FinTech Firm Exploring AI Opportunities Offered by GitHub Copilot 


An international FinTech firm, headquartered in South Carolina, with offices across the US and over 50 employees around the world.

The company’s flagship product is an asset and portfolio management platform that centralizes operational and financial data, allowing its customers to better visualize and analyze performance.


The firm had a legacy codebase (an active software system built with outdated technology) that needed updating/refactoring. 

A Keeper Solutions team (made up of a solutions architect, an AI tech lead, a frontend developer and a backend developer) was tasked with exploring all the ways in which GitHub Copilot could be used to enhance refactoring efficiency vs using traditional methodologies alone. 

GitHub Copilot is an AI-powered code completion tool developed by GitHub and OpenAI that suggests code completions as developers type and turns natural language prompts into coding suggestions.


During this exploratory project, which took the form of a 4-week sprint, Keeper Solutions’ team wanted to assess if GitHub Copilot could:

                          1. Improve refactoring efficiency, quantifying the time savings achieved by using GitHub Copilot during the refactoring process. This included elements such as code simplification, reducing technical debt, and optimizing performance.
                          2. Enhance code quality, assessing if Copilot improved code quality, readability and maintainability post-refactoring.
                          3. Enhance developer experience, evaluating user feedback on the platform’s usability and effectiveness from the developers, who were given a basic introduction into GitHub Copilot the week prior.
Results, Findings and Recommendations

Below are a number of findings that our team identified after a 4-week exploratory project:

                        1. The more a developer uses GitHub Copilot, the more effective the system becomes as it learns patterns and offers better code snippets for the task at hand. 
                        2. On the flip side, the more a developer uses the platform, the better they become at providing prompts to the AI technology. This, again, increased speed. 
                        3. Two thirds of developer’s activities were positively impacted by using Copilot, resulting in a 45% time saving on average.
                        4. Simple unit testing was seen as one of the most helpful and effective Copilot features. Copilot suggests multiple tests in a matter of seconds. The generation of these tests offers significant time savings.
                        5. The process of writing JSDocs was simplified hugely through using Copilot. Again, this was a huge timesaver for frontend developers.  
                        6. The time it took to refactor legacy codebase was reduced significantly.
                        7. Copilot’s ability to analyze existing tests, and suggest areas of vulnerability saved backend developers several hours for each test file.
                        8. Instant generation of docstrings (which allows programmers to understand what a system does without having to read the entire piece of code) was noted as one of the top time-savers. 
                        9. If the company were to continue using GitHub Copilot long term, our team highlighted the importance of establishing AI governance standards and defining AI guardrails to ensure safe use and to mitigate potential risks associated with AI-generated code. 
                        10. Our team also recommended building a prompt library, which would increase Copilot effectiveness and efficiency even further.
Project Impact

Prior to examining how Copilot could be used to enhance refactoring efficiency, the company’s team of 35 full-time developers was having difficulties with product delivery, with one crucial module at least 9 months behind schedule. 

After the 4-week sprint, Keeper Solutions was able to suggest a solution that involved the use of GitHub Copilot to increase efficiency. This solution provided the client with an opportunity to reduce engineering capacity from 35 to just under 20 developers while improving code quality and increasing development velocity. 

In this way, incorporating AI into the software development process saved the firm money, helped to streamline the business and increased overall performance. 

Read Our Case Studies

Supporting Atlanta-based Startup, Artis Technologies, from Idea to FinTech Success
Allowing FinTech Start-up, Rivet Labs, to Effortlessly Scale Up Software Resources
Working with Umba: A Digital Bank for the Emerging Markets