Streamlining Software Development: AI Tool Saves Time on Requirements Analysis and Documentation

AI tool development

Our AI-driven tool helped a typical software development team achieve exceptional results in approaching requirements analysis and project documentation within just three weeks. It saved our client a significant amount of time and money.


Our client is a medium-sized software development company in the technology field working with businesses from various sectors. They needed to upgrade their requirements documentation process to significantly enhance the efficiency and accuracy of requirement analysis and minimize gaps at the earliest stage of the investigation. 


The client searched for a solution to deal with inefficiency and over-reliance on manual labor while creating project documentation, such as compiling user stories and acceptance criteria in a specific format (Gherkin). They found that the traditional methods were time-consuming and slowed down their projects during the most critical moment: requirements analysis. The client was also concerned about convenience when using the future tool: it had to be integrated with the analysts’ everyday environments, such as JIRA, Google Docs, or Slack.

Our Approach

Our team, consisting of a project manager, developer, and prompt engineer, used an agile approach to address challenges. 

We divided the work into phases, focusing first on prompt development and then on platform integration with Azure and Slack. Afterward, we focused on extensive testing and optimization.


  • We developed an AI-driven prompt tool that decomposes requirements into user stories with acceptance criteria in Gherkin format.
    • We realized that sometimes there is not enough available information, so we added functionality where the tool asks additional questions. The questions are then categorized and sent to a customer when more information is needed. This feature is useful for rough estimation and initial analysis of our client’s customers’ requests.
  • Using the FewShotPrompt template and OpenAI API, our tool composes requirements with the story following the Gherkin template, Gherkin-style Acceptance Criteria, and a list of additional questions to refine with stakeholders. The tool then sends it as a reply via Slack bot.
  • The application is built upon the Flask framework, hosted on Azure, and is integrated with Slack over an API. 
  • We tested and optimized the prompt to get the most reliable results, optimizing performance as well as costs. We minimized the amount of tokens used for the prompt, and after optimization, it decreased by about 7 times.

The Outcome

  • Time Savings:
    Implementing our AI Decomposition (AiD) Tool significantly reduces the time required to break down requirements into user stories and acceptance criteria.
  • Cost Savings:
    By automating routine tasks, the client experienced a 12% decrease in operational costs associated with software development, contributing to improved profitability.
  • Enhanced Reputation: The tool facilitated internal skill development in AI-related environments. With streamlined requirements documentation, the client enhanced their reputation as AI specialists, attracting more business opportunities in the competitive technology market.

Learn more about AI PoC (Proof of Concept) Development with First Line Software

Want to know more details about this case study?

Get in touch

Related work

Interested in talking?

Whether you have a problem that needs solving or a great idea you’d like to explore, our team is always on hand to help you.