There's a recurring debate in Business Analyst communities: will AI replace the profession?
It's a fair question. But I think it deserves to be asked differently.
AI can already speed up part of a BA's daily work. Writing specs, summarizing meetings, structuring requirements, preparing tests... Tasks that used to take several hours can now be done much faster.
But producing a document is not the same as understanding a problem.
And that's precisely where the stakes lie.
AI doesn't necessarily mean the end of the Business Analyst role. It challenges certain ways of doing the job, especially when the added value rests mainly on producing documents rather than on analysis, understanding needs, and the quality of decisions.
What AI automates, and that's a good thing
Let's be honest: part of a BA's time goes to repetitive tasks.
Writing a user story from a conversation that's already well scoped. Turning meeting notes into a structured document. Producing yet another deliverable from an existing template. Summarizing a fifty-email thread to pull out the key points.
These tasks are useful. But they don't always require you to carry out every step by hand.
Today, AI can help produce a first version of these deliverables in a few minutes. You give it the material, set the context, guide the writing, then check and refine the result.
The typing and formatting work can be cut down considerably.
The gain obviously depends on the complexity of the topic, the quality of the available information, and the level of verification needed. A simple summary doesn't take the same effort as a set of specs involving several systems and complex business rules.
But the potential is real.
And the point isn't only to produce faster. The real advantage appears when the time you save lets you put more energy into analysis, conversations with the teams, and exploring solutions.
AI can also become a great learning tool.
Want to understand how an API works? Decipher a data structure? Make sense of a code snippet or explore the possible causes of a technical error?
You can ask for an explanation adapted to your level, dig into the points that are still fuzzy, and go through several examples until you understand the topic better.
This doesn't replace hands-on experimentation, technical documentation, or a developer's expertise. But it can considerably lower the barriers to learning and let a BA tackle technical topics that would have been harder to explore alone before.
For anyone who wants to broaden their skills, it's an especially interesting opportunity.
What AI doesn't automatically understand
AI processes information. It can analyze text, spot apparent inconsistencies, suggest hypotheses, and generate convincing answers.
But having information doesn't necessarily mean understanding the context it belongs to.
Let's take an example.
An accounting manager explains that their team has been working around the existing IT system for several months. You could ask an AI to suggest a process improvement or to write the specs for a new feature.
It could produce a structured, perfectly plausible answer.
But before looking for a solution, several questions deserve to be asked.
Why is the team working around the system? Is it because it's too slow, because a feature is missing, because the official process no longer matches the reality on the ground, or because users were never properly supported?
The answer completely changes the nature of the problem and the solutions you can consider.
This is where the BA's work comes into its own. You need to meet the people involved, understand their constraints, observe real practices, and weigh several points of view against each other.
Sometimes you also have to deal with disagreements, conflicting priorities, and unspoken things that don't appear in any document.
AI can help prepare these conversations and analyze what comes out of them. It can even suggest relevant questions you hadn't thought of.
But it's up to the BA to put this information back in context, check the assumptions, and help make decisions with stakeholders.
So the value of the job doesn't lie simply in the ability to produce an answer. It also lies in the ability to tell whether you're asking the right question.
A concrete example: integrating an API
This is where the distinction becomes especially visible: technical topics.
Imagine you're working on integrating an external API. For a simple case, with fields that map directly between two systems, AI can help you understand the documentation, identify the useful fields, propose a mapping, and prepare a first version of the functional documentation.
That's already a nice gain.
But things get complicated when several calls have to be chained to get the expected result, when some operations depend on business rules, or when the behavior varies depending on the customer's context.
At that point, generating a mapping is no longer enough.
Why do these calls have to be made in this order? Which business rule justifies this transformation? What happens if a piece of information is missing? How should the system react when a call fails? What consequences could a change have on existing processes?
AI can suggest leads, analyze examples, and help explore the possible scenarios. But its suggestions have to be checked against the real documentation, how the system actually works, and the business constraints.
It can also get things wrong, assume a feature exists, or propose logic that is technically plausible but unsuited to the context.
So the BA needs to understand the flow well enough to challenge the suggestions, identify the areas of uncertainty, and work effectively with the developers.
The more complex the topic, the more important the quality of the context you provide and the verification of the results become.
The goal isn't to turn every BA into a developer. It's to give them the means to better understand the systems they analyze and to take advantage of AI without blindly delegating their judgment to it.
Does AI threaten certain types of BAs?
This is probably the least comfortable part of this transformation.
For years, some organizations defined the Business Analyst role mainly around gathering requirements, writing specs, and producing standardized deliverables.
The BA collects the requests, fills in the templates, writes the documents, and hands the result over to the technical teams.
When the value of a position rests essentially on these activities, automation really does change the game.
If a tool can produce a first version of specs in a few minutes, an organization can legitimately wonder how much time and how many resources should still go into that production.
That doesn't mean the profession is disappearing. It means some tasks take less time, and the expectations of the people who do them may change.
We should avoid a shortcut, though: producing documents doesn't mean doing bad BA work. Documentation remains essential on many projects. The difference lies in what comes with it.
A document can be the result of real analysis, complex workshops, and difficult trade-offs. It can also be an almost mechanical formalization of information that's already available.
AI won't have the same impact on these two situations.
