PlotStudio AI is a desktop application for agentic data analysis. Users upload a dataset and ask questions in plain English, and the software uses a set of coordinated AI agents to carry out the analysis. The agents handle distinct roles, including routing the query, planning the analytical approach, writing and executing Python code, and synthesizing the results into a written report. The system runs multiple iterations, self-correcting its code until the analysis completes, and produces output that includes visualizations, statistical tests, tables, and a narrative interpretation of the findings. When a dataset is uploaded, PlotStudio profiles it, scores its data quality, and generates a cleaning plan describing what to drop or preserve. Each question a user asks produces a structured analysis page that is saved in the workspace, remains accessible over time, and can be referenced by the AI when answering later questions. A Verification Mode lets users review the AI's analysis plan before it runs, ask why a particular method was chosen, propose an alternative methodology, edit the plan in plain English, view a summary of changes, and undo edits. PlotStudio includes domain-specific capabilities across 12 specializations and can detect the relevant field from the data or accept up to two domains selected by the user, applying methods such as instrumental variable estimation, GARCH, and Cox proportional hazards where appropriate. The application also provides an interactive data explorer, suggested questions, multiple analysis modes, cross-analysis insights, and data lineage tracking. Analysis runs on the user's own machine, and PlotStudio does not see, store, or access the underlying data; AI calls are sent directly from the user's device to the model provider's API. Enterprise users can connect their own Azure OpenAI instance. Completed analyses can be exported as PDF reports with embedded charts and equations or as Jupyter notebooks containing the code and outputs. PlotStudio supports a range of file formats, does not require knowledge of Python or SQL, and is built by AR5 Labs LLC.
