Discover how Arrk Group combined AI tools with agile delivery to develop a customer’s web application in around 75% less time than originally estimated.
Customer
The customer operates in the Business Intelligence sector, providing businesses with a comprehensive dataset that exceeds many competitor offerings and enables users to focus on critical insights through an AI-powered interface.
The Challenge
The customer wanted to explore an enhanced data discovery experience and test the market with a baseline web application to support their go-to-market strategy. Traditional development approaches risked increasing costs, extending delivery times and delaying launch. They required an agile approach that would streamline delivery and produce high-quality outcomes quickly.
Solution
Arrk Group delivered the baseline web application in weeks using React and Python, supported by AI tools such as ChatGPT and Sourcegraph Cody.
Cody accelerated React development by generating boilerplate code and improving development workflows. AI-assisted tools also sped up debugging through error analysis and helped refine UI and UX components. This approach enabled rapid iteration, efficient integration, and the delivery of a high-quality solution within a compressed timescale.
We combined agile practices with AI-enabled ways of working to respond quickly to questions and change requests from the customer team. The effective use of AI tools helped us deliver with a lean team while reducing coordination overhead. As a result, the customer achieved their go-to-market objectives efficiently and with minimal investment.
Key Outcomes
- Delivered a baseline product in three weeks, significantly faster than originally estimated.
- Established a repeatable AI-enabled agile delivery approach now used on other customer projects.
Conclusion
By streamlining development and using AI-enabled tools to support delivery, Arrk helped the customer bring a new product to market in a significantly shorter timeframe, requiring around 75% less effort than originally estimated.



