Gigasoft Solves AI's Biggest Charting Code Problem: Hallucinated Property Names
illi News/10326944

DALLAS - illiNews -- When developers ask AI assistants to write charting code, something predictable happens. The AI generates property names that do not exist. If the developer uses that code, it will not compile — and they are left searching through documentation to find the correct property path. For charting libraries with 1,000+ properties, this cycle of hallucinated names, compile errors, and manual correction is the single biggest time sink in AI-assisted development.

Gigasoft, Inc. has addressed this directly in ProEssentials v10 with pe_query.py, a Python-based AI assistance tool that works with Claude, ChatGPT, GitHub Copilot, Gemini, and local AI models. Unlike documentation-search approaches that reduce hallucination probabilistically, pe_query.py eliminates it deterministically by validating every property path against ground truth extracted from the compiled DLL binary.

The system gives any AI assistant on-demand access to 1,104 properties, 80 methods, 40 events, and 167 enumerations extracted directly from the ProEssentials assembly. Before the AI delivers code to the developer, it runs a validate command that checks each .NET property path against this authoritative source. Invalid paths receive correction suggestions pointing to the actual API element. The result: the developer receives code with every property path verified before they see it.

More on illi News

• Carlsbad Hotel Named Best of La Quinta Award Winner
• Scoop Social Co. Launches a New Era of Mobile Hospitality — One Truck, Two Experiences
• Record Sales Growth After Strategic Acquisitions; New Distribution Agreements for Established Premium Cigar Supplier: Green Leaf Innovations $GRLF
• R2 Copilot Addresses Critical Privacy Issues as Enterprise AI Spending and Security Incidents Rise
• Innovative Environmental Technologies Unveils New Website Featuring Free AI Tools for the Environmental Industry

"Every other charting vendor tells developers to review AI-generated code for accuracy," said Robert Dede, founder and lead engineer at Gigasoft. "We decided to solve the problem instead of warning about it. The validate command makes hallucinated property names structurally impossible."

But validation is only part of the system. The pe_query.py tool provides AI assistants with on-demand access to the complete ProEssentials API with ground truth validation, 32 knowledge files covering architecture, patterns, and best practices, 116 working code examples in C# and C++, a 4 MB unified documentation JSON containing rich descriptions for every property, and an 800-synonym feature index that maps natural language queries to the API, examples, and documentation content. The pe_query.py tool ties all of these resources together — enabling the AI not only to write validated code, but to intelligently extrapolate complete solutions for scenarios that do not exist in any example.

"A customer emailed us with a Gantt charting question," said Dede. "I loaded their requirements into a Claude Opus 4.6 Extended conversation with our knowledge files. Claude built a full Gantt implementation using our GraphAnnotation and custom Y-axis features — code that does not exist in any of our 116 examples. It mostly worked on the first attempt. That is not autocomplete. That is an AI that understands the architecture well enough to invent new solutions."

Gigasoft recommends Claude Opus 4.6 Extended with the Projects feature for the best results. With ProEssentials knowledge files loaded, Claude can answer technical support questions and write validated charting code with accuracy approaching that of Gigasoft's own engineering support team.

More on illi News

• CCHR Warns: Psychiatric Diagnoses Without Biological Proof Now Used to Justify Euthanasia
• Impact Filtration Appoints Alejandro Sturniolo as Head of Sustainability to Engineer High-Performance, Water-Positive Infrastructure
• Hedge Funds Benchmark Against Tickeron AI After Platform Posts 129% Annualized Return Amid Tariff Volatility
• NAWLA Members Approve Bylaws Amendments With Strong Support
• Naperville: Annual Arbor Day Tree Sale Begins Online at 8 a.m. March 23

The pe_query.py tool and its knowledge files are local assets included with the installation — no proprietary cloud service required. The system works with any AI assistant, and for air-gapped defense environments, classified programs, and organizations with strict data governance requirements, the entire workflow can run against local models with no external connectivity. An interactive API Explorer at https://gigasoft.com/documentation lets developers see the same query capabilities that power the AI assistance system.

The AI assistance system ships as part of ProEssentials v10.0.0.20, which also introduced expanded GPU compute shader rendering. ProEssentials constructs chart images entirely on the GPU using Direct3D compute shaders, rendering only when data changes. The engine handles 100 million data points in approximately 15 milliseconds using a zero-copy method that reads the developer's existing array without duplication.

ProEssentials covers WPF, WinForms, C++ MFC, Delphi VCL, and ActiveX from a single native engine. A full technical comparison evaluating ProEssentials against SciChart, LightningChart, Syncfusion, and DevExpress across GPU architecture, performance, AI assistance, pricing, and support is published at https://gigasoft.com/why-proessentials

The pe_query.py tool and AI assistance workflow are documented at https://gigasoft.com/ai-code-assistant

About Gigasoft, Inc.

Gigasoft, Inc., based in the Dallas-Fort Worth area, develops ProEssentials, a GPU-accelerated charting component library for WPF, WinForms, C++ MFC, Delphi VCL, and ActiveX. The library is used in medical, industrial, scientific, defense, and financial applications worldwide. Technical support is provided directly by the engineering team. Founded 1993.

Contact: Robert Dede, Gigasoft, Inc. | gigasoft.com