Probly, an AI-powered spreadsheet application combining traditional spreadsheet functionality with Python data analysis capabilities, has sparked interest in the developer community. The tool aims to bridge the gap between conventional spreadsheets and more advanced data analysis by leveraging artificial intelligence to provide intelligent suggestions and automated analysis.
Real-World Testing Reveals Current Limitations
Early adopters have begun testing Probly with practical use cases, revealing both promise and limitations. One user attempted to categorize bank transactions into household expense groups, a common personal finance task. Despite the tool's well-executed interface and smooth setup process, the AI component struggled with accurate categorization, often mislabeling expenses or defaulting to generic categories.
I have a pressing need to come up with a household budget... The task I set is realworld - please categorize my bank transactions into household expense groups - and proved too much for my ChatGPT o1 account - most lines were labelled as 'other', bank charges were labelled 'fuel', etc.
This feedback highlights a key challenge in current AI applications: while the technology shows promise, real-world financial categorization tasks still present significant hurdles for even advanced language models.
Rethinking AI's Role in Data Analysis
The community discussion around Probly has sparked interesting conversations about the optimal role of AI in data analysis workflows. Rather than simply having AI write code, some users suggest that tools like Probly represent a more intuitive approach—asking questions directly and having AI determine what code is needed, execute it, and deliver results.
This paradigm shift focuses on outcomes rather than the intermediary step of code generation, potentially making data analysis more accessible to non-programmers. However, others note that seeing the generated code provides transparency into the AI's process, which may be valuable for verification and learning purposes.
Feature Requests Point to Future Development
Community feedback has already generated several feature requests that could shape Probly's development roadmap. These include Docker support for self-hosted environments, integration with alternative AI models like Ollama for users seeking OpenAI alternatives, Google Sheets compatibility, and options for embedding the tool as a component library.
The interest in containerization and self-hosting options suggests that privacy-conscious users and organizations are interested in the technology but require deployment flexibility. Meanwhile, requests for integration with existing tools like Google Sheets indicate that users want to enhance their current workflows rather than completely replace them.
Probly Technical Stack
- Frontend: Next.js 14, TypeScript, React
- Spreadsheet: Handsontable, HyperFormula
- Python Runtime: Pyodide (WebAssembly)
- LLM: OpenAI API
- Visualization: ECharts
Key Features
- Interactive Spreadsheet with formula support
- Python Analysis: Run Python code directly on spreadsheet data
- Data Visualization capabilities
- AI-Powered suggestions and automated analysis
Community Feature Requests
- Docker/docker-compose support for self-hosting
- Ollama integration as OpenAI alternative
- Google Sheets support
- Standalone npm component library for embedding
- Configuration to accumulate custom prompts
Technical Foundation Shows Promise
Probly's technical architecture combines modern web technologies with powerful data processing capabilities. Built on Next.js 14 with TypeScript and React for the frontend, it utilizes Handsontable and HyperFormula for spreadsheet functionality. The Python runtime is handled by Pyodide (WebAssembly), allowing Python code execution directly in the browser, while visualization is powered by Apache ECharts.
This stack enables a unique combination of spreadsheet usability with programming power, all within a browser environment that requires no server-side processing for many operations. The inclusion of WebAssembly technology through Pyodide represents a significant technical achievement, allowing complex data processing to happen client-side.
As AI-powered productivity tools continue to evolve, Probly represents an interesting exploration of how traditional spreadsheet workflows might be enhanced through artificial intelligence. While current limitations exist, the community's engagement suggests there's substantial interest in tools that can bridge the gap between simple spreadsheets and more complex data analysis environments.