How can a business analyst move into AI operations?

Most directly through building rather than reviewing. A business analyst in retail banking already maps processes, writes requirements and defines what done looks like. Those skills write an AI agent's brief, choose the real cases it is certified against and set its limits. Learn how governed agents earn depth, then aim for builder or agent owner roles.

Which business analysis skills carry straight across?

• Process mapping. Before an agent can take over account servicing or complaint triage, someone has to map every path a case can take, including the awkward ones.
• Acceptance criteria. Certifying an agent means replaying it against real past cases and scoring each answer against a floor. Writing those cases and their correct outcomes is acceptance testing.
• Stakeholder translation. Compliance, operations and product all want different things from an agent. You already broker that.
• Scope discipline. An agent's role scope is one of the five bounds every action must meet. Deciding what is in and out of scope is familiar ground.

What changes compared with traditional business analysis?

Three things.

First, the specification is live. An agent's brief, sources and limits change through its life, and each change can require re-certification. You are not handing over a document and moving on.

Second, evidence replaces sign-off. In a governed system, an agent's depth is earned on the bank's own cases, not configured. The role sets a ceiling; the test evidence sets the depth; the lower wins. Your test cases decide how far the agent may go.

Third, certificates expire. An agent's certificate lapses after 90 days, and a failed run drops it to read-only the same day. Someone has to plan re-certification. That someone is often the analyst who built the test set.

What does the agent do, and what does the analyst do?

What does this look like for one person?

Take a business analyst we will call Nadia at a retail bank moving card dispute intake to an agent. She maps the dispute types: fraud, goods not received, duplicate charge, subscription not cancelled. She gathers three hundred closed disputes covering each type and the awkward edge cases, like a merchant refund that crossed with the dispute. Those become the certification set.

The agent passes most gates but fails on escalation for disputes involving a vulnerable customer. Nadia rewrites the brief so those are always held, and the agent is certified at depth 2: it drafts, a person decides. When the bank later wants it at depth 3 for low-value disputes, Nadia designs the evidence needed. She is now the agent's owner in all but title.

Which roles should you target?

Titles vary. Look for agent builder, automation analyst, AI product analyst, operations design lead or AI operations analyst. Duties to look for: designing agent workflows, defining test cases, certification, approval design and change control. Banks under model risk supervision often want people who can document decisions clearly, which business analysts do.

Some analysts start with a short spell in review to learn how agents fail in practice. The route through review is in how to move from business analyst to AI agent reviewer, and the analyst path in how to become an AI operations analyst as a former business analyst. Read what an AI operations analyst does to compare. Neighbouring routes for compliance analysts and fraud analysts are worth comparing. The full map is on AI agent jobs and careers.

For evidence employers can check, the Builder track at Pryme Intelligence Academy ends in a capstone agent certified in Attest to depth 2 or higher, which is close to the work described here.