Autoplot

Product Introduction

1. Definition: Autoplot is a native macOS scientific workspace application designed for exploratory data analysis (EDA) and publication-quality scientific plotting. It is a technical desktop software suite that integrates data import, cleaning, statistical analysis, visualization, and figure composition into a single, unified environment.
2. Core Value Proposition: Autoplot exists to eliminate the fragmented, multi-tool workflow common in scientific research. Its primary value is providing a native Mac data analysis workspace that unifies data import, statistical fitting, GPU-accelerated visualization, and vector graphic export, all while ensuring full data privacy by keeping raw data on the local machine and leveraging local or privacy-focused AI assistance.

Main Features

1. Unified Scientific Workspace: Autoplot functions as a single document-based application where users can import tabular data, create derived variables, run analyses, generate plots, and compose multi-panel figures. It replaces the need to juggle between terminal scripts, Python IDEs, basic plotting tools, and image editors. The workspace maintains a dynamic link between data, analysis cards, and visualizations, so changes propagate automatically.
2. Native, GPU-Accelerated Visualization Engine: The application uses Apple's Metal framework for GPU-backed rendering. This enables smooth, interactive manipulation (rotation, zoom, pan) of large 3D scatter plots, surface plots, and complex correlation networks without performance degradation, even at full dataset sizes. It supports a wide range of plot types including X-Y plots, histograms with fits, 2D heat maps with contours, 3D surfaces, and categorical charts.
3. Local-First AI Assistant & Analysis Toolkit: Autoplot includes an optional AI assistant that operates under a strict data privacy model. The assistant can generate Python code for data splicing, filtering, custom statistics, and Matplotlib plotting based on user intent, but executes this code locally within the app's embedded Python stack (NumPy, SciPy, Matplotlib). For complex prompts, it can leverage Apple's Private Cloud Compute infrastructure, which is stateless and cryptographically verifiable, ensuring raw data values are never exposed for model training.
4. Vetted Analysis "Cards" and Reusable Outputs: The software provides pre-built, verified analysis modules ("cards") for common scientific tasks such as calculating log-binned probability density functions (PDFs), complementary cumulative distribution functions (CCDFs), power-law and truncated power-law fits, finite-size scaling (FSS), and scaling relations. Every analysis output becomes a reusable variable within the project, enabling iterative workflows.
5. Composition and Vector Export: Users can annotate plots directly with labels, legends, and callouts that persist through re-styling and re-analysis. Multiple plots can be arranged on a "board" to create multi-panel figures for publications. The system exports to print-ready, vector PDF format with embedded fonts, as well as PNG and JPG at journal-standard DPI settings. The entire analysis can also be exported as runnable Python/Matplotlib code for reproducibility.

Problems Solved

1. Pain Point: The "copy-paste treadmill" between disparate tools (SFTP clients, text editors, terminal for Python/Matplotlib, spreadsheet software, and graphic design tools) for data analysis and figure generation is inefficient and error-prone.
2. Target Audience: Primary user personas include academic researchers and scientists (especially in physics, biology, and engineering) who need to analyze experimental data and produce publication-quality figures; R&D professionals in industry working under NDAs who require powerful, local-first analysis tools; and university students in STEM fields who need to produce lab reports and thesis figures without mastering a full programming pipeline.
3. Use Cases: Essential scenarios include: rapidly exploring and fitting heavy-tailed distributions from simulation data; creating annotated, multi-panel figures for journal submission; screening high-dimensional datasets using correlation matrices and networks; interactively visualizing 3D point cloud or surface data; and performing reproducible analysis where the Python code can be archived alongside the final figure.

Unique Advantages

1. Differentiation: Unlike cloud-based data science notebooks (e.g., Google Colab) which require data upload, Autoplot is a local-native macOS application, ensuring data never leaves the user's machine. Compared to general-purpose tools like Python/Matplotlib (which require coding) or cross-platform GUI apps (often Electron-based), Autoplot offers a deeper, more performant native macOS integration with Metal GPU acceleration and a unified, non-linear workflow that bridges the gap between analysis and publication.
2. Key Innovation: The integration of a privacy-constrained AI assistant within a local-first desktop application is a significant innovation. The assistant acts as an intelligent interface to the full application and Python stack, but is architecturally designed to avoid raw data exfiltration, using schema understanding and local execution or Apple's Private Cloud Compute for heavier tasks. This provides the utility of AI-powered coding assistance without the privacy risks of standard cloud chatbots.

Frequently Asked Questions (FAQ)

1. Is Autoplot a cloud service or a desktop application? Autoplot is a native desktop application for macOS. All data processing, analysis, and visualization occur locally on your Mac. The optional AI assistant processes prompts either on-device or via Apple's stateless Private Cloud Compute servers, ensuring raw data does not leave your machine for model training or storage.
2. How does Autoplot handle large datasets compared to browser-based tools? As a native Metal-accelerated macOS app, Autoplot can handle large datasets directly from disk without browser memory limits or upload steps. Its GPU-backed rendering keeps 3D plots and complex visualizations interactive, a task that typically causes web-based visualization tools to lag or crash.
3. Can I reproduce my analysis outside of Autoplot? Yes. Any analysis performed using the built-in "cards" or generated by the AI assistant can be exported as standard Python code using NumPy, SciPy, and Matplotlib. This ensures full reproducibility and allows the workflow to be shared or audited by collaborators who may not use Autoplot.
4. What kind of data formats and import methods does Autoplot support? Autoplot specializes in tabular scientific data. It can import local files (like CSVs) directly and supports fetching data from SFTP servers. It provides native tools for setting delimiters, headers, and encodings, and can merge multiple files into a single project. For complex import logic, the AI assistant can write and execute local Python scripts.
5. Who is the ideal user for Autoplot versus learning Python/Matplotlib directly? Autoplot is ideal for scientists, students, and R&D professionals who need to achieve robust, publication-ready results quickly without wanting to become expert programmers. It encapsulates best-practice analysis modules and provides an intuitive GUI, while still offering the full power and transparency of Python via its assistant and code export for customization and verification.