3 Ways Business Analysts Can Scale Their Work with AI

If you work as a Business Analyst, you’re probably asking yourself “How can I find some practical ways to use AI for the work I do?” And it makes sense.

You don’t want to get left behind but you’re also cautious about using a tool that is, to some extent, still in its infancy.

In broad terms, you want to use AI (aka ChatGPT, Claude, or Deepseek) to boost your productivity, refine the documentation process, and deliver more value to customers.

In this article, I’ll share three practical scenarios where you can use Artificial Intelligence (AI) do this. From automating manual tasks to proactively identifying risks, these strategies will help you become a more efficient Business Analyst. These tactics are non-technical with no coding involved. It’s aimed at beginners and those who are relatively new to AI.

This is the first in a series of tutorials from the team at Klariti, dedicated to helping BAs like yourself with the latest AI tools and techniques.

Scenario #1: Automate use cases

I’ll be honest. I was initially hesitant to embrace Large Language Models (LLMs) in my role. AI felt a bit intimidating, and I wasn’t sure where to start. I also had a concern that it might replace the ‘human touch’ that is so important in my role.

However, the sheer volume of manual effort involved in writing use cases from stakeholder interviews, coupled with the desire to improve efficiency, nudged me to explore the possibilities.

Not sure about you but a significant portion of my day is spent manually transcribing, organizing, and synthesizing information to create initial use case documents. This is tedious, time-consuming, and prone to mistakes on my side (e.g. misinterpreting complex technical requirements).

While I use tools like to record meetings, even after I convert the recordings into usable drafts it still requires significant manual effort to get them into something I can really work with. Is there a better way?

How LLM can improve this process

To overcome these challenges, I decided to experiment with an LLM to automate the use case creation process. I started with ChatGPT but use Google Gemini right now. My Dev colleagues use Claude a lot as it’s good with coding.

Initially, I was cautious, but the more I worked with the LLM, the more my confidence grew, and I began to see its potential. Here’s how I got the LLM to improve my workflow:

1. Automated Transcription and Summarization:
• Process: I uploaded audio/video recordings of stakeholder interviews to the LLM. The LLM transcribed the interview, identified requirements, and summarized the main points discussed. Google NotebookLM is great for this.
• Improvement: This reduces the time I would have spent writing up the interviews, so I can now work on more strategic tasks. In the past, this process could take me several hours for a single interview, now it’s done in minutes.
2. Use Case Draft Generation:
• Process: Using the summarized interview content, I instructed ChatGPT to generate initial drafts of use cases. It identifies the actor, system interaction, pre-conditions, post-conditions, main flow, and alternate flows. I also provided the LLM with a template for the use case document to ensure consistency. That way it knows exactly what I’m expecting.
• Improvement: The LLM then generates a structured starting point for the use case, which saves me starting with a blank document. The template as input ensures consistent formatting across all use cases, leading to a more professional output.
3. Requirement Gap Identification:
• Process: Next, I feed the generated use case drafts back into the LLM and prompt it to identify potential gaps or inconsistencies in the requirements. For example, I might ask the LLM, “Are there any potential security vulnerabilities identified in these use cases?” or “Are there any conflicting requirements between Use Case A and Use Case B?”
• Improvement: This helps identify potential issues early in the process, preventing rework and ensuring a more complete set of requirements. Catching these issues early is invaluable. The LLM acts as a ‘second set of eyes’, so very little get missed.
4. Sentiment Analysis and Priority Ranking:
• Process: I also use the LLM to perform ‘sentiment analysis’ on the interview transcripts in order to understand the stakeholders’ emotional responses to different features. This helps prioritize features based on stakeholder enthusiasm and perceived value.
• Improvement: This insight into stakeholder priorities helps me prioritize the most impactful features first, which leads to greater satisfaction and quicker adoption. Better Net Promoter Score (NPS) too.

Value to Customers

Ultimately, my goal is to deliver maximum value to our stakeholders. AI allow me to focus on understanding their real needs and translates those into actionable requirements.

And the unexpected bonus: because the LLM takes some of the manual effort away, I find I’m better rested and have more energy to be creative, which helps me and the customers.

