AI-driven development lifecycle transformation for a healthcare technology company

Development teams

Development teams were dealing with a high volume of manual work across code generation, automated test creation, and code review. In addition, developers also occasionally handled incoming end-user issues that were often caused by end client-side misconfigurations rather than code defects, requiring investigation and responses without any actual code changes. These challenges slowed feature delivery across platforms, reduced developer productivity, extended review cycles, and increased overall development costs.

To address these issues, Itransition introduced a series of targeted AI-driven initiatives:

• AI-assisted automated test creation
  The AI agent analyzes the existing project codebase and automatically generates test cases, increasing testing coverage and accelerating test creation. This helps identify and fix issues earlier, while reducing manual effort
• AI-assisted code review
  The AI assistant enhances the code review process by providing structured context and highlighting complex areas of the code. It generates visual diagrams and summaries of changes, helping reviewers quickly understand functionality and dependencies. It also suggests improvements, enabling faster and more informed reviews with less manual analysis
• AI-assisted development
  An integrated AI assistant ecosystem that supports developers across the full development workflow, including coding, debugging, testing, research (e.g. identifying optimal approaches, libraries, and implementation patterns), and documentation (e.g. generating descriptions of implemented logic and changes)

In addition, Itransition addressed a key capacity issue outside core development:

• AI intake reviewer
  The AI agent assists in reviewing incoming user issues, helping determine whether a problem is caused by solution misconfiguration or requires code changes. By resolving or redirecting non-development issues early, it minimizes interruptions and allows developers to focus on core engineering tasks

Business analysis teams

BA teams faced three key challenges:

• Time-consuming requirements preparation and risk of gaps
  Business analysts had to gather information from multiple sources – meetings, Jira tasks, legacy documentation, etc. This process was time-intensive and error-prone, creating a risk of overlooked dependencies, edge cases, or system constraints when formalizing requirements, especially given the complexity of projects, legacy systems, and fragmented information sources. If such gaps were to be discovered later in the project during refinement or testing, addressing them would become significantly more costly
• Lack of formal requirements in fast-track initiatives
  In some cases, development teams created demo solutions or PoCs based on high-level business concepts without formal BA involvement. When these solutions needed to be moved into production, BAs had to retrospectively document requirements based on already implemented functionality, often without full visibility into system logic and edge cases
• Manual effort in release documentation
  Product owners and business analysts spent significant time manually preparing release notes and compiling release documentation

To address these challenges, we embedded AI agents into BA workflows:

• Requirements review
  The AI agent reviews requirements before they reach the team, validating them against internal standards, checklists, and the knowledge base. This helps identify gaps, inconsistencies, and missing dependencies early
• Requirements generation
  The AI assistant generates structured user stories based on a combination of inputs, including guidance from BAs, meeting transcripts, emails, and supporting documents. Instead of manually formalizing requirements, BAs provide high-level input and contextual materials, while the agent consolidates these inputs into complete, well-structured requirements
• Release notes generation
  The AI agent automatically generates release notes and completes release documentation based on user stories and BA inputs

Quality assurance teams

QA teams faced several efficiency and scalability challenges:

• Time-consuming analysis of automated test results
  Teams managed projects with over 2,500 automated tests each, requiring regular analysis. Each run generated large volumes of test result data including failures that required analysis. This output required manual review, categorization, and triage by QA engineers, making the process time-consuming
• Limited scalability of test automation
  Test creation largely relied on manual scripting, typically requiring experienced QA automation engineers with strong technical and programming skills. The limited availability of such experts constrained the team’s ability to expand test coverage
• Inefficient test case creation for manual QA
  Manual QA engineers spent significant time compiling test cases from existing libraries of reusable test actions (e.g. opening a page, clicking a button, verifying a result). While many of these actions were already available, finding and combining them into complete test scenarios required substantial effort, reducing overall productivity

We addressed these key bottlenecks by introducing the following AI-driven solutions:

• AI-improved test automation reporting
  Enhanced Allure-based reporting, automatically transferring test run data to an AI agent for analysis. The agent processes mixed errors, groups them by root cause and relevance, and generates a structured report for QA engineers. This eliminates the need for manual triage of raw results, significantly reducing analysis effort and accelerating defect investigation
• AI-driven test automation
  The AI agent enables auto test creation without requiring deep coding expertise, allowing a broader range of QA engineers - not just automation specialists - to build and maintain automated tests. The agent also adapts to changes in UI and application logic, automatically updating tests as the system evolves
• AI-based test compiler
  Converts user stories and feature descriptions into ready-to-run automated tests, generating structured test scripts to expand overall test coverage. As the solution can also detect and fix failing scenarios, it proved particularly valuable in projects with limited or no QA capacity, enabling developers to maintain test coverage independently and reducing reliance on QA for routine updates, while supporting QA teams in accelerating test creation.