Gigasoft Tool Aims to Eliminate AI Charting Code Errors
Gigasoft's ProEssentials v10 charting library has introduced a tool called pe_query.py designed to solve errors in AI-generated code. The tool validates generated code against the compiled DLL binary, which the company claims eliminates the common problem of AI assistants hallucinating incorrect property names.
- The `pe_query.py` tool's validation method is distinct from common approaches like Retrieval-Augmented Generation (RAG), which ground AI responses in documentation. Instead, it deterministically verifies code by treating the compiled .NET DLL assembly as the definitive source of truth, programmatically inspecting its members, types, and properties at runtime. - This validation is likely accomplished using a library such as Python.NET, which enables Python code to interact with the .NET Common Language Runtime (CLR). This allows the script to use a mechanism called .NET Reflection to dynamically load and query the ProEssentials assembly, accessing its metadata to confirm the existence and correct naming of API properties. - The problem of AI hallucination is prevalent in code generation; a 2025 study found that 5.2% of package names recommended by commercial AI models were non-existent. For a developer, this could manifest as an AI suggesting a property like `chart.PlottingMethod.CandlestickOHLC` when the actual property in the library is `chart.PlottingMethod.OpenHighLowClose`, leading to compilation errors and time spent debugging. - ProEssentials has a long history in specialized, high-performance charting, having been founded in 1993 and releasing its first C++ charting libraries in 1995. The library has since evolved to support multiple interfaces including WinForms, WPF, ActiveX, and VCL, indicating a focus on mission-critical and embedded applications often found in engineering and finance. - The library is specifically marketed for financial applications, offering chart types such as High-Low-Close, Candlestick, and multiple, overlapping axes for complex data visualization. The v10 release also emphasizes performance by offloading rendering tasks to the GPU via Direct3D, a critical feature for handling the large, real-time datasets common in quantitative analysis. - For freelance developers and solo founders, the tool aims to reduce the friction of adopting a new, extensive library by lowering the time cost of experimentation with AI-generated code. The pricing model is a one-time, per-developer perpetual license which includes ongoing support and royalty-free distribution, a factor for those building a commercial product.