• Faster turnaround time: Quicker delivery of initial requirements documentation allows the Development team to start working sooner, reducing the ‘time to market’ (T2M) for new features.
• Higher quality requirements: More complete and well-structured requirements lead to fewer misunderstandings and reduce the risk of project delays and cost overruns, improving overall quality.
• Reduced project risk: Early identification of requirement gaps and inconsistencies leads to a more stable and predictable project outcome, mitigating potential risks.
• Better alignment with stakeholder needs: Sentiment analysis ensures that the project focuses on delivering the features that stakeholders value most, increasing adoption and satisfaction.

Performance Improvements

Like I said, I now use Google Gemini (and some ChatGPT) to offload repetitive and time-consuming tasks, so I can focus on higher-value activities.

And, as I continue to use LLMs, such as Google NotebookLM, I discover more ways to improve my performance. You begin to find different use cases at work. In terms of performance, it helps with:

• Time Savings: Less time spent on transcribing, summarizing, and use case drafting.
• Improved Accuracy: It transcribes and summarize information with higher accuracy than manual methods, which reduces errors.
• Increased Efficiency: This allows me to focus on more strategic activities, such as stakeholder engagement, solution design, and validation, improving my efficiency. Everyone gains.

Scenario #2: Automate initial test cases

Here’s another example of how to use AI for business analyst tasks.

After using the LLM to generate initial use case drafts, I realized that another significant bottleneck was the creation of test cases.

Manually creating test cases for each use case is time-consuming and requires a deep understanding of the system and potential failure points. This process is prone to overlooking edge cases and can be very repetitive. Before using an LLM, test cases were easily 25% of my workload, after documenting the use cases.

How LLM can improve this process

After using ChatGPT to draft the use cases, I felt more confident in exploring its capabilities for test case generation. This builds directly on the previous scenario and further optimizes the overall development process.

1. Automated Test Case Generation:
• Process: I fed the finalized use case document (generated with the LLM as described in the previous scenario) into the LLM. I instructed the LLM to generate a set of test cases, covering both positive and negative scenarios, boundary conditions, and potential error states. I provided specific templates or formats for the test cases, as well as guidelines on the level of detail required, and asked it to generate a large quantity of cases, covering many possibilities. At this point, I’m confident that I won’t miss out on critical tests.
• Improvement: This significantly reduces the time spent manually writing test cases. The LLM can quickly generate a wide range of test scenarios based on the use case, so we have more test coverage. Again, this frees me from the more mundane tasks.
2. Test Case Prioritization and Risk Assessment:
• Process: After generating the initial set of test cases, I use the LLM to prioritize them based on risk and impact. It can analyze the use case and identify the areas that are most critical to the system’s functionality and security. It then prioritizes the test cases that cover those areas.
• Improvement: This ensures that the most important test cases are executed first, allowing for faster identification and resolution of critical issues.
3. Test Data Generation:
• Process: Instruct the AI to generate realistic test data for each test case. This includes data for both positive and negative scenarios, as well as edge cases.
• Improvement: This eliminates the need to manually create test data, which can be error-prone process.
4. Traceability Matrix Creation:
• Process: Utilize the LLM to automatically create a traceability matrix that links each test case back to the corresponding use case requirement.
• Improvement: This ensures that all requirements are adequately tested and provides a clear audit trail for compliance purposes.

By automating the test case generation process, I can ensure that the software is thoroughly tested and meets the highest quality standards. This leads to a more reliable and user-friendly product for our customers. With more confidence, I can push the LLM to produce test data for many more scenarios, ensuring that we don’t make mistakes.

• Higher quality software: More comprehensive testing leads to fewer bugs and a more stable product.
• Reduced time to market: Faster test case generation allows for quicker testing cycles, reducing the time it takes to release new features.
• Improved user experience: A more reliable product leads to a better user experience and increased customer satisfaction.

This new application of the LLM further streamlines my workflow and allows me to deliver even greater value to the project. As I become more better at using the LLM, I am constantly finding new ways to optimize my processes.

