Business analyst to AI: what transfers and what to learn

Why business analysts have a head start

I've spent 13+ years in business and IT, and I'm a business analyst at heart. The part of a project I always loved was the part before the code: figuring out what people actually need. AI made that the most important part of the job.

In April 2026, Google's CEO said 75% of all new code at Google is generated by AI. When AI writes most of the code, the bottleneck moves to knowing exactly what to build and proving it works. From my own projects, the work is shifting toward roughly 40% requirements, 20% coding and 40% testing. That's my estimate, not a published statistic, but it explains why analysts have a real opening.

What transfers directly

• Requirements and scoping. Turning "we need a dashboard" into who it's for, what goes in, what comes out and what done means is exactly how you get good results from AI.
• Acceptance criteria. Written well, they become the test cases you run the AI against.
• Process mapping. Most AI projects automate one step of a process you can already draw.
• Stakeholder management. Someone still has to agree on scope, explain trade-offs and get people to actually use the thing.
• Domain knowledge. Knowing how invoices, claims or orders really flow is what lets you spot when the AI is confidently wrong.

If you've done these for years, you already have the hardest part. (Here's what the job actually involves.)

The three gaps worth closing

1. Testing AI output. Analysts write acceptance criteria, but often hand testing to QA. With AI you need to test yourself: hidden examples, a failure rate, and a retest after every change. I call it the Break-It Rule.
2. Knowing the tools. One chat model, one automation tool like n8n or Make, and a way to build a quick app. Here's the toolkit I actually use.
3. Building something clickable. Analysts are used to handing a spec to developers. Now you can build the first working version yourself with AI, then hand it over. That shift is the whole career move.

Notice what's not on the list: learning to code from scratch. You'll read what the AI writes and ask it to explain things, but you don't need to write code by hand. (More on that here.)

A 30-day plan for analysts

• Week 1: Take a requirement you wrote recently. Turn its acceptance criteria into ten test examples with known right answers.
• Week 2: Build an AI step that does the task, using six examples. Test it on the four you held back, and write down the score.
• Week 3: Connect it to real work with an automation tool, and keep a person reviewing anything the AI isn't sure about.
• Week 4: Write a one-page case study with the problem, the failure rate before and after your fix, and the time saved.

Never use your employer's or a client's real data in a public project. Use public documents or realistic sample data instead.

How to describe the move

Keep "business analyst" in your history. It's an advantage, not something to hide. Add a line about AI with a result: Built [what] that [did what], [result with a number].

For example, the format looks like "Scoped and built an AI workflow that turns supplier invoices into structured data; 9 of 10 correct on unseen invoices, the rest flagged for review." Use your own real numbers.

Titles to search for include AI Solutions Engineer, AI Business Analyst, AI Automation Specialist and Applied AI Analyst. Read the description: if it says "requirements," "stakeholders," "automate" or "prototype," it's written for you.

Where to start this week

Coming from testing or project management instead? See QA to AI or project manager to AI. Not sure which gap is biggest for you? Take the 2-minute readiness quiz. It shows which of the four skill layers you're strongest in and gives you a three-step plan.

If you want the guided version of the 30-day plan, the AI Portfolio Sprint walks you through three projects in 21 days, the way AI work actually gets built inside companies.