AI-Driven Data Automation for Products Taxonomical Categorization

ZAGREB, Croatia - Jan. 3, 2026 - PRLog -- Matasoft, a developer of innovative productivity software, today showcased a significant advancement in practical business artificial intelligence. A new public case study details how their flagship product, (Un)Perplexed Spready, successfully automated a complex data taxonomy migration project using integrated AI models, turning weeks of potential manual work into a seamless, automated process.

The case study documents a real-world task: accurately re-categorizing a catalog of 2,751 furniture products from one hierarchical taxonomy system to another. Traditionally, this work requires extensive manual review, lookups, and research, consuming valuable analyst time and introducing risk of human error.

(Un)Perplexed Spready solved this by embedding a powerful large language model (LLM), GLM-4.7, directly into a spreadsheet formula. The AI was instructed to map each product to the new taxonomy, intelligently using existing data and—when information was unclear—performing its own web research via integrated tools to verify product details.

"The line between a spreadsheet and an intelligent data assistant has officially blurred," said a Matasoft spokesperson. "This isn't about generating text; it's about applying reasoning and research to a core, time-consuming business problem. We're moving from automating calculations to automating analysis within AI tools."

Key Highlights from the Demonstration:

• AI Inside the Grid: The task was executed using a single formula, =ASK_LOCAL3(), which passed product data and a taxonomy list to the AI model alongside a natural language instruction.
• Cost-Effective, Powerful Stack: The demo leveraged the open-source GLM-4.7 model served via Ollama Cloud, noted for its high performance and cost-effectiveness compared to commercial APIs.
• Augmented with Live Research: Integrated tools (Matasoft Web Search/Reader) allowed the AI to autonomously browse and extract information from retail sites like Amazon to inform its decisions, showcasing a fully autonomous research-and-classify workflow.
• Tangible Business ROI: The demo provides a clear template for automating data standardization, classification, and enrichment tasks across industries such as e-commerce, retail, supply chain, and customer data management.

This demonstration underscores a major shift in data productivity. Professionals are no longer limited to formula-based logic; they can now delegate tasks requiring understanding, context, and external verification to an AI directly within their spreadsheet workspace.

(Un)Perplexed Spready supports multiple AI backends, including local private models via Ollama for data-sensitive environments and cloud-based models for scalable power, making advanced AI accessible without requiring infrastructure investment.

For a complete technical walkthrough, including the exact AI prompt, formula syntax, and analysis of the results, view the full case study: https://matasoft.hr/QTrendControl/index.php/un-perplexed-...