The Business Analyst’s AI Cheat Sheet

You have Claude, Gemini, Co-Pilot, ChatGPT, Grok—but do you know where and how to use them? I’ll be honest, when ChatGPT rolled out to the public, I was a little skeptical of its capabilities. I neither believed it could make a dent in the work I do, nor thought it would be capable of surpassing the skillset I have acquired over my career.

Things changed when the industry started pushing more for AI. I started dabbling a lot more in AI and slowly came to the realization that it is a great companion. If you use it well, it can be your smartest executive assistant who knows a bit of everything.

Imagine an assistant who can create meeting invites, build screen mockups, talk to you about complicated SOQL queries, optimize your batch logic, create a personal profile (like this whole website), or layout a one-year plan for your career advancement. That is AI.

However, I often see practitioners stuck with this question: How do I use AI?

How to Integrate AI into the Business Analyst Workflow

There is no universal answer on how to use AI. Your workflow is unique to you. Your style of work is unique to you. The goal here is to tailor your AI usage to your style of work. Start experimenting.

What has worked for me?

I started injecting AI into my workflow gradually, increasing complexity over time. I’m still not an AI Superuser who has five different agents working under me or anything. Currently, I leverage AI—and when I say AI, I’m referring to publicly available or enterprise AI LLMs.

Here are some of the areas where I offload work to AI:

P.S. This is by no means an exhaustive list

AI Use Area	Description
Data Cleanup	Quickly making a CSV out of unstructured data instead of using Excel.
Data Cleanup	De-duplicating entries within a CSV.
Data Analysis	Running a quick sanity check on data integrity.
Data Mapping	Mapping source to target data by simply providing the two data dictionaries.
Data Creation	Quickly creating dummy data that follows a pattern if needed.
UX Design	Asking for best practices on screen design to be educated before handing it over to the UX team.
UX Design	Generating screen microcopy.
UX Design	Extracting screen elements and label names from existing screens.
UX Design	Creating mockups/wireframes from descriptions to present to clients or the dev team for initial feedback.
Requirements Gathering	Asking AI for talk tracks on simplifying complex system jargon for non-technical stakeholders.
Requirements Gathering	Reviewing requirement synthesis with AI to check for edge cases or logical gaps.
System Design	Reviewing system architecture best practices (e.g., Salesforce-specific platform questions).
System Design	Evaluating batch logic efficiency.
System Design	Asking overall feasibility questions based on platform limitations.
System Design	Creating user stories from meeting notes and raw descriptions.
Project Management	Scheduling calls or drafting meeting agendas from just a brief description.
Project Management	Recapping email inboxes and querying which specific emails require immediate responses.
Project Management	Reviewing outgoing emails for logical fallacies, tone adjustments, or factual inaccuracies.
Personal	Navigating career trajectories and finding the best certification options based on a specific practice or platform.

Where do I start?

There is no single right answer for this. Use it wherever you see fit. Start using AI instead of basic spellcheck for your emails. Start asking foundational questions about the platform you work on. Start using AI to summarize meetings.

While I whole-heartedly recommend watching a quick video on Prompt Engineering (you will find plenty of them on YouTube—pick any one), don’t get intimidated by the jargon. Here is an excellent example of a Prompt Engineering video from Youtube. The terminology is a little over the top, but it gives you a solid idea of how to prompt efficiently. As you adopt more AI, your prompt writing will naturally become sharper.

Once you get the hang of basic prompting, it is crucial to understand the power of context-rich prompting.

What is context-rich prompting?

This is easiest to explain with a direct contrast:

Lazy Prompt	Context-Rich Prompt
Write a user story for a feature to prompt a user to view bank disclosures prior to selecting a prepaid card payment.	Assume yourself as a Senior Business Analyst on a public benefits portal that uses an external banking vendor. The business needs users to view the banking terms and disclosures prior to selecting their prepaid card solution as a payment method. Write user stories considering a modal popup that will track whether the user clicks the link before letting them continue.

Essentially context rich prompting is giving the LLM enough information to work with so that it can generate an answer that is more detailed and tailored for your use case.

Where do I stop?

I intentionally added this header because when it comes to using AI, you absolutely must know when to STOP. Before you paste a single line of project work into an AI tool, you have to understand the boundaries of enterprise consulting.

Never paste client intellectual property or PII: Unless your client has explicitly approved a dedicated, secure enterprise LLM instance where project information is protected, never share IP into a public LLM. Keep data entirely anonymized.
The “Trust but Verify” Rule: Anything that comes out of AI must be verified. AI is prone to hallucinations—just like us when we haven’t had our morning cup of coffee.

Treat AI output like work delivered by a brilliant intern: it looks great on the surface, but it requires your senior oversight to review, edit, and validate it before it ever touches production.

You should use AI as an accelerator, not a replacement for your own intellect. AI is still in its infancy, and you should use it in a way that keeps you in complete control of the final output. You review the results. You edit the results. Do not under any circumstances allow a client to look at your deliverables and think of them as “AI slop.”

Be authentic. Be smart enough to show that you thoroughly understand the details of what you are talking about.