Duklas' customer (NDA) develops a financial analytics product delivered as an Excel add-in. It's used by banks, insurance companies, and investment houses, where calculation accuracy and consistency are business-critical.

One enterprise client needed to run the same calculations directly inside their internal systems, without relying on Excel.

The business goals were to:

✅ Onboard an enterprise client
✅ Deliver a fast, accurate, low-risk solution without rewriting years of financial logic
✅ Enable a clear decision: proceed to further development or stop

The solution was to build an internal API that replicated Excel-based calculation logic and make it available for integration with company's systems.

Duklas proposed a prototype-first approach to validate API integration feasibility before full-scale development.

AI was used as a development accelerator, while responsibility for logic and quality control remained with a senior developer.

Duklas' senior developer wrote prompts, reviewed the code and test cases before deployment, ensured the logic was correct, the code was clean, and that AI did not hallucinate.

Key steps:

1️⃣ Create reference implementation
Market and portfolio data were fed to the Excel add-in. The output calculations were used as “source-of-truth”.

2️⃣ Generate API code with AI
The same market and portfolio data along with the Excel outputs were fed to AI. It generated the initial API code, replicating the add-in’s logic.

3️⃣ Generate tests with AI
AI created test cases using the Excel calculations as the expected results.

4️⃣ Run feedback loop
Compared outputs, adjusted, repeated. Accuracy was the main concern, so the process repeated until the results matched Excel exactly.

The prototype met the primary objective — to deliver a fast, accurate, low-risk solution ready for enterprise integration.

✅ A working prototype was built within hours instead of weeks of development.

✅ The prototype delivered correct results, ready for customer rollout.

✅ The entire prototype was created with AI guided by a senior developer without sacrificing quality.

This allowed the customer to validate API integration feasibility immediately. In the following weeks, Duklas finalized the solution and successfully deployed it into the customer’s environment.

As a result, GIAnalyzer expanded its offering, introducing custom API development as an additional service for enterprise customers.