiguity, and edge cases that the model doesn't have. The speed gain is real; the accountability for correctness stays with you.

For analytics: rather than being the person who runs every stakeholder data request, become the person who sets up self-service tools well and teaches the business to use them, then focus on the analysis that actually requires framing a question the tool can't.

For no-code: get ahead of citizen development rather than resisting it. The analyst who defines the guardrails and keeps the resulting sprawl coherent is doing higher-value work than the one filing requirements tickets.

What I Think

The 2030 horizon feels appropriately close. This isn't a job facing a speculative future capability; it's facing three shipped product categories that each remove a reason the role was necessary. The senior business analyst who is really a facilitator, negotiator, and judgment layer is well protected. The analyst whose actual day is writing up requirements, running standard reports, and shuttling questions between teams is doing work that is being automated in front of them.

What I'd watch is whether organizations recognize the distinction. The risk isn't that AI does the whole job. It's that a company sees the documentation get faster, concludes it needs fewer analysts, and cuts without noticing which analysts were doing the part that mattered. That has been the pattern in adjacent fields, and there's no reason to expect this one to be different.

"The job was to stand between the business and the build. AI didn't take the job. It removed three of the reasons anyone had to stand there."