• Significant time savings: Reduced time spent on manually creating test cases and test data.
• Improved test coverage: More comprehensive test case generation ensures that all aspects of the system are thoroughly tested.
• Enhanced traceability: Automated traceability matrix creation simplifies compliance and audit processes.
• Increased efficiency: Allows me to focus on more strategic testing activities, such as exploratory testing and defect analysis.

Scenario #3:  AI “What-If” Analysis.

While we gather requirements, create use cases, and generate test cases, one area that often gets less attention (and I admit, sometimes gets rushed due to time constraints) is proactive risk identification and mitigation planning.

We often rely on our past experiences and standard checklists, which can lead to overlooking novel risks specific to the project or emerging from the external environment. As a BA, I want to feel that the project has been assessed from every angle.

Learn more: How to Create Monte Carlo Simulations for Excel using DeepSeek

Building upon my growing confidence with LLMs, I’ve discovered a creative application to enhance risk management. Instead of simply reacting to known risks, I’m using the LLM to proactively explore potential “what-if” scenarios and develop mitigation strategies.

1. LLM-Powered “What-If” Scenario Generation:
• Process: Provide the LLM with project documentation (requirements, use cases, technical specifications, budget information, timelines), information on the company’s internal policies and procedures, industry trends, and even potential geopolitical or economic factors. Instruct the LLM to generate a list of potential risks that could impact the project, taking into account both internal and external factors. I’m then use the LLM to generate a wide range of “What if” scenarios and now see it as another ‘member’ of the team.
• Improvement: The LLM can identify risks that I might not have considered due to cognitive biases or limited experience. It can analyze vast amounts of data and identify patterns and connections that would be difficult for a human to spot. This is like having access to a highly knowledgeable and objective risk assessment expert.
2. Impact Assessment and Probability Estimation:
• Process: For each identified risk, ask the LLM to assess its potential impact on the project’s scope, timeline, budget, and quality. Also, ask it to estimate the probability of each risk occurring, based on available data and expert opinions.
• Improvement: This provides a more objective and data-driven assessment of the risks, allowing for better prioritization and resource allocation.
3. Mitigation Strategy Development:
• Process: Based on the impact assessment and probability estimation, instruct the AI to generate a range of potential mitigation strategies for each risk. These strategies can include preventive measures, contingency plans, and fallback options.
• Improvement: The LLM will then suggest mitigation strategies that I might not have considered. It can also analyze the potential costs and benefits of each strategy, allowing for informed decision-making.
4. Risk Monitoring and Reporting:
• Process: Use the LLM to continuously monitor the project environment for signs of emerging risks. The LLM can analyze news articles, social media feeds, industry reports, and other data sources to identify potential threats. It can also generate regular risk reports, highlighting the most critical risks and their mitigation strategies.
• Improvement: This ensures that risks are identified and addressed in a timely manner, minimizing their potential impact on the project.

By proactively identifying and mitigating risks, I can significantly increase the likelihood of project success and deliver greater value to our customers.

• Increased project success rate: Proactive risk management reduces the likelihood of project delays, cost overruns, and scope creep.
• Improved stakeholder confidence: Demonstrates a proactive approach to risk management builds trust and confidence with stakeholders.
• Greater return on investment: By mitigating potential risks, I can help ensure that the project delivers the expected benefits and return on investment.

What have I learnt? This creative application of the LLM changes my role from a reactive problem-solver to a more proactive risk manager. It demonstrates a commitment to innovation and a willingness to embrace new technologies to improve project outcomes.

• Reduced firefighting: Proactive risk management minimizes the need to react to unexpected problems.
• Improved decision-making: Data-driven risk assessments allow for more informed decision-making.
• Increased efficiency: Automating risk monitoring and reporting frees up time for other tasks.
• Enhanced reputation: Demonstrating a proactive approach to risk management enhances my reputation as a skilled and innovative business analyst.

Your Next Step

Here’s a suggestion. Select one of these scenarios – perhaps use case automation, as it offers a great entry point – and put it into practice. As you experiment and experience the results firsthand, your confidence will grow.

I’d highly recommend that you use this chance to use AI to hand off manual activities so you have more free time to do ‘deep work.’

This article is just the first in Klariti’s tutorials designed to help Business Analysts use AI. I’ve love to hear about your experiences and answer any questions you have. I’m over here on LinkedIn