For the BA who sees documentation as a tool in service of analysis, AI can become a great accelerator. It can cut the time spent on repetitive tasks and free up more effort for understanding the issues, the quality of solutions, and coordinating stakeholders.
But that means evolving your practices, learning to use the tools, and making the value of your work visible beyond the deliverables you produce.
So the question isn't simply whether AI can perform certain tasks. It's also how the profession will reorganize itself around these new possibilities.
The BA can also become a driver of AI integration
There's another opportunity, one that gets mentioned less often.
The Business Analyst knows the business processes, the difficulties the teams run into, and the friction points that slow activities down. That puts them in a good position to help identify the situations where AI could bring real value.
It can start with very concrete questions.
Which repetitive tasks eat up a lot of time? Which steps require tedious information searching? Where do teams have to manually rephrase, sort, or summarize information? Which activities could be assisted by AI without compromising the quality of the result?
Take a document validation process.
Today, a person may have to go through several documents, compare information, check that certain elements are present, and flag inconsistencies.
AI could help extract the useful information, compare the contents, and prepare a summary of the points to check.
The BA can then help rethink the process: determine which steps can be assisted, which checks must remain mandatory, who keeps responsibility for the decision, and how to handle errors.
You also need to assess the constraints around confidentiality, security, cost, and reliability. Not every task that can be automated in theory is worth automating in a given context.
The point isn't to add AI everywhere. It's to identify the use cases where it genuinely solves a problem.
This kind of thinking draws on exactly the BA's skills: understanding a process, identifying a need, evaluating solutions, and supporting the changes they imply.
Technical development, architecture, security, and governance remain responsibilities shared with the relevant specialists. The BA brings their knowledge of the business and helps bridge technical possibilities and real needs.
It's a natural extension of the role, as long as you stay focused on the value created rather than on how new the tool is.
How AI is changing day-to-day work in practice
The transformation isn't limited to writing specs. Several BA activities can already benefit from this kind of assistance.
Writing specs
Rather than starting from a blank page, you can give the AI a brief, workshop notes, and business rules to get a first structure.
You can then fill in what's missing, challenge ambiguous wording, and check that edge cases are properly covered.
The time savings are especially interesting when the format is recurring. But a well-presented spec isn't necessarily a correct one. Validation remains essential.
Synthesizing and analyzing information
Needs can be scattered across meeting minutes, emails, user feedback, and project documents.
AI can help bring this information together, surface recurring themes, identify apparent contradictions, and prepare a summary.
It's a way to speed up the preparatory work, as long as you go back to the sources when the meaning of a piece of information is uncertain or several interpretations are possible.
Preparing tests
From a functional spec, AI can suggest test cases, nominal scenarios, edge cases, and error situations.
It can also generate fictional datasets to make test preparation easier.
This lets you explore a set of scenarios more quickly and spot some situations that might have been overlooked.
But the generated cases aren't automatically complete or relevant. They may miss an important business rule, rest on a wrong assumption, or include scenarios that are impossible in the real system.
So the BA needs to check that the tests reflect the project's needs and risks. The team must also take into account the constraints of the test environment and data management.
Upfront investigation
This may be the most important point.
AI can help prepare a user interview, suggest clarifying questions, organize the results of a workshop, or explore several possible solutions.
But it shouldn't become an excuse to avoid getting out into the field.
Observing a real process, listening to users, understanding a team's constraints, and weighing the different perceptions of the problem against each other remain essential.
AI can make these activities more effective. It doesn't excuse you from doing them.
How to position yourself as a BA
If you're a Business Analyst today, I think the right question isn't only: "Is AI going to replace me?"
It's also: "Am I using AI to improve my work and spend more time on what really matters?"
A few concrete ideas can help you move forward.
Start with repetitive tasks. Pick an activity you do regularly, like summarizing a workshop or preparing test cases. Experiment with a tool that's approved in your organization, compare the result to your usual method, and check the real gain.
Learn by exploring. Use AI to better understand the APIs, data structures, technical concepts, and tools you come across in your projects. Check the answers and compare them against reliable sources.
Strengthen your analysis skills. The ability to clarify a need, identify fragile assumptions, and understand stakeholders' constraints becomes all the more important as producing content gets easier.
Make your value visible. Don't present your work only through the documents you've produced. Also show the problems you helped clarify, the risks you identified, the decisions you made easier, and the improvements you brought to processes.
Experiment without losing your critical thinking. A convincing answer isn't necessarily a correct one. Results must be checked, sensitive information protected, and tools used in line with your organization's rules.
It's not about chasing every new tool or automating everything overnight. It's about gradually developing a practice that combines AI's capabilities with judgment, experience, and an understanding of context.
What's next
AI is already transforming part of the Business Analyst's work. It speeds up certain tasks, makes it easier to explore technical topics, and opens up new possibilities for improving processes.
But it doesn't automatically make analysis better. It can also just help you produce an answer to the wrong problem faster.
So the difference will come as much from how you use these tools as from your ability to understand what they produce, assess their limits, and integrate them intelligently into your daily work.
That's exactly what I want to explore in the upcoming articles: tools, methods, experiments, and concrete lessons learned, to understand what really works.
Subscribe to the newsletter to regularly receive analyses, practical tools, and prompts you can use right away in your daily BA work.