Matrices vs GRID

Matrices transforms data analysis by combining AI-powered spreadsheets with an innovative platform for training AI agents in simulated environments. It automates research and data gathering by using AI agents that fill spreadsheets with data from multiple sources, providing confidence scores for transparency. Beyond spreadsheets, Matrices is pioneering a system of virtual 'escape rooms'—complex simulations where AI agents learn tasks through trial and error without real-world risks. This approach enables AI to practice and improve on tasks like market research, coding, and business operations in safe, scalable environments.

The platform supports two main user roles: engineers who build configurable simulation components, and creative designers who craft scenarios for AI to solve. This ecosystem aims to scale AI training by crowdsourcing simulation creation, eventually covering a wide range of real-world tasks. Matrices targets founders, investors, sales professionals, CEOs, and AI researchers who benefit from automated data workflows and advanced AI training environments.

Matrices emphasizes natural language commands to reduce the learning curve, allowing users to automate complex workflows without coding. Its extensible design adapts to diverse needs, from simple data tasks to intricate AI training challenges. The company is actively expanding its simulation library and plans to open its level editor to the public, inviting collaboration to accelerate AI automation across industries.

Technically, Matrices leverages reinforcement learning by placing AI agents in simulated environments where they can experiment and learn from feedback. This method moves beyond imitation learning, enabling AI to achieve higher competence in practical tasks. The platform’s modular architecture supports integration with various web apps and data sources, making it a versatile tool for both data analysis and AI development.

Overall, Matrices offers a unique blend of AI spreadsheet automation and a scalable framework for training AI agents in realistic, risk-free settings. This dual focus positions it as a forward-looking solution for businesses and AI practitioners aiming to automate complex workflows and accelerate AI capabilities.

GRID is a spreadsheet engine built for agentic products. Developers install @grid-is/spreadsheet-engine from npm and embed Excel-compatible calculations directly in browsers, Node.js servers, or on-premise infrastructure. Agents call structured spreadsheet operations instead of generating code, which cuts token use and keeps recalculation deterministic.

The platform ships as a family of npm packages: a headless engine for logic, a themeable Spreadsheet Viewer for read-only display, a Spreadsheet Editor for in-product editing, and an Excel add-in for domain-specific skills inside Excel. GRID also publishes agent skills (npx skills add GRID-is/skills) so coding agents like Cursor and Claude Code can scaffold spreadsheet editors quickly.

GRID targets software teams that need spreadsheet reliability inside their own products. The engine runs nearly 200,000 automated unit tests and supports real-world Excel and Google Sheets compatibility. Major investment banks and a leading LLM provider use GRID as core spreadsheet infrastructure.

Founded in 2018 in Reykjavik, Iceland, GRID also offers Calculator Studio for interactive calculators built on spreadsheet models. Evaluation licenses are free for development and testing; commercial licenses are required when shipping to customers.